- Spitfire List - http://spitfirelist.com -

The Cambridge Analytica Microcosm in Our Panoptic Macrocosm

Let the Great Unfriend­ing Com­mence! Specif­i­cal­ly, the mass unfriend­ing of Face­book. Which would be a well deserved unfriend­ing after the scan­dalous rev­e­la­tions in a recent series of arti­cles cen­tered around the claims of Christo­pher Wylie, a Cam­bridge Ana­lyt­i­ca whis­tle-blow­er who helped found the firm and worked there until late 2014 until he and oth­ers grew increas­ing­ly uncom­fort­able with the far right goals and ques­tion­able actions of the firm.

And it turns out those ques­tion­able actions by Cam­bridge involve a far larg­er and more scan­dalous Face­book pol­i­cy brought forth by anoth­er whis­tle-blow­er, Sandy Parak­i­las, the plat­form oper­a­tions man­ag­er at Face­book respon­si­ble for polic­ing data breach­es by third-par­ty soft­ware devel­op­ers between 2011 and 2012.

So here’s a rough break­down of what’s been learned so far:

Accord­ing to Christo­pher Wylie, Cam­bridge Ana­lyt­i­ca was “har­vest­ing” mas­sive amount data off of Face­book from peo­ple who did not give their per­mis­sion by uti­liz­ing a Face­book loop­hole. This “friends per­mis­sions” loop­hole allowed app devel­op­ers to scrape infor­ma­tion not just from the Face­book pro­files of the peo­ple that agree to use their apps but also their friends’ pro­files too. In oth­er words, if your Face­book friend down­loaded Cam­bridge Ana­lyt­i­ca’s app, Cam­bridge Ana­lyt­i­ca was allowed to grab pri­vate infor­ma­tion from your Face­book pro­file with­out your per­mis­sion. And you would nev­er know it.

So how many pro­files was Cam­bridge Ana­lyt­i­ca allowed to “har­vest” uti­liz­ing this “friends per­mis­sion” fea­ture? About 50 mil­lion, and only a tiny frac­tion (~270,000) of that 50 mil­lion peo­ple actu­al­ly agreed to use Cam­bridge Ana­lyt­i­ca’s app. The rest were all their friends. So Face­book lit­er­al­ly used the con­nec­tiv­i­ty of Face­book users against them.

Keep in mind that this isn’t a new rev­e­la­tion. There were reports last year about how Cam­bridge Ana­lyt­i­ca paid ~100,000 peo­ple a dol­lar or two (via Ama­zon’s Mechan­i­cal Turks micro-task plat­form) to take an online sur­vey. But the only way they could be paid was to down­load an app that gave Cam­bridge Ana­lyt­i­ca access to the pro­files of all their Face­book friends, even­tu­al­ly yield­ing ~30 mil­lion “har­vest­ed” pro­files [1]. Although accord­ing to these new reports that num­ber is clos­er to 50 mil­lion pro­files.

Before that, there was also a report from Decem­ber of 2015 about Cam­bridge Ana­lyt­i­ca’s build­ing of “psy­cho­graph­ic pro­files” for the Ted Cruz cam­paign. And that report also includ­ed the fact that this involved Face­book data har­vest­ed large­ly with­out users’ per­mis­sions [2].

So the fact that Cam­bridge Ana­lyt­i­ca was secret­ly har­vest­ing pri­vate Face­book user data with­out their per­mis­sions isn’t the big rev­e­la­tion here. What’s new is the rev­e­la­tion that what Cam­bridge Ana­lyt­i­ca did was inte­gral to Face­book’s busi­ness mod­el for years and very wide­spread.

This is where Sandy Parak­i­las comes into the pic­ture. Accord­ing to Parak­i­las, this pro­file-scrap­ing loop­hole that Cam­bridge Ana­lyt­i­ca was exploit­ing with its app was rou­tine­ly exploit­ing by pos­si­bly hun­dreds of thou­sands of oth­er app devel­op­ers for years. Yep. It turns out that Face­book had an arrange­ment going back to 2007 where the com­pa­ny would get a 30 per­cent cut in the mon­ey app devel­op­ers make off their Face­book apps and in exchange these devel­op­ers were giv­en the abil­i­ty to scrape the pro­files of not just the peo­ple who used their apps but also their friends. In oth­er words, Face­book was essen­tial­ly sell­ing the pri­vate infor­ma­tion of its users to app devel­op­ers. Secret­ly. Well, except it was­n’t a secret to all those app devel­op­ers. That’s also part of this scan­dal

This “friends per­mis­sion” fea­ture start­ed get­ting phased out around 2012, although it turns out Cam­bridge Ana­lyt­i­ca was one of the very last apps allowed to use it up into 2014.

Face­book has tried to defend itself by assert­ing that Face­book was only mak­ing this avail­able for things like aca­d­e­m­ic research and that Cam­bridge Ana­lyt­i­ca was there­fore mis­us­ing that data. And aca­d­e­m­ic research was in fact the cov­er sto­ry Cam­bridge Ana­lyt­i­ca used. Cam­bridge Ana­lyt­ic actu­al­ly set up a shell com­pa­ny, Glob­al Sci­ence Research (GRS), that was run by a Cam­bridge Uni­ver­si­ty pro­fes­sor, Alek­san­dr Kogan, and claimed to be pure­ly inter­est­ed in using that Face­book data for aca­d­e­m­ic research. The col­lect­ed data was then sent off to Cam­bridge Ana­lyt­i­ca. But accord­ing to Parak­i­las, Face­book was allow­ing devel­op­ers to uti­lize this “friends per­mis­sions” fea­ture rea­sons as vague as “improv­ing user expe­ri­ences”. Parak­i­las saw plen­ty of apps har­vest­ing this data for com­mer­cial pur­pos­es. Even worse, both Parak­i­las and Wylie paint a pic­ture of Face­book releas­ing this data and then doing almost noth­ing to ensure that it’s not mis­used.

So we’ve learned that Face­book was allow­ing app devel­op­ers to “har­vest” pri­vate data on Face­book users with­out their per­mis­sions from 2007–2014, and now we get to per­haps the most chill­ing part: Accord­ing to Parak­i­las, this data almost cer­tain­ly float­ing around in the black mar­ket. And it was so easy to set up an app and start col­lect­ing this kind of data that any­one with basic app cre­ate skills could start trawl­ing Face­book for data. And a major­i­ty of Face­book users prob­a­bly had their pro­files secret­ly “har­vest­ed” dur­ing this peri­od. If true, that means there’s like­ly a mas­sive black mar­ket of Face­book user pro­files just float­ing around out there and Face­book has done lit­tle to noth­ing to address this.

Parak­i­las, whose job it was to police data breach­es by third-par­ty soft­ware devel­op­ers from 2011–2012, under­stand­ably grew quite con­cerned over the risks to user data inher­ent in this busi­ness mod­el. So what did Face­book’s lead­er­ship do when he raised these con­cerns? They essen­tial­ly asked him “do you real­ly want to know how this data is being use” atti­tude and active­ly dis­cour­aged him from inves­ti­gat­ing how this data may be abused. Inten­tion­al­ly not know­ing about abus­es was oth­er part of the busi­ness mod­el. Crack­ing down on “rogue devel­op­ers” was very rare and the approval of Face­book CEO Mark Zucker­berg him­self was required to get an app kicked off the plat­form.

Face­book has been pub­licly deny­ing alle­ga­tions like this for years. It was the pub­lic denials that led Parak­i­las to come for­ward.

And it gets worse. It turns out that Alek­san­dr Kogan, the Uni­ver­si­ty of Cam­bridge aca­d­e­m­ic who end­ed up team­ing up with Cam­bridge Ana­lyt­i­ca and built the app that har­vest­ed the data, has a remark­ably close work­ing rela­tion­ship with Face­book. So close that Kogan actu­al­ly co-authored an aca­d­e­m­ic study pub­lished in 2015 with Face­book employ­ees. In addi­tion, one of Kogan’s part­ners in the data har­vest­ing, Joseph Chan­cel­lor, was also an author on the study and went on to join Face­book a few months after it was pub­lished.

It also looks like Steve Ban­non was over­see­ing this entire process, although he claims to know noth­ing.

Oh, and Palan­tir, the pri­vate intel­li­gence firm with deep ties to the US nation­al secu­ri­ty state owned by far right Face­book board mem­ber Peter Thiel, appears to have had an infor­mal rela­tion­ship with Cam­bridge Ana­lyt­i­ca this whole time, with Palan­tir employ­ees report­ed­ly trav­el­ing to Cam­bridge Ana­lyt­i­ca’s office to help build the psy­cho­log­i­cal pro­files. And this state of affairs is an exten­sion of how the inter­net has been used from its very con­cep­tion a half cen­tu­ry ago.

And that’s all part of why the Great Unfriend­ing of Face­book real­ly is long over­due. It’s one real­ly big rea­son to delete your Face­book account com­prised of many many many small egre­gious rea­sons.

So let’s start tak­ing a look at those many small rea­sons to delete your Face­book account with a look at a New York Times sto­ry about Christo­pher Wylie and his sto­ry of the ori­gins of Cam­bridge Ana­lyt­i­ca and the cru­cial role Face­book “har­vest­ing” played in pro­vid­ing the com­pa­ny with the data it need­ed to car­ry out the goals of its chief financiers: wag­ing the kind of ‘cul­ture war’ the bil­lion­aire far right Mer­cer fam­i­ly and Steve Ban­non want­ed to wage [3]:

The New York Times

How Trump Con­sul­tants Exploit­ed the Face­book Data of Mil­lions

by Matthew Rosen­berg, Nicholas Con­fes­sore and Car­ole Cad­wal­ladr;
03/17/2018

As the upstart vot­er-pro­fil­ing com­pa­ny Cam­bridge Ana­lyt­i­ca [4] pre­pared to wade into the 2014 Amer­i­can midterm elec­tions, it had a prob­lem.

The firm had secured a $15 mil­lion invest­ment from Robert Mer­cer [5], the wealthy Repub­li­can donor, and wooed his polit­i­cal advis­er, Stephen K. Ban­non, with the promise of tools that could iden­ti­fy the per­son­al­i­ties of Amer­i­can vot­ers and influ­ence their behav­ior. But it did not have the data to make its new prod­ucts work.

So the firm har­vest­ed pri­vate infor­ma­tion from the Face­book pro­files of more than 50 mil­lion users with­out their per­mis­sion, accord­ing to for­mer Cam­bridge employ­ees, asso­ciates and doc­u­ments, mak­ing it one of the largest data leaks in the social network’s his­to­ry. The breach allowed the com­pa­ny to exploit the pri­vate social media activ­i­ty of a huge swath of the Amer­i­can elec­torate, devel­op­ing tech­niques that under­pinned its work on Pres­i­dent Trump’s cam­paign in 2016.

An exam­i­na­tion by The New York Times and The Observ­er of Lon­don reveals how Cam­bridge Analytica’s dri­ve to bring to mar­ket a poten­tial­ly pow­er­ful new weapon put the firm — and wealthy con­ser­v­a­tive investors seek­ing to reshape pol­i­tics — under scruti­ny from inves­ti­ga­tors and law­mak­ers on both sides of the Atlantic.

Christo­pher Wylie, who helped found Cam­bridge and worked there until late 2014, said of its lead­ers: “Rules don’t mat­ter for them. For them, this is a war, and it’s all fair.”

“They want to fight a cul­ture war in Amer­i­ca,” he added. “Cam­bridge Ana­lyt­i­ca was sup­posed to be the arse­nal of weapons to fight that cul­ture war.”

Details of Cambridge’s acqui­si­tion [6] and use of Face­book data [2] have sur­faced in sev­er­al accounts since the busi­ness began work­ing on the 2016 cam­paign, set­ting off a furi­ous debate [7] about the mer­its of the firm’s so-called psy­cho­graph­ic mod­el­ing tech­niques.

But the full scale of the data leak involv­ing Amer­i­cans has not been pre­vi­ous­ly dis­closed — and Face­book, until now, has not acknowl­edged it. Inter­views with a half-dozen for­mer employ­ees and con­trac­tors, and a review of the firm’s emails and doc­u­ments, have revealed that Cam­bridge not only relied on the pri­vate Face­book data but still pos­sess­es most or all of the trove.

Cam­bridge paid to acquire the per­son­al infor­ma­tion through an out­side researcher who, Face­book says, claimed to be col­lect­ing it for aca­d­e­m­ic pur­pos­es.

Dur­ing a week of inquiries from The Times, Face­book down­played the scope of the leak and ques­tioned whether any of the data still remained out of its con­trol. But on Fri­day, the com­pa­ny post­ed a state­ment [8] express­ing alarm and promis­ing to take action.

“This was a scam — and a fraud,” Paul Gre­w­al, a vice pres­i­dent and deputy gen­er­al coun­sel at the social net­work, said in a state­ment to The Times ear­li­er on Fri­day. He added that the com­pa­ny was sus­pend­ing Cam­bridge Ana­lyt­i­ca, Mr. Wylie and the researcher, Alek­san­dr Kogan, a Russ­ian-Amer­i­can aca­d­e­m­ic, from Face­book. “We will take what­ev­er steps are required to see that the data in ques­tion is delet­ed once and for all — and take action against all offend­ing par­ties,” Mr. Gre­w­al said.

Alexan­der Nix, the chief exec­u­tive of Cam­bridge Ana­lyt­i­ca [9], and oth­er offi­cials had repeat­ed­ly denied obtain­ing or using Face­book data, most recent­ly dur­ing a par­lia­men­tary hear­ing last month. But in a state­ment to The Times, the com­pa­ny acknowl­edged that it had acquired the data, though it blamed Mr. Kogan for vio­lat­ing Facebook’s rules and said it had delet­ed the infor­ma­tion as soon as it learned of the prob­lem two years ago.

In Britain, Cam­bridge Ana­lyt­i­ca is fac­ing inter­twined inves­ti­ga­tions by Par­lia­ment and gov­ern­ment reg­u­la­tors into alle­ga­tions that it per­formed ille­gal work on the “Brex­it” cam­paign. The coun­try has strict pri­va­cy laws, and its infor­ma­tion com­mis­sion­er announced on Sat­ur­day [10] that she was look­ing into whether the Face­book data was “ille­gal­ly acquired and used.”

In the Unit­ed States, Mr. Mercer’s daugh­ter, Rebekah, a board mem­ber, Mr. Ban­non and Mr. Nix received warn­ings from their lawyer that it was ille­gal to employ for­eign­ers in polit­i­cal cam­paigns, accord­ing to com­pa­ny doc­u­ments and for­mer employ­ees.

Con­gres­sion­al inves­ti­ga­tors have ques­tioned Mr. Nix about the company’s role in the Trump cam­paign. And the Jus­tice Department’s spe­cial coun­sel, Robert S. Mueller III, has demand­ed [11] the emails of Cam­bridge Ana­lyt­i­ca employ­ees who worked for the Trump team as part of his inves­ti­ga­tion into Russ­ian inter­fer­ence in the elec­tion.

While the sub­stance of Mr. Mueller’s inter­est is a close­ly guard­ed secret, doc­u­ments viewed by The Times indi­cate that the firm’s British affil­i­ate claims to have worked in Rus­sia and Ukraine. And the Wik­iLeaks founder, Julian Assange, dis­closed in Octo­ber [12] that Mr. Nix had reached out to him dur­ing the cam­paign in hopes of obtain­ing pri­vate emails belong­ing to Mr. Trump’s Demo­c­ra­t­ic oppo­nent, Hillary Clin­ton.

The doc­u­ments also raise new ques­tions about Face­book, which is already grap­pling with intense crit­i­cism over the spread of Russ­ian pro­pa­gan­da and fake news. The data Cam­bridge col­lect­ed from pro­files, a por­tion of which was viewed by The Times, includ­ed details on users’ iden­ti­ties, friend net­works and “likes.” Only a tiny frac­tion of the users had agreed to release their infor­ma­tion to a third par­ty.

“Pro­tect­ing people’s infor­ma­tion is at the heart of every­thing we do,” Mr. Gre­w­al said. “No sys­tems were infil­trat­ed, and no pass­words or sen­si­tive pieces of infor­ma­tion were stolen or hacked.”

Still, he added, “it’s a seri­ous abuse of our rules.”

Read­ing Vot­ers’ Minds

The Bor­deaux flowed freely as Mr. Nix and sev­er­al col­leagues sat down for din­ner at the Palace Hotel in Man­hat­tan in late 2013, Mr. Wylie recalled in an inter­view. They had much to cel­e­brate.

Mr. Nix, a brash sales­man, led the small elec­tions divi­sion at SCL Group, a polit­i­cal and defense con­trac­tor. He had spent much of the year try­ing to break into the lucra­tive new world of polit­i­cal data, recruit­ing Mr. Wylie, then a 24-year-old polit­i­cal oper­a­tive with ties to vet­er­ans of Pres­i­dent Obama’s cam­paigns. Mr. Wylie was inter­est­ed in using inher­ent psy­cho­log­i­cal traits to affect vot­ers’ behav­ior and had assem­bled a team of psy­chol­o­gists and data sci­en­tists, some of them affil­i­at­ed with Cam­bridge Uni­ver­si­ty.

The group exper­i­ment­ed abroad, includ­ing in the Caribbean and Africa, where pri­va­cy rules were lax or nonex­is­tent and politi­cians employ­ing SCL were hap­py to pro­vide gov­ern­ment-held data, for­mer employ­ees said.

Then a chance meet­ing brought Mr. Nix into con­tact with Mr. Ban­non, the Bre­it­bart News fire­brand who would lat­er become a Trump cam­paign and White House advis­er, and with Mr. Mer­cer, one of the rich­est men on earth [13].

Mr. Nix and his col­leagues court­ed Mr. Mer­cer, who believed a sophis­ti­cat­ed data com­pa­ny could make him a king­mak­er in Repub­li­can pol­i­tics, and his daugh­ter Rebekah, who shared his con­ser­v­a­tive views. Mr. Ban­non was intrigued by the pos­si­bil­i­ty of using per­son­al­i­ty pro­fil­ing to shift America’s cul­ture and rewire its pol­i­tics, recalled Mr. Wylie and oth­er for­mer employ­ees, who spoke on the con­di­tion of anonymi­ty because they had signed nondis­clo­sure agree­ments. Mr. Ban­non and the Mer­cers declined to com­ment.

Mr. Mer­cer agreed to help finance a $1.5 mil­lion pilot project to poll vot­ers and test psy­cho­graph­ic mes­sag­ing in Virginia’s guber­na­to­r­i­al race in Novem­ber 2013, where the Repub­li­can attor­ney gen­er­al, Ken Cuc­cinel­li, ran against Ter­ry McAu­li­ffe, the Demo­c­ra­t­ic fund-rais­er. Though Mr. Cuc­cinel­li lost, Mr. Mer­cer com­mit­ted to mov­ing for­ward.

The Mer­cers want­ed results quick­ly, and more busi­ness beck­oned. In ear­ly 2014, the investor Toby Neuge­bauer and oth­er wealthy con­ser­v­a­tives were prepar­ing to put tens of mil­lions of dol­lars behind a pres­i­den­tial cam­paign for Sen­a­tor Ted Cruz of Texas, work that Mr. Nix was eager to win.

...

Mr. Wylie’s team had a big­ger prob­lem. Build­ing psy­cho­graph­ic pro­files on a nation­al scale required data the com­pa­ny could not gath­er with­out huge expense. Tra­di­tion­al ana­lyt­ics firms used vot­ing records and con­sumer pur­chase his­to­ries to try to pre­dict polit­i­cal beliefs and vot­ing behav­ior.

But those kinds of records were use­less for fig­ur­ing out whether a par­tic­u­lar vot­er was, say, a neu­rot­ic intro­vert, a reli­gious extro­vert, a fair-mind­ed lib­er­al or a fan of the occult. Those were among the psy­cho­log­i­cal traits the firm claimed would pro­vide a unique­ly pow­er­ful means of design­ing polit­i­cal mes­sages.

Mr. Wylie found a solu­tion at Cam­bridge University’s Psy­cho­met­rics Cen­tre. Researchers there had devel­oped a tech­nique to map per­son­al­i­ty traits based on what peo­ple had liked on Face­book. The researchers paid users small sums to take a per­son­al­i­ty quiz and down­load an app, which would scrape some pri­vate infor­ma­tion from their pro­files and those of their friends, activ­i­ty that Face­book per­mit­ted at the time. The approach, the sci­en­tists said, could reveal more about a per­son than their par­ents or roman­tic part­ners knew — a claim that has been dis­put­ed.

When the Psy­cho­met­rics Cen­tre declined to work with the firm, Mr. Wylie found some­one who would: Dr. Kogan, who was then a psy­chol­o­gy pro­fes­sor at the uni­ver­si­ty and knew of the tech­niques. Dr. Kogan built his own app and in June 2014 began har­vest­ing data for Cam­bridge Ana­lyt­i­ca. The busi­ness cov­ered the costs — more than $800,000 — and allowed him to keep a copy for his own research, accord­ing to com­pa­ny emails and finan­cial records.

All he divulged to Face­book, and to users in fine print, was that he was col­lect­ing infor­ma­tion for aca­d­e­m­ic pur­pos­es, the social net­work said. It did not ver­i­fy his claim. Dr. Kogan declined to pro­vide details of what hap­pened, cit­ing nondis­clo­sure agree­ments with Face­book and Cam­bridge Ana­lyt­i­ca, though he main­tained that his pro­gram was “a very stan­dard vanil­la Face­book app.”

He ulti­mate­ly pro­vid­ed over 50 mil­lion raw pro­files to the firm, Mr. Wylie said, a num­ber con­firmed by a com­pa­ny email and a for­mer col­league. Of those, rough­ly 30 mil­lion — a num­ber pre­vi­ous­ly report­ed by The Inter­cept [6] — con­tained enough infor­ma­tion, includ­ing places of res­i­dence, that the com­pa­ny could match users to oth­er records and build psy­cho­graph­ic pro­files. Only about 270,000 users — those who par­tic­i­pat­ed in the sur­vey — had con­sent­ed to hav­ing their data har­vest­ed.

Mr. Wylie said the Face­book data was “the sav­ing grace” that let his team deliv­er the mod­els it had promised the Mer­cers.

“We want­ed as much as we could get,” he acknowl­edged. “Where it came from, who said we could have it — we weren’t real­ly ask­ing.”

Mr. Nix tells a dif­fer­ent sto­ry. Appear­ing before a par­lia­men­tary com­mit­tee last month, he described Dr. Kogan’s con­tri­bu­tions as “fruit­less.”

An Inter­na­tion­al Effort

Just as Dr. Kogan’s efforts were get­ting under­way, Mr. Mer­cer agreed to invest $15 mil­lion in a joint ven­ture with SCL’s elec­tions divi­sion. The part­ners devised a con­vo­lut­ed cor­po­rate struc­ture, form­ing a new Amer­i­can com­pa­ny, owned almost entire­ly by Mr. Mer­cer, with a license to the psy­cho­graph­ics plat­form devel­oped by Mr. Wylie’s team, accord­ing to com­pa­ny doc­u­ments. Mr. Ban­non, who became a board mem­ber and investor, chose the name: Cam­bridge Ana­lyt­i­ca.

The firm was effec­tive­ly a shell. Accord­ing to the doc­u­ments and for­mer employ­ees, any con­tracts won by Cam­bridge, orig­i­nal­ly incor­po­rat­ed in Delaware, would be ser­viced by Lon­don-based SCL and over­seen by Mr. Nix, a British cit­i­zen who held dual appoint­ments at Cam­bridge Ana­lyt­i­ca and SCL. Most SCL employ­ees and con­trac­tors were Cana­di­an, like Mr. Wylie, or Euro­pean.

But in July 2014, an Amer­i­can elec­tion lawyer advis­ing the com­pa­ny, Lau­rence Levy, warned that the arrange­ment could vio­late laws lim­it­ing the involve­ment of for­eign nation­als in Amer­i­can elec­tions.

In a memo to Mr. Ban­non, Ms. Mer­cer and Mr. Nix, the lawyer, then at the firm Bracewell & Giu­liani, warned that Mr. Nix would have to recuse him­self “from sub­stan­tive man­age­ment” of any clients involved in Unit­ed States elec­tions. The data firm would also have to find Amer­i­can cit­i­zens or green card hold­ers, Mr. Levy wrote, “to man­age the work and deci­sion mak­ing func­tions, rel­a­tive to cam­paign mes­sag­ing and expen­di­tures.”

In sum­mer and fall 2014, Cam­bridge Ana­lyt­i­ca dived into the Amer­i­can midterm elec­tions, mobi­liz­ing SCL con­trac­tors and employ­ees around the coun­try. Few Amer­i­cans were involved in the work, which includ­ed polling, focus groups and mes­sage devel­op­ment for the John Bolton Super PAC, con­ser­v­a­tive groups in Col­orado and the cam­paign of Sen­a­tor Thom Tillis, the North Car­oli­na Repub­li­can.

Cam­bridge Ana­lyt­i­ca, in its state­ment to The Times, said that all “per­son­nel in strate­gic roles were U.S. nation­als or green card hold­ers.” Mr. Nix “nev­er had any strate­gic or oper­a­tional role” in an Amer­i­can elec­tion cam­paign, the com­pa­ny said.

Whether the company’s Amer­i­can ven­tures vio­lat­ed elec­tion laws would depend on for­eign employ­ees’ roles in each cam­paign, and on whether their work count­ed as strate­gic advice under Fed­er­al Elec­tion Com­mis­sion rules.

Cam­bridge Ana­lyt­i­ca appears to have exhib­it­ed a sim­i­lar pat­tern in the 2016 elec­tion cycle, when the com­pa­ny worked for the cam­paigns of Mr. Cruz and then Mr. Trump. While Cam­bridge hired more Amer­i­cans to work on the races that year, most of its data sci­en­tists were cit­i­zens of the Unit­ed King­dom or oth­er Euro­pean coun­tries, accord­ing to two for­mer employ­ees.

Under the guid­ance of Brad Parscale, Mr. Trump’s dig­i­tal direc­tor in 2016 and now the cam­paign man­ag­er for his 2020 re-elec­tion effort, Cam­bridge per­formed a vari­ety of ser­vices, for­mer cam­paign offi­cials said. That includ­ed design­ing tar­get audi­ences for dig­i­tal ads and fund-rais­ing appeals, mod­el­ing vot­er turnout, buy­ing $5 mil­lion in tele­vi­sion ads and deter­min­ing where Mr. Trump should trav­el to best drum up sup­port.

Cam­bridge exec­u­tives have offered con­flict­ing accounts about the use of psy­cho­graph­ic data on the cam­paign. Mr. Nix has said that the firm’s pro­files helped shape Mr. Trump’s strat­e­gy — state­ments dis­put­ed by oth­er cam­paign offi­cials — but also that Cam­bridge did not have enough time to com­pre­hen­sive­ly mod­el Trump vot­ers.

In a BBC inter­view last Decem­ber, Mr. Nix said that the Trump efforts drew on “lega­cy psy­cho­graph­ics” built for the Cruz cam­paign.

After the Leak

By ear­ly 2015, Mr. Wylie and more than half his orig­i­nal team of about a dozen peo­ple had left the com­pa­ny. Most were lib­er­al-lean­ing, and had grown dis­en­chant­ed with work­ing on behalf of the hard-right can­di­dates the Mer­cer fam­i­ly favored.

Cam­bridge Ana­lyt­i­ca, in its state­ment, said that Mr. Wylie had left to start a rival firm, and that it lat­er took legal action against him to enforce intel­lec­tu­al prop­er­ty claims. It char­ac­ter­ized Mr. Wylie and oth­er for­mer “con­trac­tors” as engag­ing in “what is clear­ly a mali­cious attempt to hurt the com­pa­ny.”

Near the end of that year, a report in The Guardian revealed [2] that Cam­bridge Ana­lyt­i­ca was using pri­vate Face­book data on the Cruz cam­paign, send­ing Face­book scram­bling. In a state­ment at the time, Face­book promised that it was “care­ful­ly inves­ti­gat­ing this sit­u­a­tion” and would require any com­pa­ny mis­us­ing its data to destroy it.

Face­book ver­i­fied the leak and — with­out pub­licly acknowl­edg­ing it — sought to secure the infor­ma­tion, efforts that con­tin­ued as recent­ly as August 2016. That month, lawyers for the social net­work reached out to Cam­bridge Ana­lyt­i­ca con­trac­tors. “This data was obtained and used with­out per­mis­sion,” said a let­ter that was obtained by the Times. “It can­not be used legit­i­mate­ly in the future and must be delet­ed imme­di­ate­ly.”

Mr. Gre­w­al, the Face­book deputy gen­er­al coun­sel, said in a state­ment that both Dr. Kogan and “SCL Group and Cam­bridge Ana­lyt­i­ca cer­ti­fied to us that they destroyed the data in ques­tion.”

But copies of the data still remain beyond Facebook’s con­trol. The Times viewed a set of raw data from the pro­files Cam­bridge Ana­lyt­i­ca obtained.

While Mr. Nix has told law­mak­ers that the com­pa­ny does not have Face­book data, a for­mer employ­ee said that he had recent­ly seen hun­dreds of giga­bytes on Cam­bridge servers, and that the files were not encrypt­ed.

Today, as Cam­bridge Ana­lyt­i­ca seeks to expand its busi­ness in the Unit­ed States and over­seas, Mr. Nix has men­tioned some ques­tion­able prac­tices. This Jan­u­ary, in under­cov­er footage filmed by Chan­nel 4 News in Britain and viewed by The Times, he boast­ed of employ­ing front com­pa­nies and for­mer spies on behalf of polit­i­cal clients around the world, and even sug­gest­ed ways to entrap politi­cians in com­pro­mis­ing sit­u­a­tions.

All the scruti­ny appears to have dam­aged Cam­bridge Analytica’s polit­i­cal busi­ness. No Amer­i­can cam­paigns or “super PACs” have yet report­ed pay­ing the com­pa­ny for work in the 2018 midterms, and it is unclear whether Cam­bridge will be asked to join Mr. Trump’s re-elec­tion cam­paign.

In the mean­time, Mr. Nix is seek­ing to take psy­cho­graph­ics to the com­mer­cial adver­tis­ing mar­ket. He has repo­si­tioned him­self as a guru for the dig­i­tal ad age — a “Math Man,” he puts it [14]. In the Unit­ed States last year, a for­mer employ­ee said, Cam­bridge pitched Mer­cedes-Benz, MetLife and the brew­er AB InBev, but has not signed them on.

———-

“How Trump Con­sul­tants Exploit­ed the Face­book Data of Mil­lions” by Matthew Rosen­berg, Nicholas Con­fes­sore and Car­ole Cad­wal­ladr; The New York Times; 03/17/2018 [3]

“They want to fight a cul­ture war in Amer­i­ca,” he added. “Cam­bridge Ana­lyt­i­ca was sup­posed to be the arse­nal of weapons to fight that cul­ture war.”

Cam­bridge Ana­lyt­i­ca was sup­posed to be the arse­nal of weapons to fight the cul­ture war Cam­bridge Ana­lyt­i­ca’s lead­er­ship want­ed to wage. But that arse­nal could­n’t be built with­out data on what makes us ‘tick’. That’s where Face­book pro­file har­vest­ing came in:

The firm had secured a $15 mil­lion invest­ment from Robert Mer­cer [5], the wealthy Repub­li­can donor, and wooed his polit­i­cal advis­er, Stephen K. Ban­non, with the promise of tools that could iden­ti­fy the per­son­al­i­ties of Amer­i­can vot­ers and influ­ence their behav­ior. But it did not have the data to make its new prod­ucts work.

So the firm har­vest­ed pri­vate infor­ma­tion from the Face­book pro­files of more than 50 mil­lion users with­out their per­mis­sion, accord­ing to for­mer Cam­bridge employ­ees, asso­ciates and doc­u­ments, mak­ing it one of the largest data leaks in the social network’s his­to­ry. The breach allowed the com­pa­ny to exploit the pri­vate social media activ­i­ty of a huge swath of the Amer­i­can elec­torate, devel­op­ing tech­niques that under­pinned its work on Pres­i­dent Trump’s cam­paign in 2016.

An exam­i­na­tion by The New York Times and The Observ­er of Lon­don reveals how Cam­bridge Analytica’s dri­ve to bring to mar­ket a poten­tial­ly pow­er­ful new weapon put the firm — and wealthy con­ser­v­a­tive investors seek­ing to reshape pol­i­tics — under scruti­ny from inves­ti­ga­tors and law­mak­ers on both sides of the Atlantic.

Christo­pher Wylie, who helped found Cam­bridge and worked there until late 2014, said of its lead­ers: “Rules don’t mat­ter for them. For them, this is a war, and it’s all fair.”
...

And the acqui­si­tion of these 50 mil­lion Face­book pro­files has nev­er been acknowl­edge by Face­book, until now. And most or per­haps all of that data is still in the hands of Cam­bridge Ana­lyt­i­ca:

...
But the full scale of the data leak involv­ing Amer­i­cans has not been pre­vi­ous­ly dis­closed — and Face­book, until now, has not acknowl­edged it. Inter­views with a half-dozen for­mer employ­ees and con­trac­tors, and a review of the firm’s emails and doc­u­ments, have revealed that Cam­bridge not only relied on the pri­vate Face­book data but still pos­sess­es most or all of the trove.
...

And Face­book isn’t alone in sud­den­ly dis­cov­er­ing that its data was “har­vest­ed” by Cam­bridge Ana­lyt­i­ca. Cam­bridge Ana­lyt­i­ca itself would­n’t admit this either. Until now. Now Cam­bridge Ana­lyt­i­ca admits it did indeed obtained Face­book’s data. But the com­pa­ny blames it all on Alek­san­dr Kogan, the Cam­bridge Uni­ver­si­ty aca­d­e­m­ic who ran the front-com­pa­ny that paid peo­ple to take the psy­cho­log­i­cal pro­file sur­veys, for vio­lat­ing Face­book’s data usage rules. It also claims it delet­ed all the “har­vest­ed” infor­ma­tion two years ago as soon as it learned there was a prob­lem. That’s Cam­bridge Ana­lyt­i­ca’s new sto­ry and it’s stick­ing to it. For now:

...
Alexan­der Nix, the chief exec­u­tive of Cam­bridge Ana­lyt­i­ca [9], and oth­er offi­cials had repeat­ed­ly denied obtain­ing or using Face­book data, most recent­ly dur­ing a par­lia­men­tary hear­ing last month. But in a state­ment to The Times, the com­pa­ny acknowl­edged that it had acquired the data, though it blamed Mr. Kogan for vio­lat­ing Facebook’s rules and said it had delet­ed the infor­ma­tion as soon as it learned of the prob­lem two years ago.
...

But Christo­pher Wylie has a very dif­fer­ent rec­ol­lec­tion of events. In 2013, Wylie was a 24-year-old polit­i­cal oper­a­tive with ties to vet­er­ans of Pres­i­dent Obama’s cam­paigns inter­est­ed in using psy­cho­log­i­cal traits to affect vot­ers’ behav­ior. He even had a team of psy­chol­o­gists and data sci­en­tists, some of them affil­i­at­ed with Cam­bridge Uni­ver­si­ty (where Alek­san­dr Kogan was also work­ing at the time). And that exper­tise in psy­cho­log­i­cal pro­fil­ing for polit­i­cal pur­pos­es is why Mr. Nix recruit­ed Wylie and his team.

Then Nix has a chance meet­ing with Steve Ban­non and Robert Mer­cer. Mer­cer shows inter­est in the com­pa­ny because he believes it can make him a Repub­li­can king­mak­er, while Ban­non was focused on the pos­si­bil­i­ty of using per­son­al­i­ty pro­fil­ing to shift America’s cul­ture and rewire its pol­i­tics. The Mer­cers end up invest­ing $1.5 mil­lion in a pilot project: polling vot­ers and test­ing psy­cho­graph­ic mes­sag­ing in Virginia’s 2013 guber­na­to­r­i­al race:

...
The Bor­deaux flowed freely as Mr. Nix and sev­er­al col­leagues sat down for din­ner at the Palace Hotel in Man­hat­tan in late 2013, Mr. Wylie recalled in an inter­view. They had much to cel­e­brate.

Mr. Nix, a brash sales­man, led the small elec­tions divi­sion at SCL Group, a polit­i­cal and defense con­trac­tor. He had spent much of the year try­ing to break into the lucra­tive new world of polit­i­cal data, recruit­ing Mr. Wylie, then a 24-year-old polit­i­cal oper­a­tive with ties to vet­er­ans of Pres­i­dent Obama’s cam­paigns. Mr. Wylie was inter­est­ed in using inher­ent psy­cho­log­i­cal traits to affect vot­ers’ behav­ior and had assem­bled a team of psy­chol­o­gists and data sci­en­tists, some of them affil­i­at­ed with Cam­bridge Uni­ver­si­ty.

The group exper­i­ment­ed abroad, includ­ing in the Caribbean and Africa, where pri­va­cy rules were lax or nonex­is­tent and politi­cians employ­ing SCL were hap­py to pro­vide gov­ern­ment-held data, for­mer employ­ees said.

Then a chance meet­ing brought Mr. Nix into con­tact with Mr. Ban­non, the Bre­it­bart News fire­brand who would lat­er become a Trump cam­paign and White House advis­er, and with Mr. Mer­cer, one of the rich­est men on earth [13].

Mr. Nix and his col­leagues court­ed Mr. Mer­cer, who believed a sophis­ti­cat­ed data com­pa­ny could make him a king­mak­er in Repub­li­can pol­i­tics, and his daugh­ter Rebekah, who shared his con­ser­v­a­tive views. Mr. Ban­non was intrigued by the pos­si­bil­i­ty of using per­son­al­i­ty pro­fil­ing to shift America’s cul­ture and rewire its pol­i­tics, recalled Mr. Wylie and oth­er for­mer employ­ees, who spoke on the con­di­tion of anonymi­ty because they had signed nondis­clo­sure agree­ments. Mr. Ban­non and the Mer­cers declined to com­ment.

Mr. Mer­cer agreed to help finance a $1.5 mil­lion pilot project to poll vot­ers and test psy­cho­graph­ic mes­sag­ing in Virginia’s guber­na­to­r­i­al race in Novem­ber 2013, where the Repub­li­can attor­ney gen­er­al, Ken Cuc­cinel­li, ran against Ter­ry McAu­li­ffe, the Demo­c­ra­t­ic fund-rais­er. Though Mr. Cuc­cinel­li lost, Mr. Mer­cer com­mit­ted to mov­ing for­ward.
...

So the pilot project pro­ceed, but there was a prob­lem: Wylie’s team sim­ply did not have the data it need­ed. They only had the kind of data tra­di­tion­al ana­lyt­ics firms had: vot­ing records and con­sumer pur­chase his­to­ries. And get­ting the kind of data they want­ed to gain insight into vot­er neu­roti­cisms and psy­cho­log­i­cal traits could be very expen­sive:

...
The Mer­cers want­ed results quick­ly, and more busi­ness beck­oned. In ear­ly 2014, the investor Toby Neuge­bauer and oth­er wealthy con­ser­v­a­tives were prepar­ing to put tens of mil­lions of dol­lars behind a pres­i­den­tial cam­paign for Sen­a­tor Ted Cruz of Texas, work that Mr. Nix was eager to win.

...

Mr. Wylie’s team had a big­ger prob­lem. Build­ing psy­cho­graph­ic pro­files on a nation­al scale required data the com­pa­ny could not gath­er with­out huge expense. Tra­di­tion­al ana­lyt­ics firms used vot­ing records and con­sumer pur­chase his­to­ries to try to pre­dict polit­i­cal beliefs and vot­ing behav­ior.

But those kinds of records were use­less for fig­ur­ing out whether a par­tic­u­lar vot­er was, say, a neu­rot­ic intro­vert, a reli­gious extro­vert, a fair-mind­ed lib­er­al or a fan of the occult. Those were among the psy­cho­log­i­cal traits the firm claimed would pro­vide a unique­ly pow­er­ful means of design­ing polit­i­cal mes­sages.
...

And that’s where Alek­san­dr Kogan enters the pic­ture: First, Wylie found that Cam­bridge University’s Psy­cho­met­rics Cen­tre had exact­ly the kind of set up he need­ed. Researchers there claimed to have devel­oped tech­niques for map­ping per­son­al­i­ty traits based on what peo­ple “liked” on Face­book. Bet­ter yet, this team already had an app that paid users small sums to take a per­son­al­i­ty quiz and down­load an app that would scrape pri­vate infor­ma­tion from their Face­book pro­files and from their friends’ Face­book pro­files. In oth­er words, Cam­bridge University’s Psy­cho­met­rics Cen­tre was already employ­ing exact­ly the same kind of “har­vest­ing” mod­el Kogan and Cam­bridge Ana­lyt­i­ca even­tu­al­ly end­ed up doing.

But there was a prob­lem for Wylie and his team: Cam­bridge University’s Psy­cho­met­rics Cen­tre declined to work with them:

...
Mr. Wylie found a solu­tion at Cam­bridge University’s Psy­cho­met­rics Cen­tre. Researchers there had devel­oped a tech­nique to map per­son­al­i­ty traits based on what peo­ple had liked on Face­book. The researchers paid users small sums to take a per­son­al­i­ty quiz and down­load an app, which would scrape some pri­vate infor­ma­tion from their pro­files and those of their friends, activ­i­ty that Face­book per­mit­ted at the time. The approach, the sci­en­tists said, could reveal more about a per­son than their par­ents or roman­tic part­ners knew — a claim that has been dis­put­ed.
...

But it was­n’t a par­tic­u­lar­ly big prob­lem because Wylie found anoth­er Cam­bridge Uni­ver­si­ty psy­chol­o­gy pro­fes­sor who was famil­iar with the tech­niques and will­ing to do the job: Alek­san­dr Kogan. So Kogan built his own psy­cho­log­i­cal pro­file app and began har­vest­ing data for Cam­bridge Ana­lyt­i­ca in June 2014. Kogan was even allowed to keep the har­vest­ed data for his own research accord­ing to his con­tract with Cam­bridge Ana­lyt­i­ca. Accord­ing to Face­book, the only thing Kogan told them and told the users of his app in the fine print was that he was col­lect­ing infor­ma­tion for aca­d­e­m­ic pur­pos­es. Although Face­book did­n’t appear to have ever attempt­ed to ver­i­fy that claim:

...
When the Psy­cho­met­rics Cen­tre declined to work with the firm, Mr. Wylie found some­one who would: Dr. Kogan, who was then a psy­chol­o­gy pro­fes­sor at the uni­ver­si­ty and knew of the tech­niques. Dr. Kogan built his own app and in June 2014 began har­vest­ing data for Cam­bridge Ana­lyt­i­ca. The busi­ness cov­ered the costs — more than $800,000 — and allowed him to keep a copy for his own research, accord­ing to com­pa­ny emails and finan­cial records.

All he divulged to Face­book, and to users in fine print, was that he was col­lect­ing infor­ma­tion for aca­d­e­m­ic pur­pos­es, the social net­work said. It did not ver­i­fy his claim. Dr. Kogan declined to pro­vide details of what hap­pened, cit­ing nondis­clo­sure agree­ments with Face­book and Cam­bridge Ana­lyt­i­ca, though he main­tained that his pro­gram was “a very stan­dard vanil­la Face­book app.”
...

In the end, Kogan’s app man­aged to “har­vest” 50 mil­lion Face­book pro­files based on a mere 270,000 peo­ple actu­al­ly sign­ing up for Kogan’s app. So for each per­son who signed up for the app there were ~185 oth­er peo­ple who had their pro­files sent to Kogan too.

And 30 mil­lion of those pro­files con­tained infor­ma­tion like places of res­i­dence that allowed them to match that Face­book pro­file with oth­er records (pre­sum­ably non-Face­book records) and build psy­cho­graph­ic pro­files, imply­ing that those 30 mil­lion records were mapped to real life peo­ple:

...
He ulti­mate­ly pro­vid­ed over 50 mil­lion raw pro­files to the firm, Mr. Wylie said, a num­ber con­firmed by a com­pa­ny email and a for­mer col­league. Of those, rough­ly 30 mil­lion — a num­ber pre­vi­ous­ly report­ed by The Inter­cept [6] — con­tained enough infor­ma­tion, includ­ing places of res­i­dence, that the com­pa­ny could match users to oth­er records and build psy­cho­graph­ic pro­files. Only about 270,000 users — those who par­tic­i­pat­ed in the sur­vey — had con­sent­ed to hav­ing their data har­vest­ed.

Mr. Wylie said the Face­book data was “the sav­ing grace” that let his team deliv­er the mod­els it had promised the Mer­cers.
...

So this har­vest­ing starts in mid-2014, but by ear­ly 2015, Wylie and more than half his orig­i­nal team leave the firm to start a rival firm, although it sounds lie con­cerns over the far right cause they were work­ing for was also behind their depar­ture:

...
By ear­ly 2015, Mr. Wylie and more than half his orig­i­nal team of about a dozen peo­ple had left the com­pa­ny. Most were lib­er­al-lean­ing, and had grown dis­en­chant­ed with work­ing on behalf of the hard-right can­di­dates the Mer­cer fam­i­ly favored.

Cam­bridge Ana­lyt­i­ca, in its state­ment, said that Mr. Wylie had left to start a rival firm, and that it lat­er took legal action against him to enforce intel­lec­tu­al prop­er­ty claims. It char­ac­ter­ized Mr. Wylie and oth­er for­mer “con­trac­tors” as engag­ing in “what is clear­ly a mali­cious attempt to hurt the com­pa­ny.”
...

Final­ly, this whole scan­dal goes pub­lic. Well, at least par­tial­ly: At the end of 2015, the Guardian reports this Face­book pro­file col­lec­tion scheme Cam­bridge Ana­lyt­i­ca was doing for the Ted Cruz cam­paign. Face­book does­n’t pub­licly acknowl­edge the truth of this report, but it did pub­licly state that it was “care­ful­ly inves­ti­gat­ing this sit­u­a­tion.” Face­book also sent a let­ter to Cam­bridge Ana­lyt­i­ca demand­ing that it destroy this data...except the let­ter was­n’t sent until August of 2016.

...
Near the end of that year, a report in The Guardian revealed [2] that Cam­bridge Ana­lyt­i­ca was using pri­vate Face­book data on the Cruz cam­paign, send­ing Face­book scram­bling. In a state­ment at the time, Face­book promised that it was “care­ful­ly inves­ti­gat­ing this sit­u­a­tion” and would require any com­pa­ny mis­us­ing its data to destroy it.

Face­book ver­i­fied the leak and — with­out pub­licly acknowl­edg­ing it — sought to secure the infor­ma­tion, efforts that con­tin­ued as recent­ly as August 2016. That month, lawyers for the social net­work reached out to Cam­bridge Ana­lyt­i­ca con­trac­tors. “This data was obtained and used with­out per­mis­sion,” said a let­ter that was obtained by the Times. “It can­not be used legit­i­mate­ly in the future and must be delet­ed imme­di­ate­ly.”
...

Face­book now claims that Cam­bridge Ana­lyt­i­ca “SCL Group and Cam­bridge Ana­lyt­i­ca cer­ti­fied to us that they destroyed the data in ques­tion.” But, of course, this was a lie. The New York Times was shown sets of the raw data.

And even more dis­turb­ing, a for­mer Cam­bridge Ana­lyt­i­ca employ­ee claims he recent­ly saw hun­dreds of giga­bytes on Cam­bridge Ana­lyt­i­ca’s servers. Unen­crypt­ed. Which means that data could poten­tial­ly be grabbed by any Cam­bridge Ana­lyt­i­ca employ­ee with access to that serv­er:

...
Mr. Gre­w­al, the Face­book deputy gen­er­al coun­sel, said in a state­ment that both Dr. Kogan and “SCL Group and Cam­bridge Ana­lyt­i­ca cer­ti­fied to us that they destroyed the data in ques­tion.”

But copies of the data still remain beyond Facebook’s con­trol. The Times viewed a set of raw data from the pro­files Cam­bridge Ana­lyt­i­ca obtained.

While Mr. Nix has told law­mak­ers that the com­pa­ny does not have Face­book data, a for­mer employ­ee said that he had recent­ly seen hun­dreds of giga­bytes on Cam­bridge servers, and that the files were not encrypt­ed.
...

So, to sum­ma­rize the key points from this New York Times arti­cle:

1. In 2013, Cam­bridge Ana­lyt­i­ca is formed when Alexan­der Nix, then a sales­man for the small elec­tions divi­sion at SCL Group, recruits Christo­pher Wylie and a team of psy­chol­o­gist to help devel­op a “polit­i­cal data” unit at the com­pa­ny, with an eye on the 2014 US mid-terms.

2. By chance, Nix and Wylie meet Steve Ban­non and Robert Mer­cer, who are quick­ly sold on the idea of psy­cho­graph­ic pro­fil­ing for polit­i­cal pur­pos­es. Ban­non was intrigue by the idea of using this data to wage the “cul­ture war.” Mer­cer agrees to invest $1.5 Bil­lion in a pilot project involv­ing the Vir­ginia guber­na­to­r­i­al race. Their suc­cess is lim­it­ed as Wylie soon dis­cov­ers that they don’t have the data they real­ly need to car­ry out their psy­cho­graph­ic pro­fil­ing project. But Robert Mer­cer remained com­mit­ted to the project.

3. Wylie found that Cam­bridge University’s Psy­cho­met­rics Cen­tre had exact­ly the kind of data they were seek­ing. Data that was being col­lect­ed via an app admin­is­tered through Face­book, where peo­ple were paid small amounts a mon­ey to take a sur­vey, and in exchange Cam­bridge University’s Psy­cho­met­rics Cen­tre was allowed to scrape their Face­book pro­file as well as the pro­files of all their Face­book friends.

4. Cam­bridge University’s Psy­cho­met­rics Cen­tre reject­ed Wylies offer to work with them, but there was anoth­er Cam­bridge Uni­ver­si­ty psy­chol­o­gy pro­fes­sor who was will­ing to do so, Alek­san­dr Kogan. Kogan pro­ceed­ed to start a com­pa­ny (as a front for Cam­bridge Ana­lyt­i­ca) and devel­op his own app, get­ting ~270,000 peo­ple to down­load it and give their per­mis­sion for their pro­files to be col­lect­ed. But using the “friends per­mis­sion” fea­ture, Kogan’s app end­ed col­lect­ing anoth­er ~50 mil­lion Face­book pro­files from the friends of those 270,000 peo­ple. ~30 mil­lion of those pro­files were matched to US vot­ers.

5. By ear­ly 2015, Wylie and his left-lean­ing team mem­bers leave Cam­bridge Ana­lyt­i­ca and form their own com­pa­ny, appar­ent­ly due to con­cerns over the far right goals of the firm.

6. Cam­bridge Ana­lyt­i­ca goes on to work for the Ted Cruz cam­paign. In late 2015, it’s report­ed that Cam­bridge Ana­lyt­i­ca work for Cruz involved work­ing with Face­book data from peo­ple who did­n’t give it per­mis­sion. Face­book issues a vague state­ment about how it’s going to inves­ti­gate.

7. In August 2016, Face­book sends a let­ter to Cam­bridge Ana­lyt­i­ca assert­ing that the data was obtained and used with­out per­mis­sion and must be delet­ed imme­di­ate­ly. The New York Times was just shown copies of exact­ly that data to write this arti­cle. Hun­dreds of giga­bytes of data that is com­plete­ly out­side Face­book’s con­trol.

8. Cam­bridge Ana­lyt­i­ca CEO (now for­mer CEO) Alexan­der Nix told law­mak­ers that the firm did­n’t pos­sess any Face­book data. So he was clear­ly com­plete­ly lying.

9. Final­ly, a for­mer Cam­bridge Ana­lyt­i­ca employ­ee showed the New York Times hun­dreds of giga­bytes of Face­book data. And it was unen­crypt­ed, so any­one with access to it could make a copy and give it to who­ev­er they want.

And that’s what we learned from just the New York Times’s ver­sion of this sto­ry. The Guardian Observ­er was also talk­ing with Christo­pher Wylie and oth­er Cam­bridge Ana­lyt­i­ca whis­tle-blow­ers. And while it large­ly cov­ers the same sto­ry as the New York Times report, the Observ­er arti­cle con­tains some addi­tion­al details.
1. For starters, the fol­low­ing arti­cle notes that the Facebook’s “plat­form pol­i­cy” allowed only col­lec­tion of friends’ data to improve user expe­ri­ence in the app and barred it being sold on or used for adver­tis­ing. That’s impor­tant to note because the stat­ed use of the data grabbed by Alek­san­dr Kogan’s app was for research pur­pos­es. But “improv­ing user expe­ri­ence in the app” is a far more gener­ic rea­son for grab­bing that data than aca­d­e­m­ic research pur­pos­es. And that hints at some­thing we’re going to see below from a Face­book whis­tle-blow­er: that all sorts of app devel­op­ers were grab­bing this kind of data using the ‘friends’ loop­hole for rea­sons that had absolute­ly noth­ing to do with aca­d­e­m­ic pur­pos­es and this was deemed fine by Face­book.

2. Face­book did­n’t for­mal­ly sus­pend Cam­bridge Ana­lyt­i­ca and Alek­san­dr Kogan from the plat­form until one day before the Observ­er arti­cle was pub­lished, which is more than two years after the ini­tial reports in late 2015 about the Cam­bridge Ana­lyt­i­ca mis­us­ing Face­book data for the Ted Cruz cam­paign. So if Face­book felt like Cam­bridge Ana­lyt­i­ca and Alek­san­dr Kogan was improp­er­ly obtain­ing and mis­us­ing its data it sure tried hard not to let on until the very last moment.

3. Simon Mil­ner, Facebook’s UK pol­i­cy direc­tor, told the UK MP when asked if Cam­bridge Ana­lyt­i­ca had Face­book data that, “They may have lots of data but it will not be Face­book user data. It may be data about peo­ple who are on Face­book that they have gath­ered them­selves, but it is not data that we have pro­vid­ed.” Which, again, as we’re going to see, was a total lie accord­ing to a Face­book whis­tle-blow­er because Face­book was rou­tine­ly pro­vid­ing exact­ly the kind of data Kogan’s app was col­lect­ing to thou­sands of devel­op­ers.

4. Alek­san­dr Kogan had a license from Face­book to col­lect pro­file data, but for research pur­pos­es, so when he used the data for com­mer­cial pur­pos­es he was vio­lat­ing his agree­ment, accord­ing to the arti­cle. Also, Kogan main­tains every­thing he did was legal, and says he had a “close work­ing rela­tion­ship” with Face­book, which had grant­ed him per­mis­sion for his apps. And as we’re going to see in sub­se­quent arti­cles, it does indeed look like Kogan is cor­rect and he was very open about using the data from the Cam­bridge Ana­lyt­i­ca app for com­mer­cial pur­pos­es and Face­book had no prob­lem with this.

5. In addi­tion to being a Cam­bridge Uni­ver­si­ty pro­fes­sor, Alek­san­dr Kogan has links to a Russ­ian uni­ver­si­ty and took Russ­ian grants for research. This will undoubt­ed­ly raise spec­u­la­tion about the pos­si­bil­i­ty that Kogan’s data was hand­ed over to the Krem­lin and used in the social-media influ­enc­ing cam­paign car­ried out by the Krem­lin-linked Inter­net Research Agency. If so, it’s still impor­tant to keep in mind that, based on what we’re going to see from Face­book whis­tle-blow­er Sandy Parak­i­las, the Krem­lin could have eas­i­ly set up all sorts of Face­book apps for col­lect­ing this kind of data because appar­ent­ly any­one could do it as long as the data was for “improv­ing the user expe­ri­ence”. That’s how obscene this sit­u­a­tion is. Kogan was not at all need­ed to pro­vide this data to the Krem­lin because it was so easy for any­one to obtain. In oth­er words, we should assume all sorts of gov­ern­ments have this kind of data.

6. The legal let­ter sent by Face­book to Cam­bridge Ana­lyt­i­ca in August 2016 demand­ing that it delete the data was sent just days before it was offi­cial­ly announced that Steve Ban­non was tak­ing over as cam­paign man­ag­er for Trump and bring­ing Cam­bridge Ana­lyt­i­ca with him. That sure does seem like Face­book knew about Ban­non’s involve­ment with Cam­bridge Ana­lyt­i­ca and the fact that Ban­non was going to become Trump’s cam­paign man­ag­er and bring Cam­bridge Ana­lyt­i­ca into the cam­paign.

7. Steve Bannon’s lawyer said he had no com­ment because his client “knows noth­ing about the claims being assert­ed”. He added: “The first Mr Ban­non heard of these reports was from media inquiries in the past few days.”

So as we can see, like the prover­bial onion, the more lay­ers you peel back on the sto­ry Cam­bridge Ana­lyt­i­ca and Face­book have been ped­dling about how this data was obtained and used, the more acrid and mal­odor­ous it gets. With a dis­tinct tinge of BS [15]:

The Guardian

Revealed: 50 mil­lion Face­book pro­files har­vest­ed for Cam­bridge Ana­lyt­i­ca in major data breach

Whistle­blow­er describes how firm linked to for­mer Trump advis­er Steve Ban­non com­piled user data to tar­get Amer­i­can vot­ers

Car­ole Cad­wal­ladr and Emma Gra­ham-Har­ri­son

Sat 17 Mar 2018 18.03 EDT

The data ana­lyt­ics firm that worked with Don­ald Trump’s elec­tion team and the win­ning Brex­it cam­paign har­vest­ed mil­lions of Face­book pro­files of US vot­ers, in one of the tech giant’s biggest ever data breach­es, and used them to build a pow­er­ful soft­ware pro­gram to pre­dict and influ­ence choic­es at the bal­lot box.

A whistle­blow­er has revealed to the Observ­er how Cam­bridge Ana­lyt­i­ca – a com­pa­ny owned by the hedge fund bil­lion­aire Robert Mer­cer, and head­ed at the time by Trump’s key advis­er Steve Ban­non – used per­son­al infor­ma­tion tak­en with­out autho­ri­sa­tion in ear­ly 2014 to build a sys­tem that could pro­file indi­vid­ual US vot­ers, in order to tar­get them with per­son­alised polit­i­cal adver­tise­ments.

Christo­pher Wylie, who worked with a Cam­bridge Uni­ver­si­ty aca­d­e­m­ic to obtain the data, told the Observ­er: “We exploit­ed Face­book to har­vest mil­lions of people’s pro­files. And built mod­els to exploit what we knew about them and tar­get their inner demons. That was the basis the entire com­pa­ny was built on.

Doc­u­ments seen by the Observ­er, and con­firmed by a Face­book state­ment, show that by late 2015 the com­pa­ny had found out that infor­ma­tion had been har­vest­ed on an unprece­dent­ed scale. How­ev­er, at the time it failed to alert users and took only lim­it­ed steps to recov­er and secure the pri­vate infor­ma­tion of more than 50 mil­lion indi­vid­u­als.

The New York Times is report­ing [3] that copies of the data har­vest­ed for Cam­bridge Ana­lyt­i­ca could still be found online; its report­ing team had viewed some of the raw data.

The data was col­lect­ed through an app called thi­sisy­our­dig­i­tal­life, built by aca­d­e­m­ic Alek­san­dr Kogan, sep­a­rate­ly from his work at Cam­bridge Uni­ver­si­ty. Through his com­pa­ny Glob­al Sci­ence Research (GSR), in col­lab­o­ra­tion with Cam­bridge Ana­lyt­i­ca, hun­dreds of thou­sands of users were paid to take a per­son­al­i­ty test and agreed to have their data col­lect­ed for aca­d­e­m­ic use.

How­ev­er, the app also col­lect­ed the infor­ma­tion of the test-tak­ers’ Face­book friends, lead­ing to the accu­mu­la­tion of a data pool tens of mil­lions-strong. Facebook’s “plat­form pol­i­cy” allowed only col­lec­tion of friends’ data to improve user expe­ri­ence in the app and barred it being sold on or used for adver­tis­ing. The dis­cov­ery of the unprece­dent­ed data har­vest­ing, and the use to which it was put, rais­es urgent new ques­tions about Facebook’s role in tar­get­ing vot­ers in the US pres­i­den­tial elec­tion. It comes only weeks after indict­ments of 13 Rus­sians [16] by the spe­cial coun­sel Robert Mueller which stat­ed they had used the plat­form to per­pe­trate “infor­ma­tion war­fare” against the US.

Cam­bridge Ana­lyt­i­ca and Face­book are one focus of an inquiry into data and pol­i­tics by the British Infor­ma­tion Commissioner’s Office. Sep­a­rate­ly, the Elec­toral Com­mis­sion is also inves­ti­gat­ing what role Cam­bridge Ana­lyt­i­ca played in the EU ref­er­en­dum.

...

On Fri­day, four days after the Observ­er sought com­ment for this sto­ry, but more than two years after the data breach was first report­ed, Face­book announced [8] that it was sus­pend­ing Cam­bridge Ana­lyt­i­ca and Kogan from the plat­form, pend­ing fur­ther infor­ma­tion over mis­use of data. Sep­a­rate­ly, Facebook’s exter­nal lawyers warned the Observ­er it was mak­ing “false and defam­a­to­ry” alle­ga­tions, and reserved Facebook’s legal posi­tion.

The rev­e­la­tions pro­voked wide­spread out­rage. The Mass­a­chu­setts Attor­ney Gen­er­al Mau­ra Healey announced that the state would be launch­ing an inves­ti­ga­tion. “Res­i­dents deserve answers imme­di­ate­ly from Face­book and Cam­bridge Ana­lyt­i­ca,” she said on Twit­ter.

The Demo­c­ra­t­ic sen­a­tor Mark Warn­er said the har­vest­ing of data on such a vast scale for polit­i­cal tar­get­ing under­lined the need for Con­gress to improve con­trols. He has pro­posed an Hon­est Ads Act to reg­u­late online polit­i­cal adver­tis­ing the same way as tele­vi­sion, radio and print. “This sto­ry is more evi­dence that the online polit­i­cal adver­tis­ing mar­ket is essen­tial­ly the Wild West. Whether it’s allow­ing Rus­sians to pur­chase polit­i­cal ads, or exten­sive micro-tar­get­ing based on ill-got­ten user data, it’s clear that, left unreg­u­lat­ed, this mar­ket will con­tin­ue to be prone to decep­tion and lack­ing in trans­paren­cy,” he said.

Last month both Face­book and the CEO of Cam­bridge Ana­lyt­i­ca, Alexan­der Nix, told a par­lia­men­tary inquiry on fake news: that the com­pa­ny did not have or use pri­vate Face­book data.

Simon Mil­ner, Facebook’s UK pol­i­cy direc­tor, when asked if Cam­bridge Ana­lyt­i­ca had Face­book data, told MPs: “They may have lots of data but it will not be Face­book user data. It may be data about peo­ple who are on Face­book that they have gath­ered them­selves, but it is not data that we have pro­vid­ed.”

Cam­bridge Analytica’s chief exec­u­tive, Alexan­der Nix, told the inquiry: “We do not work with Face­book data and we do not have Face­book data.”

Wylie, a Cana­di­an data ana­lyt­ics expert who worked with Cam­bridge Ana­lyt­i­ca and Kogan to devise and imple­ment the scheme, showed a dossier of evi­dence about the data mis­use to the Observ­er which appears to raise ques­tions about their tes­ti­mo­ny. He has passed it to the Nation­al Crime Agency’s cyber­crime unit and the Infor­ma­tion Commissioner’s Office. It includes emails, invoic­es, con­tracts and bank trans­fers that reveal more than 50 mil­lion pro­files – most­ly belong­ing to reg­is­tered US vot­ers – were har­vest­ed from the site in one of the largest-ever breach­es of Face­book data. Face­book on Fri­day said that it was also sus­pend­ing Wylie from access­ing the plat­form while it car­ried out its inves­ti­ga­tion, despite his role as a whistle­blow­er.

At the time of the data breach, Wylie was a Cam­bridge Ana­lyt­i­ca employ­ee, but Face­book described him as work­ing for Eunoia Tech­nolo­gies, a firm he set up on his own after leav­ing his for­mer employ­er in late 2014.

The evi­dence Wylie sup­plied to UK and US author­i­ties includes a let­ter from Facebook’s own lawyers sent to him in August 2016, ask­ing him to destroy any data he held that had been col­lect­ed by GSR, the com­pa­ny set up by Kogan to har­vest the pro­files.

That legal let­ter was sent sev­er­al months after the Guardian first report­ed the breach and days before it was offi­cial­ly announced that Ban­non was tak­ing over as cam­paign man­ag­er for Trump and bring­ing Cam­bridge Ana­lyt­i­ca with him.

“Because this data was obtained and used with­out per­mis­sion, and because GSR was not autho­rised to share or sell it to you, it can­not be used legit­i­mate­ly in the future and must be delet­ed imme­di­ate­ly,” the let­ter said.

Face­book did not pur­sue a response when the let­ter ini­tial­ly went unan­swered for weeks because Wylie was trav­el­ling, nor did it fol­low up with foren­sic checks on his com­put­ers or stor­age, he said.

“That to me was the most aston­ish­ing thing. They wait­ed two years and did absolute­ly noth­ing to check that the data was delet­ed. All they asked me to do was tick a box on a form and post it back.”

Paul-Olivi­er Dehaye, a data pro­tec­tion spe­cial­ist, who spear­head­ed the inves­tiga­tive efforts into the tech giant, said: “Face­book has denied and denied and denied this. It has mis­led MPs and con­gres­sion­al inves­ti­ga­tors and it’s failed in its duties to respect the law.

“It has a legal oblig­a­tion to inform reg­u­la­tors and indi­vid­u­als about this data breach, and it hasn’t. It’s failed time and time again to be open and trans­par­ent.”

A major­i­ty of Amer­i­can states have laws requir­ing noti­fi­ca­tion in some cas­es of data breach, includ­ing Cal­i­for­nia, where Face­book is based.

Face­book denies that the har­vest­ing of tens of mil­lions of pro­files by GSR and Cam­bridge Ana­lyt­i­ca was a data breach. It said in a state­ment that Kogan “gained access to this infor­ma­tion in a legit­i­mate way and through the prop­er chan­nels” but “did not sub­se­quent­ly abide by our rules” because he passed the infor­ma­tion on to third par­ties.

Face­book said it removed the app in 2015 and required cer­ti­fi­ca­tion from every­one with copies that the data had been destroyed, although the let­ter to Wylie did not arrive until the sec­ond half of 2016. “We are com­mit­ted to vig­or­ous­ly enforc­ing our poli­cies to pro­tect people’s infor­ma­tion. We will take what­ev­er steps are required to see that this hap­pens,” Paul Gre­w­al, Facebook’s vice-pres­i­dent, said in a state­ment. The com­pa­ny is now inves­ti­gat­ing reports that not all data had been delet­ed.

Kogan, who has pre­vi­ous­ly unre­port­ed links to a Russ­ian uni­ver­si­ty and took Russ­ian grants for research, had a licence from Face­book to col­lect pro­file data, but it was for research pur­pos­es only. So when he hoovered up infor­ma­tion for the com­mer­cial ven­ture, he was vio­lat­ing the company’s terms. Kogan main­tains every­thing he did was legal, and says he had a “close work­ing rela­tion­ship” with Face­book, which had grant­ed him per­mis­sion for his apps.

The Observ­er has seen a con­tract dat­ed 4 June 2014, which con­firms SCL, an affil­i­ate of Cam­bridge Ana­lyt­i­ca, entered into a com­mer­cial arrange­ment with GSR, entire­ly premised on har­vest­ing and pro­cess­ing Face­book data. Cam­bridge Ana­lyt­i­ca spent near­ly $1m on data col­lec­tion, which yield­ed more than 50 mil­lion indi­vid­ual pro­files that could be matched to elec­toral rolls. It then used the test results and Face­book data to build an algo­rithm that could analyse indi­vid­ual Face­book pro­files and deter­mine per­son­al­i­ty traits linked to vot­ing behav­iour.

The algo­rithm and data­base togeth­er made a pow­er­ful polit­i­cal tool [17]. It allowed a cam­paign to iden­ti­fy pos­si­ble swing vot­ers and craft mes­sages more like­ly to res­onate.

“The ulti­mate prod­uct of the train­ing set is cre­at­ing a ‘gold stan­dard’ of under­stand­ing per­son­al­i­ty from Face­book pro­file infor­ma­tion,” the con­tract spec­i­fies. It promis­es to cre­ate a data­base of 2 mil­lion “matched” pro­files, iden­ti­fi­able and tied to elec­toral reg­is­ters, across 11 states, but with room to expand much fur­ther.

At the time, more than 50 mil­lion pro­files rep­re­sent­ed around a third of active North Amer­i­can Face­book users, and near­ly a quar­ter of poten­tial US vot­ers. Yet when asked by MPs if any of his firm’s data had come from GSR, Nix said: “We had a rela­tion­ship with GSR. They did some research for us back in 2014. That research proved to be fruit­less and so the answer is no.”

Cam­bridge Ana­lyt­i­ca said that its con­tract with GSR stip­u­lat­ed that Kogan should seek informed con­sent for data col­lec­tion and it had no rea­son to believe he would not.

GSR was “led by a seem­ing­ly rep­utable aca­d­e­m­ic at an inter­na­tion­al­ly renowned insti­tu­tion who made explic­it con­trac­tu­al com­mit­ments to us regard­ing its legal author­i­ty to license data to SCL Elec­tions”, a com­pa­ny spokesman said.

SCL Elec­tions, an affil­i­ate, worked with Face­book over the peri­od to ensure it was sat­is­fied no terms had been “know­ing­ly breached” and pro­vid­ed a signed state­ment that all data and deriv­a­tives had been delet­ed, he said. Cam­bridge Ana­lyt­i­ca also said none of the data was used in the 2016 pres­i­den­tial elec­tion.

Steve Bannon’s lawyer said he had no com­ment because his client “knows noth­ing about the claims being assert­ed”. He added: “The first Mr Ban­non heard of these reports was from media inquiries in the past few days.” He direct­ed inquires to Nix.

———-

“Revealed: 50 mil­lion Face­book pro­files har­vest­ed for Cam­bridge Ana­lyt­i­ca in major data breach” by Car­ole Cad­wal­ladr and Emma Gra­ham-Har­ri­son; The Guardian; 03/17/2018 [15]

“Christo­pher Wylie, who worked with a Cam­bridge Uni­ver­si­ty aca­d­e­m­ic to obtain the data, told the Observ­er: “We exploit­ed Face­book to har­vest mil­lions of people’s pro­files. And built mod­els to exploit what we knew about them and tar­get their inner demons. That was the basis the entire com­pa­ny was built on.””

Exploit­ing every­one’s inner demons. Yeah, that sounds like some­thing Steve Ban­non [18] and Robert Mer­cer [19] would be inter­est­ed in. And it explains why Face­book data would have been poten­tial­ly so use­ful for exploit­ing those demons. Recall that the orig­i­nal non-Face­book data that Christo­pher Wylie and ini­tial Cam­bridge Ana­lyt­i­ca team was work­ing with with in 2013 and 2014 was­n’t seen as effec­tive. It did­n’t have that inner-demon-influ­enc­ing gran­u­lar­i­ty. And then they dis­cov­ered the Face­book data avail­able through this app loop­hole and it was tak­en to a dif­fer­ent lev­el. Remem­ber when Face­book ran that con­tro­ver­sial exper­i­ment on users where they tried to manip­u­late their emo­tions by alter­ing their news feeds [20]? It sounds like that’s what Cam­bridge Ana­lyt­i­ca was basi­cal­ly try­ing to do using Face­book ads instead of the news­feed, but per­haps in a more micro­tar­get­ed way.

And that’s all because Facebook’s “plat­form pol­i­cy” allowed for the col­lec­tion of friends’ data to “improve user expe­ri­ence in the app” with the non-enforced request that the data not be sold on or used for adver­tis­ing:

...
The data was col­lect­ed through an app called thi­sisy­our­dig­i­tal­life, built by aca­d­e­m­ic Alek­san­dr Kogan, sep­a­rate­ly from his work at Cam­bridge Uni­ver­si­ty. Through his com­pa­ny Glob­al Sci­ence Research (GSR), in col­lab­o­ra­tion with Cam­bridge Ana­lyt­i­ca, hun­dreds of thou­sands of users were paid to take a per­son­al­i­ty test and agreed to have their data col­lect­ed for aca­d­e­m­ic use.

How­ev­er, the app also col­lect­ed the infor­ma­tion of the test-tak­ers’ Face­book friends, lead­ing to the accu­mu­la­tion of a data pool tens of mil­lions-strong. Facebook’s “plat­form pol­i­cy” allowed only col­lec­tion of friends’ data to improve user expe­ri­ence in the app and barred it being sold on or used for adver­tis­ing. The dis­cov­ery of the unprece­dent­ed data har­vest­ing, and the use to which it was put, rais­es urgent new ques­tions about Facebook’s role in tar­get­ing vot­ers in the US pres­i­den­tial elec­tion. It comes only weeks after indict­ments of 13 Rus­sians [16] by the spe­cial coun­sel Robert Mueller which stat­ed they had used the plat­form to per­pe­trate “infor­ma­tion war­fare” against the US.
...

Just imag­ine how many app devel­op­ers were using this over the 2007–2014 peri­od Face­book had this “plat­form pol­i­cy” that allowed data cap­tures of friends’ “to improve user expe­ri­ence in the app”. It was­n’t just Cam­bridge Ana­lyt­i­ca that took advan­tage of this. That’s a big part of the sto­ry here.

And yet when Simon Mil­ner, Facebook’s UK pol­i­cy direc­tor, was asked if Cam­bridge Ana­lyt­i­ca had Face­book data, he said, “They may have lots of data but it will not be Face­book user data. It may be data about peo­ple who are on Face­book that they have gath­ered them­selves, but it is not data that we have pro­vid­ed.”:

...
Last month both Face­book and the CEO of Cam­bridge Ana­lyt­i­ca, Alexan­der Nix, told a par­lia­men­tary inquiry on fake news: that the com­pa­ny did not have or use pri­vate Face­book data.

Simon Mil­ner, Facebook’s UK pol­i­cy direc­tor, when asked if Cam­bridge Ana­lyt­i­ca had Face­book data, told MPs: “They may have lots of data but it will not be Face­book user data. It may be data about peo­ple who are on Face­book that they have gath­ered them­selves, but it is not data that we have pro­vid­ed.”

Cam­bridge Analytica’s chief exec­u­tive, Alexan­der Nix, told the inquiry: “We do not work with Face­book data and we do not have Face­book data.”
...

And note how the arti­cle appears to say the data Cam­bridge Ana­lyt­i­ca col­lect­ed on Face­book users includ­ed “emails, invoic­es, con­tracts and bank trans­fers that reveal more than 50 mil­lion pro­files.” It’s not clear if that’s a ref­er­ence to emails, invoic­es, con­tracts and bank trans­fers that involved with set­ting up Cam­bridge Ana­lyt­i­ca or emails, invoic­es, con­tracts and bank trans­fers from Face­book users, but if that was from users that would be wild­ly scan­dalous:

...
Wylie, a Cana­di­an data ana­lyt­ics expert who worked with Cam­bridge Ana­lyt­i­ca and Kogan to devise and imple­ment the scheme, showed a dossier of evi­dence about the data mis­use to the Observ­er which appears to raise ques­tions about their tes­ti­mo­ny. He has passed it to the Nation­al Crime Agency’s cyber­crime unit and the Infor­ma­tion Commissioner’s Office. It includes emails, invoic­es, con­tracts and bank trans­fers that reveal more than 50 mil­lion pro­filesmost­ly belong­ing to reg­is­tered US vot­ers – were har­vest­ed from the site in one of the largest-ever breach­es of Face­book data. Face­book on Fri­day said that it was also sus­pend­ing Wylie from access­ing the plat­form while it car­ried out its inves­ti­ga­tion, despite his role as a whistle­blow­er.
...

So it will be inter­est­ing to see if that point of ambi­gu­i­ty is ever clar­i­fied some­where. Because wow would that be scan­dalous if emails, invoic­es, con­tracts and bank trans­fers of Face­book users were released through this “plat­form pol­i­cy”.

Either way, it looks unam­bigu­ous­ly awful for Face­book. Espe­cial­ly now that we learn that the cease and destroy let­ter Face­book sent to Cam­bridge Ana­lyt­i­ca in August of 2016 was sus­pi­cious­ly sent just days before Steve Ban­non, a founder and offi­cer of Cam­bridge Ana­lyt­i­ca, becomes Trump’s cam­paign man­ag­er and brings the com­pa­ny into the Trump cam­paign:

...
The evi­dence Wylie sup­plied to UK and US author­i­ties includes a let­ter from Facebook’s own lawyers sent to him in August 2016, ask­ing him to destroy any data he held that had been col­lect­ed by GSR, the com­pa­ny set up by Kogan to har­vest the pro­files.

That legal let­ter was sent sev­er­al months after the Guardian first report­ed the breach and days before it was offi­cial­ly announced that Ban­non was tak­ing over as cam­paign man­ag­er for Trump and bring­ing Cam­bridge Ana­lyt­i­ca with him.

“Because this data was obtained and used with­out per­mis­sion, and because GSR was not autho­rised to share or sell it to you, it can­not be used legit­i­mate­ly in the future and must be delet­ed imme­di­ate­ly,” the let­ter said.
...

And the only thing Face­book did to con­firm that the Face­book data was­n’t mis­used, accord­ing to Christo­pher Wylie, was to ask that a box be checked a box on a form:

...
Face­book did not pur­sue a response when the let­ter ini­tial­ly went unan­swered for weeks because Wylie was trav­el­ling, nor did it fol­low up with foren­sic checks on his com­put­ers or stor­age, he said.

“That to me was the most aston­ish­ing thing. They wait­ed two years and did absolute­ly noth­ing to check that the data was delet­ed. All they asked me to do was tick a box on a form and post it back.”
...

And, again, Face­book denied it’s data based passed along to Cam­bridge Ana­lyt­i­ca when ques­tioned by both the US Con­gress and UK Par­lia­ment:

...
Paul-Olivi­er Dehaye, a data pro­tec­tion spe­cial­ist, who spear­head­ed the inves­tiga­tive efforts into the tech giant, said: “Face­book has denied and denied and denied this. It has mis­led MPs and con­gres­sion­al inves­ti­ga­tors and it’s failed in its duties to respect the law.

“It has a legal oblig­a­tion to inform reg­u­la­tors and indi­vid­u­als about this data breach, and it hasn’t. It’s failed time and time again to be open and trans­par­ent.”

A major­i­ty of Amer­i­can states have laws requir­ing noti­fi­ca­tion in some cas­es of data breach, includ­ing Cal­i­for­nia, where Face­book is based.
...

And not how Face­book now admits Alek­san­dr Kogan did indeed get the data legal­ly. It just was­n’t used prop­er­ly. It’s why Face­book is say­ing it should­n’t be called a “data breach”: because it was­n’t a breach because the data was obtained prop­er­ly:

...
Face­book denies that the har­vest­ing of tens of mil­lions of pro­files by GSR and Cam­bridge Ana­lyt­i­ca was a data breach. It said in a state­ment that Kogan “gained access to this infor­ma­tion in a legit­i­mate way and through the prop­er chan­nels” but “did not sub­se­quent­ly abide by our rules” because he passed the infor­ma­tion on to third par­ties.

Face­book said it removed the app in 2015 and required cer­ti­fi­ca­tion from every­one with copies that the data had been destroyed, although the let­ter to Wylie did not arrive until the sec­ond half of 2016. “We are com­mit­ted to vig­or­ous­ly enforc­ing our poli­cies to pro­tect people’s infor­ma­tion. We will take what­ev­er steps are required to see that this hap­pens,” Paul Gre­w­al, Facebook’s vice-pres­i­dent, said in a state­ment. The com­pa­ny is now inves­ti­gat­ing reports that not all data had been delet­ed.
...

But Alek­san­dr Kogan isn’t sim­ply argu­ing that he did noth­ing wrong when he obtained that Face­book data via his app. Kogan also argues that he had a “close work­ing rela­tion­ship” with Face­book, which has grant­ed him per­mis­sion for his apps, and every­thing he did with the data was legal. So Alek­san­dr Kogan’s sto­ry is quite notable because, again, as we’ll see below, there is evi­dence that his sto­ry is clos­est to the truth of all the sto­ries we’re hear­ing: that Face­book was total­ly fine with Kogan’s apps obtain­ing the pri­vate data of mil­lions of Face­book friends. And Face­book was per­fect­ly fine with how that data was used or was at least con­scious­ly try­ing to not know how the data might be mis­used. That’s the pic­ture that’s going to emerge so keep that in mind when Kogan asserts that he had a “close work­ing rela­tion­ship” with Face­book. He prob­a­bly did based on avail­able evi­dence:

...
Kogan, who has pre­vi­ous­ly unre­port­ed links to a Russ­ian uni­ver­si­ty and took Russ­ian grants for research, had a licence from Face­book to col­lect pro­file data, but it was for research pur­pos­es only. So when he hoovered up infor­ma­tion for the com­mer­cial ven­ture, he was vio­lat­ing the company’s terms. Kogan main­tains every­thing he did was legal, and says he had a “close work­ing rela­tion­ship” with Face­book, which had grant­ed him per­mis­sion for his apps.
...

Kogan main­tains every­thing he did was legal, and guess what? It prob­a­bly was legal. That’s part of the scan­dal here.

And regard­ing those tes­ti­mony’s by Cam­bridge Ana­lyt­i­ca’s now-for­mer CEO Alexan­der Nix that the com­pa­ny nev­er worked with Face­book, note how the Observ­er got to see a copy of the con­tract Cam­bridge Ana­lyt­i­ca entered into with Kogan’s GSR and the con­tract was entire­ly premised on har­vest­ing and pro­cess­ing the Face­book data. Which, again, hints at the like­li­hood that they thought what they were doing at the time (2014) was com­plete­ly legal. They talked about it in the con­tract:

...
The Observ­er has seen a con­tract dat­ed 4 June 2014, which con­firms SCL, an affil­i­ate of Cam­bridge Ana­lyt­i­ca, entered into a com­mer­cial arrange­ment with GSR, entire­ly premised on har­vest­ing and pro­cess­ing Face­book data. Cam­bridge Ana­lyt­i­ca spent near­ly $1m on data col­lec­tion, which yield­ed more than 50 mil­lion indi­vid­ual pro­files that could be matched to elec­toral rolls. It then used the test results and Face­book data to build an algo­rithm that could analyse indi­vid­ual Face­book pro­files and deter­mine per­son­al­i­ty traits linked to vot­ing behav­iour.

...

“The ulti­mate prod­uct of the train­ing set is cre­at­ing a ‘gold stan­dard’ of under­stand­ing per­son­al­i­ty from Face­book pro­file infor­ma­tion,” the con­tract spec­i­fies. It promis­es to cre­ate a data­base of 2 mil­lion “matched” pro­files, iden­ti­fi­able and tied to elec­toral reg­is­ters, across 11 states, but with room to expand much fur­ther.

...

Cam­bridge Ana­lyt­i­ca said that its con­tract with GSR stip­u­lat­ed that Kogan should seek informed con­sent for data col­lec­tion and it had no rea­son to believe he would not.

GSR was “led by a seem­ing­ly rep­utable aca­d­e­m­ic at an inter­na­tion­al­ly renowned insti­tu­tion who made explic­it con­trac­tu­al com­mit­ments to us regard­ing its legal author­i­ty to license data to SCL Elec­tions”, a com­pa­ny spokesman said.
...

““The ulti­mate prod­uct of the train­ing set is cre­at­ing a ‘gold stan­dard’ of under­stand­ing per­son­al­i­ty from Face­book pro­file infor­ma­tion,” the con­tract spec­i­fies. It promis­es to cre­ate a data­base of 2 mil­lion “matched” pro­files, iden­ti­fi­able and tied to elec­toral reg­is­ters, across 11 states, but with room to expand much fur­ther.”

A con­tract to cre­ate a ‘gold stan­dard’ of 2 mil­lion Face­book accounts that are ‘matched’ to real life vot­ers for the use of “under­stand­ing per­son­al­i­ty from Face­book pro­file infor­ma­tion.” That was the actu­al con­tract Kogan had with Cam­bridge Ana­lyt­i­ca. All for the pur­pose of devel­op­ing a sys­tem that would allow Cam­bridge Ana­lyt­i­ca to infer your inner demons from your Face­book pro­file and then manip­u­late them.

So it’s worth not­ing how the app per­mis­sions set­up Face­book allowed from 2007–2014 of let­ting app devel­op­ers col­lect Face­book pro­file infor­ma­tion of the peo­ple who use their apps and their friends cre­at­ed this amaz­ing arrange­ment where app devel­op­ers could gen­er­ate a ‘gold stan­dard’ of of peo­ple using apps and a test set from all their friends. If the goal was get­ting peo­ple to encour­age their friends to down­load an app that would have been a very use­ful data set. But it would of course also have been an incred­i­bly use­ful data set for any­one who want­ed to col­lect the pro­file infor­ma­tion of Face­book users. Because, again, as we’re going to see, a Face­book whis­tle-blow­er is claim­ing that Face­book user pro­file infor­ma­tion was rou­tine­ly hand­ed out to app devel­op­ers.

So if an app devel­op­er want­ed to exper­i­ment on, say, how to use that avail­able Face­book pro­file infor­ma­tion to manip­u­late peo­ple, get­ting a ‘gold stan­dard’ of peo­ple to take a psy­cho­log­i­cal pro­file sur­vey would be an impor­tant step in car­ry­ing out that exper­i­ment. Because those peo­ple who take your psy­cho­log­i­cal sur­vey form the data set you can use to train your algo­rithms that take Face­book pro­file infor­ma­tion as the input and cre­ate psy­cho­log­i­cal pro­file data as the out­put.

And that’s what Alek­san­dr Kogan’s app was doing: grab­bing psy­cho­log­i­cal infor­ma­tion from the sur­vey while simul­ta­ne­ous­ly grab­bing the Face­book pro­file data from the test-tak­ers, along with the Face­book pro­file data of all their friends. Kogan’s ‘gold stan­dard’ train­ing set was the peo­ple who actu­al­ly used his app and hand­ed over a bunch of per­son­al­i­ty infor­ma­tion from the sur­vey and the test set would have been the tens of mil­lions of friends whose data was also col­lect­ed. Since the goal of Cam­bridge Ana­lyt­i­ca was to infer per­son­al­i­ty char­ac­ter­is­tics from peo­ple’s Face­book pro­files, pair­ing the per­son­al­i­ty sur­veys from the ~270,000 peo­ple who took the app sur­vey to their Face­book pro­files allowed Cam­bridge Ana­lyt­i­ca to train their algo­rithms that guessed at per­son­al­i­ty char­ac­ter­is­tics from the Face­book pro­file infor­ma­tion. Then they had all the rest of the pro­file infor­ma­tion on the rest of the ~50 mil­lion peo­ple to apply those algo­rithms.

Recall how Trump’s 2016 cam­paign dig­i­tal direc­tor, Brad Parscale, curi­ous­ly downlplayed the util­i­ty of Cam­bridge Ana­lyt­i­ca’s data dur­ing inter­views where he was brag­ging about how they were using Face­book’s ad micro-tar­get­ing fea­tures to run “A/B test­ing on ste­ri­ods” on micro-tar­get­ed audi­ences i.e. strate­gi­cal­ly expos­ing micro-tar­get­ed Face­book audi­ences sets of ads that dif­fered in some spe­cif­ic way design to explore a par­tic­u­lar psy­cho­log­i­cal dimen­sion of that micro-audi­ence [21]. So it’s worth not­ing that the “A/B test­ing on steroids” Brad Parscale referred to was prob­a­bly focused on the ~30 mil­lion of that ~50 mil­lion set of peo­ple that Cam­bridge Ana­lyt­i­ca obtained a Face­book pro­file who could be matched back to real peo­ple. Those 30 mil­lion Face­book users that Cam­bridge Ana­lyt­i­ca had Face­book pro­file data on were the test set. And the algo­rithms designed to guess the psy­cho­log­i­cal make­up of peo­ple from their Face­book pro­files that Cam­bridge Ana­lyt­i­ca refined on the train­ing set of ~270,000 Face­book users who took the psy­cho­log­i­cal pro­files were like­ly unleashed on that test set of ~30 mil­lion peo­ple.

So when we find out that the Cam­bridge Ana­lyt­i­ca con­tract with Alek­san­dr Kogan’s GSR com­pa­ny includ­ed lan­guage like build­ing a “gold stan­dard”, keep in mind that this implied that there was a lot of test­ing to do after the algo­rith­mic refine­ments based on that gold stan­dard. And the ~30–50 mil­lion pro­files they col­lect­ed from the friends of the ~270,000 peo­ple who down­loaded Kogan’s app made for quite a test set.

Also keep in mind that the denials that Cam­bridge Ana­lyt­i­ca worked with Face­book data by for­mer CEO Alexan­der Nix aren’t the only laugh­able denials of Cam­bridge Ana­lyt­i­ca’s offi­cers. Any denials by Steve Ban­non and his lawyers that he knew about Cam­bridge Ana­lyt­i­ca’s use of Face­book pro­file data should also be seen laugh­able, start­ing with the denials from Ban­non’s lawyers that he knows noth­ing about what Wylie and oth­ers are claim­ing:

...
Steve Bannon’s lawyer said he had no com­ment because his client “knows noth­ing about the claims being assert­ed”. He added: “The first Mr Ban­non heard of these reports was from media inquiries in the past few days.” He direct­ed inquires to Nix.

Steve Ban­non: the Boss Who Knows Noth­ing (Or So He Says)

Steve Ban­non “knows noth­ing about the claims being assert­ed.” LOL! Yeah, well, not accord­ing to Christo­pher Wylie, who, in the fol­low­ing arti­cle, has some rather sig­nif­i­cant claims about the role Steve Ban­non in all this. Accord­ing to Wylie:

1. Steve Ban­non was the per­son over­see­ing the acqui­si­tion of Face­book data by Cam­bridge Ana­lyt­i­ca. As Wylie put it, “We had to get Ban­non to approve every­thing at this point. Ban­non was Alexan­der Nix’s boss.” Now, when Wylie says Ban­non was Nix’s boss, note that Ban­non served as vice pres­i­dent and sec­re­tary of Cam­bridge Ana­lyt­i­ca from June 2014 to August 2016. And Nix was CEO dur­ing this peri­od. So tech­ni­cal­ly Nix was the boss. But it sounds like Ban­non was effec­tive­ly the boss, accord­ing to Wylie.

2. Wylie acknowl­edges that it’s unclear whether Ban­non knew how Cam­bridge Ana­lyt­i­ca was obtain­ing the Face­book data. But Wylie does say that both Ban­non and Rebekah Mer­cer par­tic­i­pat­ed in con­fer­ence calls in 2014 in which plans to col­lect Face­book data were dis­cussed. And Ban­non “approved the data-col­lec­tion scheme we were propos­ing”. So if Ban­non and Mer­cer did­n’t know the details of how the pur­chase of mas­sive amounts of Face­book data took place that would be pret­ty remark­able. Remark­ably uncu­ri­ous, giv­en that acquir­ing this data was at the core of what the com­pa­ny was doing and they approved of the data-col­lec­tion scheme. A scheme that involved hav­ing Alek­san­dr Kogan set up a sep­a­rate com­pa­ny. That was the “scheme” Ban­non and Mer­cer would have had to approve so the ques­tion if they did­n’t real­ize that they were acquire this Face­book data using this “friend shar­ing” fea­ture Face­book made avail­able to app devel­op­ers that would have been a sig­nif­i­cant over­sight.

The arti­cle goes on to include a few more fun facts, like...

3. Cam­bridge Ana­lyt­i­ca was doing focus group tests on vot­ers in 2014 and iden­ti­fied many of the same under­ly­ing emo­tion­al sen­ti­ments in vot­ers that formed the core mes­sage behind Don­ald Trump’s cam­paign. In focus groups for the 2014 midterms, the firm found that vot­ers respond­ed to calls for build­ing a wall with Mex­i­co, “drain­ing the swamp” int Wash­ing­ton DC, and to thin­ly veiled forms of racism toward African Amer­i­cans called “race real­ism”. The firm also test­ed vot­er atti­tudes towards Russ­ian Pres­i­dent Vladimir Putin and dis­cov­ered that there’s a lot of Amer­i­cans who real­ly like the idea of a real­ly strong author­i­tar­i­an leader. Again, this was all dis­cov­ered before Trump even jumped into the race.

4. The Trump cam­paign reject­ed ear­ly over­tures to hire Cam­bridge Ana­lyt­i­ca, which sug­gests that Trump was actu­al­ly the top choice of the Mer­cers and Ban­non, ahead of Ted Cruz.

5. Cam­bridge Ana­lyt­i­ca CEO Alexan­der Nix was caught by Chan­nel 4 News in the UK boast­ing about the secre­cy of his firm, at one point stress­ing the need to set up a spe­cial email account that self-destruc­ts all mes­sages so that “there’s no evi­dence, there’s no paper trail, there’s noth­ing.”

So based on these alle­ga­tions, Steve Ban­non was close­ly involved in approval the var­i­ous schemes to acquire Face­book data and prob­a­bly using self-destruc­t­ing emails in the process [22]:

The Wash­ing­ton Post

Ban­non over­saw Cam­bridge Analytica’s col­lec­tion of Face­book data, accord­ing to for­mer employ­ee

By Craig Tim­berg, Kar­la Adam and Michael Kran­ish
March 20, 2018 at 7:53 PM

LONDON — Con­ser­v­a­tive strate­gist Stephen K. Ban­non over­saw Cam­bridge Analytica’s ear­ly efforts to col­lect troves of Face­book data as part of an ambi­tious pro­gram to build detailed pro­files of mil­lions of Amer­i­can vot­ers, a for­mer employ­ee of the data-sci­ence firm said Tues­day.

The 2014 effort was part of a high-tech form of vot­er per­sua­sion tout­ed by the com­pa­ny, which under Ban­non iden­ti­fied and test­ed the pow­er of anti-estab­lish­ment mes­sages that lat­er would emerge as cen­tral themes in Pres­i­dent Trump’s cam­paign speech­es, accord­ing to Chris Wylie, who left the com­pa­ny at the end of that year.

Among the mes­sages test­ed were “drain the swamp” and “deep state,” he said.

Cam­bridge Ana­lyt­i­ca, which worked for Trump’s 2016 cam­paign, is now fac­ing ques­tions about alleged uneth­i­cal prac­tices, includ­ing charges that the firm improp­er­ly han­dled the data of tens of mil­lions of Face­book users. On Tues­day, the company’s board announced that it was sus­pend­ing [23] its chief exec­u­tive, Alexan­der Nix, after British tele­vi­sion released secret record­ings that appeared to show him talk­ing about entrap­ping polit­i­cal oppo­nents.

More than three years before he served as Trump’s chief polit­i­cal strate­gist, Ban­non helped launch Cam­bridge Ana­lyt­i­ca with the finan­cial back­ing of the wealthy Mer­cer fam­i­ly as part of a broad­er effort to cre­ate a pop­ulist pow­er base [24]. Ear­li­er this year, the Mer­cers cut ties [25] with Ban­non after he was quot­ed mak­ing incen­di­ary com­ments about Trump and his fam­i­ly.

In an inter­view Tues­day with The Wash­ing­ton Post at his lawyer’s Lon­don office, Wylie said that Ban­non — while he was a top exec­u­tive at Cam­bridge Ana­lyt­i­ca and head of Bre­it­bart News — was deeply involved in the company’s strat­e­gy and approved spend­ing near­ly $1 mil­lion to acquire data, includ­ing Face­book pro­files, in 2014.

“We had to get Ban­non to approve every­thing at this point. Ban­non was Alexan­der Nix’s boss,” said Wylie, who was Cam­bridge Analytica’s research direc­tor. “Alexan­der Nix didn’t have the author­i­ty to spend that much mon­ey with­out approval.”

Ban­non, who served on the company’s board, did not respond to a request for com­ment. He served as vice pres­i­dent and sec­re­tary of Cam­bridge Ana­lyt­i­ca from June 2014 to August 2016, when he became chief exec­u­tive of Trump’s cam­paign, accord­ing to his pub­licly filed finan­cial dis­clo­sure. In 2017, he joined Trump in the White House as his chief strate­gist.

Ban­non received more than $125,000 in con­sult­ing fees from Cam­bridge Ana­lyt­i­ca in 2016 and owned “mem­ber­ship units” in the com­pa­ny worth between $1 mil­lion and $5 mil­lion, accord­ing to his finan­cial dis­clo­sure.

...

It is unclear whether Ban­non knew how Cam­bridge Ana­lyt­i­ca was obtain­ing the data, which alleged­ly was col­lect­ed through an app that was por­trayed as a tool for psy­cho­log­i­cal research but was then trans­ferred to the com­pa­ny.

Face­book has said that infor­ma­tion was improp­er­ly shared and that it request­ed the dele­tion of the data in 2015. Cam­bridge Ana­lyt­i­ca offi­cials said that they had done so, but Face­book said it received reports sev­er­al days ago that the data was not delet­ed.

Wylie said that both Ban­non and Rebekah Mer­cer, whose father, Robert Mer­cer, financed the com­pa­ny, par­tic­i­pat­ed in con­fer­ence calls in 2014 in which plans to col­lect Face­book data were dis­cussed, although Wylie acknowl­edged that it was not clear they knew the details of how the col­lec­tion took place.

Ban­non “approved the data-col­lec­tion scheme we were propos­ing,” Wylie said.

...

The data and analy­ses that Cam­bridge Ana­lyt­i­ca gen­er­at­ed in this time pro­vid­ed dis­cov­er­ies that would lat­er form the emo­tion­al­ly charged core of Trump’s pres­i­den­tial plat­form, said Wylie, whose dis­clo­sures in news reports over the past sev­er­al days have rocked both his one­time employ­er and Face­book.

“Trump wasn’t in our con­scious­ness at that moment; this was well before he became a thing,” Wylie said. “He wasn’t a client or any­thing.”

The year before Trump announced his pres­i­den­tial bid, the data firm already had found a high lev­el of alien­ation among young, white Amer­i­cans with a con­ser­v­a­tive bent.

In focus groups arranged to test mes­sages for the 2014 midterms, these vot­ers respond­ed to calls for build­ing a new wall to block the entry of ille­gal immi­grants, to reforms intend­ed the “drain the swamp” of Washington’s entrenched polit­i­cal com­mu­ni­ty and to thin­ly veiled forms of racism toward African Amer­i­cans called “race real­ism,” he recount­ed.

The firm also test­ed views of Russ­ian Pres­i­dent Vladimir Putin.

“The only for­eign thing we test­ed was Putin,” he said. “It turns out, there’s a lot of Amer­i­cans who real­ly like this idea of a real­ly strong author­i­tar­i­an leader and peo­ple were quite defen­sive in focus groups of Putin’s inva­sion of Crimea.”

The con­tro­ver­sy over Cam­bridge Analytica’s data col­lec­tion erupt­ed in recent days amid news reports that an app cre­at­ed by a Cam­bridge Uni­ver­si­ty psy­chol­o­gist, Alek­san­dr Kogan, accessed exten­sive per­son­al data of 50 mil­lion Face­book users. The app, called thi­sisy­our­dig­i­tal­life, was down­loaded by 270,000 users. Facebook’s pol­i­cy, which has since changed, allowed Kogan to also col­lect data —includ­ing names, home towns, reli­gious affil­i­a­tions and likes — on all of the Face­book “friends” of those users. Kogan shared that data with Cam­bridge Ana­lyt­i­ca for its grow­ing data­base on Amer­i­can vot­ers.

Face­book on Fri­day banned the par­ent com­pa­ny of Cam­bridge Ana­lyt­i­ca, Kogan and Wylie for improp­er­ly shar­ing that data.

The Fed­er­al Trade Com­mis­sion has opened an inves­ti­ga­tion [26] into Face­book to deter­mine whether the social media plat­form vio­lat­ed a 2011 con­sent decree gov­ern­ing its pri­va­cy poli­cies when it allowed the data col­lec­tion. And Wylie plans to tes­ti­fy [27] to Democ­rats on the House Intel­li­gence Com­mit­tee as part of their inves­ti­ga­tion of Russ­ian inter­fer­ence in the elec­tion, includ­ing pos­si­ble ties to the Trump cam­paign.

Mean­while, Britain’s Chan­nel 4 News aired a video Tues­day in which Nix was shown boast­ing about his work for Trump. He seemed to high­light his firm’s secre­cy, at one point stress­ing the need to set up a spe­cial email account that self-destruc­ts all mes­sages so that “there’s no evi­dence, there’s no paper trail, there’s noth­ing.”

The com­pa­ny said in a state­ment that Nix’s com­ments “do not rep­re­sent the val­ues or oper­a­tions of the firm and his sus­pen­sion reflects the seri­ous­ness with which we view this vio­la­tion.”

Nix could not be reached for com­ment.

Cam­bridge Ana­lyt­i­ca was set up as a U.S. affil­i­ate of British-based SCL Group, which had a wide range of gov­ern­men­tal clients glob­al­ly, in addi­tion to its polit­i­cal work.

Wylie said that Ban­non and Nix first met in 2013, the same year that Wylie — a young data whiz with some polit­i­cal expe­ri­ence in Britain and Cana­da — was work­ing for SCL Group. Ban­non and Wylie met soon after and hit it off in con­ver­sa­tions about cul­ture, elec­tions and how to spread ideas using tech­nol­o­gy.

Ban­non, Wylie, Nix, Rebekah Mer­cer and Robert Mer­cer met in Rebekah Mercer’s Man­hat­tan apart­ment in the fall of 2013, strik­ing a deal in which Robert Mer­cer would fund the cre­ation of Cam­bridge Ana­lyt­i­ca with $10 mil­lion, with the hope of shap­ing the con­gres­sion­al elec­tions a year lat­er, accord­ing to Wylie. Robert Mer­cer, in par­tic­u­lar, seemed trans­fixed by the group’s plans to har­ness and ana­lyze data, he recalled.

The Mer­cers were keen to cre­ate a U.S.-based busi­ness to avoid bad optics and vio­lat­ing U.S. cam­paign finance rules, Wylie said. “They want­ed to cre­ate an Amer­i­can brand,” he said.

The young com­pa­ny strug­gled to quick­ly deliv­er on its promis­es, Wiley said. Wide­ly avail­able infor­ma­tion from com­mer­cial data bro­kers pro­vid­ed people’s names, address­es, shop­ping habits and more, but failed to dis­tin­guish on more fine-grained mat­ters of per­son­al­i­ty that might affect polit­i­cal views.

Cam­bridge Ana­lyt­i­ca ini­tial­ly worked for 2016 Repub­li­can can­di­date Sen. Ted Cruz (Tex.), who was backed by the Mer­cers. The Trump cam­paign had reject­ed ear­ly over­tures to hire Cam­bridge Ana­lyt­i­ca, and Trump him­self said in May 2016 that he “always felt” that the use of vot­er data was “over­rat­ed.”

After Cruz fad­ed, the Mer­cers switched their alle­giance to Trump and pitched their ser­vices to Trump’s dig­i­tal direc­tor, Brad Parscale. The company’s hir­ing was approved by Trump’s son-in-law, Jared Kush­n­er, who was infor­mal­ly help­ing to man­age the cam­paign with a focus on dig­i­tal strat­e­gy.

Kush­n­er said in an inter­view [28] with Forbes mag­a­zine that the cam­paign “found that Face­book and dig­i­tal tar­get­ing were the most effec­tive ways to reach the audi­ences. ...We brought in Cam­bridge Ana­lyt­i­ca.” Kush­n­er said he “built” a data hub for the cam­paign “which nobody knew about, until towards the end.”

Kushner’s spokesman and lawyer both declined to com­ment Tues­day.

Two weeks before Elec­tion Day, Nix told a Post reporter [29] at the company’s New York City office that his com­pa­ny could “deter­mine the per­son­al­i­ty of every sin­gle adult in the Unit­ed States of Amer­i­ca.”

The claim was wide­ly ques­tioned, and the Trump cam­paign lat­er said that it didn’t rely on psy­cho­graph­ic data from Cam­bridge Ana­lyt­i­ca. Instead, the cam­paign said that it used a vari­ety of oth­er dig­i­tal infor­ma­tion to iden­ti­fy prob­a­ble sup­port­ers.

Parscale said in a Post inter­view in Octo­ber 2016 that he had not “opened the hood” on Cam­bridge Analytica’s method­ol­o­gy, and said he got much of his data from the Repub­li­can Nation­al Com­mit­tee. Parscale declined to com­ment Tues­day. He has pre­vi­ous­ly said that the Trump cam­paign did not use any psy­cho­graph­ic data from Cam­bridge Ana­lyt­i­ca.

Cam­bridge Analytica’s par­ent com­pa­ny, SCL Group, has an ongo­ing con­tract with the State Department’s Glob­al Engage­ment Cen­ter. The com­pa­ny was paid almost $500,000 to inter­view peo­ple over­seas to under­stand the mind-set of Islamist mil­i­tants as part of an effort to counter their online pro­pa­gan­da and block recruits.

Heather Nauert, the act­ing under­sec­re­tary for pub­lic diplo­ma­cy, said Tues­day that the con­tract was signed in Novem­ber 2016, under the Oba­ma admin­is­tra­tion, and has not expired yet. In pub­lic records, the con­tract is dat­ed in Feb­ru­ary 2017, and the rea­son for the dis­crep­an­cy was not clear. Nauert said that the State Depart­ment had signed oth­er con­tracts with SCL Group in the past.

———-

“Ban­non over­saw Cam­bridge Analytica’s col­lec­tion of Face­book data, accord­ing to for­mer employ­ee” by Craig Tim­berg, Kar­la Adam and Michael Kran­ish; The Wash­ing­ton Post; 03/20/2018 [22]

“Con­ser­v­a­tive strate­gist Stephen K. Ban­non over­saw Cam­bridge Analytica’s ear­ly efforts to col­lect troves of Face­book data as part of an ambi­tious pro­gram to build detailed pro­files of mil­lions of Amer­i­can vot­ers, a for­mer employ­ee of the data-sci­ence firm said Tues­day.”

Steve Ban­non over­saw Cam­bridge Analytica’s ear­ly efforts to col­lect troves of Face­book data. That’s what Christo­pher Wylie claims, and giv­en Ban­non’s role as vice pres­i­dent of the com­pa­ny it’s not, on its face, an out­landish claim. And Ban­non appar­ent­ly approved the spend­ing of near­ly $1 mil­lion to acquire that Face­book data in 2014. Because, accord­ing to Wylie, Alexan­der Nix did­n’t actu­al­ly have per­mis­sion to spend that kind of mon­ey with­out approval. Ban­non, on the hand, did have per­mis­sion to make those kinds of expen­di­ture approvals. That’s how high up Ban­non was at that com­pa­ny even though he was tech­ni­cal­ly the vice pres­i­dent while Nix was the CEO:

...
In an inter­view Tues­day with The Wash­ing­ton Post at his lawyer’s Lon­don office, Wylie said that Ban­non — while he was a top exec­u­tive at Cam­bridge Ana­lyt­i­ca and head of Bre­it­bart News — was deeply involved in the company’s strat­e­gy and approved spend­ing near­ly $1 mil­lion to acquire data, includ­ing Face­book pro­files, in 2014.

“We had to get Ban­non to approve every­thing at this point. Ban­non was Alexan­der Nix’s boss,” said Wylie, who was Cam­bridge Analytica’s research direc­tor. “Alexan­der Nix didn’t have the author­i­ty to spend that much mon­ey with­out approval.”

Ban­non, who served on the company’s board, did not respond to a request for com­ment. He served as vice pres­i­dent and sec­re­tary of Cam­bridge Ana­lyt­i­ca from June 2014 to August 2016, when he became chief exec­u­tive of Trump’s cam­paign, accord­ing to his pub­licly filed finan­cial dis­clo­sure. In 2017, he joined Trump in the White House as his chief strate­gist.
...

“We had to get Ban­non to approve every­thing at this point. Ban­non was Alexan­der Nix’s boss...Alexander Nix didn’t have the author­i­ty to spend that much mon­ey with­out approval.””

And while Wylie acknowl­edges that unclear whether Ban­non knew how Cam­bridge Ana­lyt­i­ca was obtain­ing the data, Wylie does assert that both Ban­non and Rebekah Mer­cer par­tic­i­pat­ed in con­fer­ence calls in 2014 in which plans to col­lect Face­book data were dis­cussed. And, gen­er­al­ly speak­ing, if Ban­non was approval $1 mil­lion expen­di­tures on acquir­ing Face­book data he prob­a­bly sat in on at least one meet­ing where they described how they were plan­ning on actu­al­ly get­ting the data by spend­ing on that mon­ey. Don’t for­get the scheme involved pay­ing indi­vid­u­als small amounts of mon­ey to take the psy­cho­log­i­cal sur­vey on Kogan’s app, so at a min­i­mum you would expect Ban­non to know about how these apps were going to result in the gath­er­ing of Face­book pro­file infor­ma­tion:

...
It is unclear whether Ban­non knew how Cam­bridge Ana­lyt­i­ca was obtain­ing the data, which alleged­ly was col­lect­ed through an app that was por­trayed as a tool for psy­cho­log­i­cal research but was then trans­ferred to the com­pa­ny.

Face­book has said that infor­ma­tion was improp­er­ly shared and that it request­ed the dele­tion of the data in 2015. Cam­bridge Ana­lyt­i­ca offi­cials said that they had done so, but Face­book said it received reports sev­er­al days ago that the data was not delet­ed.

Wylie said that both Ban­non and Rebekah Mer­cer, whose father, Robert Mer­cer, financed the com­pa­ny, par­tic­i­pat­ed in con­fer­ence calls in 2014 in which plans to col­lect Face­book data were dis­cussed, although Wylie acknowl­edged that it was not clear they knew the details of how the col­lec­tion took place.

Ban­non “approved the data-col­lec­tion scheme we were propos­ing,” Wylie said.
...

What’s Ban­non hid­ing by claim­ing igno­rance? Well, that’s a good ques­tion after Britain’s Chan­nel 4 News aired a video Tues­day in which Nix was high­light­ing his firm’s secre­cy, includ­ing the need to set up a spe­cial email account that self-destruc­ts all mes­sages so that “there’s no evi­dence, there’s no paper trail, there’s noth­ing”:

...
Mean­while, Britain’s Chan­nel 4 News aired a video Tues­day in which Nix was shown boast­ing about his work for Trump. He seemed to high­light his firm’s secre­cy, at one point stress­ing the need to set up a spe­cial email account that self-destruc­ts all mes­sages so that “there’s no evi­dence, there’s no paper trail, there’s noth­ing.”

The com­pa­ny said in a state­ment that Nix’s com­ments “do not rep­re­sent the val­ues or oper­a­tions of the firm and his sus­pen­sion reflects the seri­ous­ness with which we view this vio­la­tion.”
...

Self-destruc­t­ing emails. That’s not sus­pi­cious or any­thing.

And note how Cam­bridge Ana­lyt­i­ca was appar­ent­ly already hon­ing in on a very ‘Trumpian’ mes­sage in 2014, long before Trump was on the radar:

...
The data and analy­ses that Cam­bridge Ana­lyt­i­ca gen­er­at­ed in this time pro­vid­ed dis­cov­er­ies that would lat­er form the emo­tion­al­ly charged core of Trump’s pres­i­den­tial plat­form, said Wylie, whose dis­clo­sures in news reports over the past sev­er­al days have rocked both his one­time employ­er and Face­book.

“Trump wasn’t in our con­scious­ness at that moment; this was well before he became a thing,” Wylie said. “He wasn’t a client or any­thing.”

The year before Trump announced his pres­i­den­tial bid, the data firm already had found a high lev­el of alien­ation among young, white Amer­i­cans with a con­ser­v­a­tive bent.

In focus groups arranged to test mes­sages for the 2014 midterms, these vot­ers respond­ed to calls for build­ing a new wall to block the entry of ille­gal immi­grants, to reforms intend­ed the “drain the swamp” of Washington’s entrenched polit­i­cal com­mu­ni­ty and to thin­ly veiled forms of racism toward African Amer­i­cans called “race real­ism,” he recount­ed.

The firm also test­ed views of Russ­ian Pres­i­dent Vladimir Putin.

“The only for­eign thing we test­ed was Putin,” he said. “It turns out, there’s a lot of Amer­i­cans who real­ly like this idea of a real­ly strong author­i­tar­i­an leader and peo­ple were quite defen­sive in focus groups of Putin’s inva­sion of Crimea.”
...

Intrigu­ing­ly, giv­en these ear­ly Trumpian find­ings in their 2014 vot­er research, it appears that the Trump cam­paign turned down ear­ly over­tures to hire Cam­bridge Ana­lyt­i­ca, which sug­gests that Trump real­ly was the top pref­er­ence for Ban­non and the Mer­cers, not Ted Cruz:

...
Cam­bridge Ana­lyt­i­ca ini­tial­ly worked for 2016 Repub­li­can can­di­date Sen. Ted Cruz (Tex.), who was backed by the Mer­cers. The Trump cam­paign had reject­ed ear­ly over­tures to hire Cam­bridge Ana­lyt­i­ca, and Trump him­self said in May 2016 that he “always felt” that the use of vot­er data was “over­rat­ed.”
...

And as the arti­cle reminds us, the Trump cam­paign has com­plete­ly denied EVER using Cam­bridge Ana­lyt­i­ca’s data [21]. Brad Parscale, Trump’s dig­i­tal direc­tor, claimed he got all the data they were work­ing with from the Repub­li­can Nation­al Com­mit­tee:

...
Two weeks before Elec­tion Day, Nix told a Post reporter [29] at the company’s New York City office that his com­pa­ny could “deter­mine the per­son­al­i­ty of every sin­gle adult in the Unit­ed States of Amer­i­ca.”

The claim was wide­ly ques­tioned, and the Trump cam­paign lat­er said that it didn’t rely on psy­cho­graph­ic data from Cam­bridge Ana­lyt­i­ca. Instead, the cam­paign said that it used a vari­ety of oth­er dig­i­tal infor­ma­tion to iden­ti­fy prob­a­ble sup­port­ers.

Parscale said in a Post inter­view in Octo­ber 2016 that he had not “opened the hood” on Cam­bridge Analytica’s method­ol­o­gy, and said he got much of his data from the Repub­li­can Nation­al Com­mit­tee. Parscale declined to com­ment Tues­day. He has pre­vi­ous­ly said that the Trump cam­paign did not use any psy­cho­graph­ic data from Cam­bridge Ana­lyt­i­ca.
...

And that denial by Parscale rais­es an obvi­ous ques­tion: when Parscale claims they only used data from the RNC, it’s clear­ly very pos­si­ble that he’s just straight up lying. But it’s also pos­si­ble that he’s lying while tech­ni­cal­ly telling the truth. Because if Cam­bridge Ana­lyt­i­ca gave its data to the RNC, it’s pos­si­ble the Trump cam­paign acquired the Cam­gridge Ana­lyt­i­ca data from the RNC at that point, giv­ing the cam­paign a degree of deni­a­bil­i­ty about the use of such scan­dalous­ly acquired data if the sto­ry of it ever became pub­lic. Like now.

Don’t for­get that data of this nature would have been poten­tial­ly use­ful for EVERY 2016 race, not just the pres­i­den­tial cam­paign. So if Ban­non and Mer­cer were intent on help­ing Repub­li­cans win across the board, hand­ing that data over to the RNC would have just made sense.

Also don’t for­get that the New York Times was shown unen­crypt­ed copies of the Face­book data col­lect­ed by Cam­bridge Ana­lyt­i­ca. If the New York Times saw this data, odds are the RNC has too. And who knows who else.

Face­book’s Sandy Parak­i­las Blows an “Utter­ly Hor­ri­fy­ing” Whis­tle

It all rais­es the ques­tion of whether or not the Repub­li­can Nation­al Com­mit­tee now pos­sess all that Cam­bridge Ana­lyt­i­ca data/Facebook data right now. And that brings us to per­haps the most scan­dalous arti­cle of all that we’re going to look at. It’s about Sandy Parak­i­las, the plat­form oper­a­tions man­ag­er at Face­book respon­si­ble for polic­ing data breach­es by third-par­ty soft­ware devel­op­ers between 2011 and 2012 who is now a whis­tle-blow­er about exact­ly the kind of “friend’s per­mis­sion” loop­hole Cam­bridge Ana­lyt­i­ca exploit­ed. And as the fol­low­ing arti­cle makes hor­rif­i­cal­ly clear:

1. It’s not just Cam­bridge Ana­lyt­i­ca or the RNC that might pos­sess this trea­sure trove of per­son­al infor­ma­tion. It’s the entire data bro­ker­age indus­try that prob­a­bly has thi­er hands on this data. Along with any­one who has picked it up through the black mar­ket.

2. It was rel­a­tive­ly easy to write an app that could exploit this “friends per­mis­sions” fea­ture and start trawl­ing Face­book for pro­file data for app users and their friends. Any­one with basic app cod­ing skills could do it.

3. Parak­i­las esti­mates that per­haps hun­dreds of thou­sands of devel­op­ers like­ly exploit­ed exact­ly the same ‘for research pur­pos­es only’ loop­hole exploit­ed by Cam­bridge Ana­lyt­i­ca. And Face­book had no way of track­ing how this data was used by devel­op­ers once it left Face­book’s servers.

4. Parak­i­las sus­pects that this amount of data will inevitably end up in the black mar­ket mean­ing there is prob­a­bly a mas­sive amount of per­son­al­ly iden­ti­fi­able Face­book data just float­ing around for the entire mar­ket­ing indus­try and any­one else (like the GOP) to data mine.

5. Parak­i­las knew of many com­mer­cial apps that were using the same “friends per­mis­sion” fea­ture to grab Face­book pro­file data use it com­mer­cial pur­pos­es.

6. Face­book’s pol­i­cy of giv­ing devel­op­ers access to Face­book users’ friends’ data was sanc­tioned in the small print in Facebook’s terms and con­di­tions, and users could block such data shar­ing by chang­ing their set­tings. That appears to be part of the legal pro­tec­tion Face­book employed when it had this pol­i­cy: don’t com­plain, it’s in the fine print.

7. Per­haps most scan­dalous of all, Face­book took a 30% cut of pay­ments made through apps in exchange for giv­ing these app devel­op­ers access to Face­book user data. Yep, Face­book was effec­tive­ly sell­ing user data, but by struc­tur­ing the sale of this data as a 30% share of the pay­ments made through the app Face­book also cre­at­ed an incen­tive to help devel­op­ers max­i­mize the prof­its they made through the app. So Face­book lit­er­al­ly set up a sys­tem that incen­tivized itself to help app devel­op­ers make as much mon­ey as pos­si­ble off of the user data they were hand­ing over.

8. Aca­d­e­m­ic research from 2010, based on an analy­sis of 1,800 Face­books apps, con­clud­ed that around 11% of third-par­ty devel­op­ers request­ed data belong­ing to friends of users. So as a 2010, ~1 in 10 Face­book apps were using this app loop­hole to grab infor­ma­tion about both the users of the app and their friends.

9. While Cam­bridge Ana­lyt­i­ca was far from alone in exploit­ing this loop­hole, it was actu­al­ly one of the very last firms giv­en per­mis­sion to be allowed to do so. Which means that par­tic­u­lar data set col­lect­ed by Cam­bridge Ana­lyt­i­ca could be unique­ly valu­able sim­ply be being larg­er and con­tain­ing and more recent data than most oth­er data sets of this nature.

10. When Parak­i­las brought up these con­cerns to Face­book’s exec­u­tives and sug­gest­ed the com­pa­ny should proac­tive­ly “audit devel­op­ers direct­ly and see what’s going on with the data” he was dis­cour­aged from the approach. One Face­book exec­u­tive advised him against look­ing too deeply at how the data was being used, warn­ing him: “Do you real­ly want to see what you’ll find?” Parak­i­las said he inter­pret­ed the com­ment to mean that “Face­book was in a stronger legal posi­tion if it didn’t know about the abuse that was hap­pen­ing”

11. Short­ly after arriv­ing at the company’s Sil­i­con Val­ley head­quar­ters, Parak­i­las was told that any deci­sion to ban an app required the per­son­al approval of Mark Zucker­berg. Although the pol­i­cy was lat­er relaxed to make it eas­i­er to deal with rogue devel­op­ers. That said, rogue devel­op­ers were rarely dealt with.

12. When Face­book even­tu­al­ly phased out this “friends per­mis­sions” pol­i­cy for app devel­op­ers, it was like­ly done out of con­cerns over the com­mer­cial val­ue of all this data they were hand­ing out. Exec­u­tives were appar­ent­ly con­cerned that com­peti­tors were going to use this data to build their own social net­works.

So, as we can see, the entire saga of Cam­bridge Ana­lyt­i­ca’s scan­dalous acqui­si­tion of pri­vate Face­book pro­files on ~50 mil­lion Amer­i­cans is some­thing Face­book made rou­tine for devel­op­ers of all sorts from 2007–2014, which means this is far from a ‘Cam­bridge Ana­lyt­i­ca’ sto­ry. It’s a Face­book sto­ry about a mas­sive prob­lem Face­book cre­at­ed for itself (for its own prof­its) [30]:

The Guardian

‘Utter­ly hor­ri­fy­ing’: ex-Face­book insid­er says covert data har­vest­ing was rou­tine

Sandy Parak­i­las says numer­ous com­pa­nies deployed these tech­niques – like­ly affect­ing hun­dreds of mil­lions of users – and that Face­book looked the oth­er way

Paul Lewis in San Fran­cis­co
Tue 20 Mar 2018 07.46 EDT

Hun­dreds of mil­lions of Face­book users are like­ly to have had their pri­vate infor­ma­tion har­vest­ed by com­pa­nies that exploit­ed the same terms as the firm that col­lect­ed data and passed it on to Cam­bridge Ana­lyt­i­ca, accord­ing to a new whistle­blow­er.

Sandy Parak­i­las, the plat­form oper­a­tions man­ag­er at Face­book respon­si­ble for polic­ing data breach­es by third-par­ty soft­ware devel­op­ers between 2011 and 2012, told the Guardian he warned senior exec­u­tives at the com­pa­ny that its lax approach to data pro­tec­tion risked a major breach.

“My con­cerns were that all of the data that left Face­book servers to devel­op­ers could not be mon­i­tored by Face­book, so we had no idea what devel­op­ers were doing with the data,” he said.

Parak­i­las said Face­book had terms of ser­vice and set­tings that “peo­ple didn’t read or under­stand” and the com­pa­ny did not use its enforce­ment mech­a­nisms, includ­ing audits of exter­nal devel­op­ers, to ensure data was not being mis­used.

Parak­i­las, whose job was to inves­ti­gate data breach­es by devel­op­ers sim­i­lar to the one lat­er sus­pect­ed of Glob­al Sci­ence Research, which har­vest­ed tens of mil­lions of Face­book pro­files and pro­vid­ed the data to Cam­bridge Ana­lyt­i­ca, said the slew of recent dis­clo­sures had left him dis­ap­point­ed with his supe­ri­ors for not heed­ing his warn­ings.

“It has been painful watch­ing,” he said, “because I know that they could have pre­vent­ed it.”

Asked what kind of con­trol Face­book had over the data giv­en to out­side devel­op­ers, he replied: “Zero. Absolute­ly none. Once the data left Face­book servers there was not any con­trol, and there was no insight into what was going on.”

Parak­i­las said he “always assumed there was some­thing of a black mar­ket” for Face­book data that had been passed to exter­nal devel­op­ers. How­ev­er, he said that when he told oth­er exec­u­tives the com­pa­ny should proac­tive­ly “audit devel­op­ers direct­ly and see what’s going on with the data” he was dis­cour­aged from the approach.

He said one Face­book exec­u­tive advised him against look­ing too deeply at how the data was being used, warn­ing him: “Do you real­ly want to see what you’ll find?” Parak­i­las said he inter­pret­ed the com­ment to mean that “Face­book was in a stronger legal posi­tion if it didn’t know about the abuse that was hap­pen­ing”.

He added: “They felt that it was bet­ter not to know. I found that utter­ly shock­ing and hor­ri­fy­ing.”

...

Face­book did not respond to a request for com­ment on the infor­ma­tion sup­plied by Parak­i­las, but direct­ed the Guardian to a Novem­ber 2017 blog­post [31] in which the com­pa­ny defend­ed its data shar­ing prac­tices, which it said had “sig­nif­i­cant­ly improved” over the last five years.

“While it’s fair to crit­i­cise how we enforced our devel­op­er poli­cies more than five years ago, it’s untrue to sug­gest we didn’t or don’t care about pri­va­cy,” that state­ment said. “The facts tell a dif­fer­ent sto­ry.”

‘A major­i­ty of Face­book users’

Parak­i­las, 38, who now works as a prod­uct man­ag­er for Uber, is par­tic­u­lar­ly crit­i­cal of Facebook’s pre­vi­ous pol­i­cy of allow­ing devel­op­ers to access the per­son­al data of friends of peo­ple who used apps on the plat­form, with­out the knowl­edge or express con­sent of those friends.

That fea­ture, called friends per­mis­sion, was a boon to out­side soft­ware devel­op­ers who, from 2007 onwards, were giv­en per­mis­sion by Face­book to build quizzes and games – like the wide­ly pop­u­lar Far­mVille – that were host­ed on the plat­form.

The apps pro­lif­er­at­ed on Face­book in the years lead­ing up to the company’s 2012 ini­tial pub­lic offer­ing, an era when most users were still access­ing the plat­form via lap­tops and com­put­ers rather than smart­phones.

Face­book took a 30% cut of pay­ments made through apps, but in return enabled their cre­ators to have access to Face­book user data.

Parak­i­las does not know how many com­pa­nies sought friends per­mis­sion data before such access was ter­mi­nat­ed around mid-2014. How­ev­er, he said he believes tens or maybe even hun­dreds of thou­sands of devel­op­ers may have done so.

Parak­i­las esti­mates that “a major­i­ty of Face­book users” could have had their data har­vest­ed by app devel­op­ers with­out their knowl­edge. The com­pa­ny now has stricter pro­to­cols around the degree of access third par­ties have to data.

Parak­i­las said that when he worked at Face­book it failed to take full advan­tage of its enforce­ment mech­a­nisms, such as a clause that enables the social media giant to audit exter­nal devel­op­ers who mis­use its data.

Legal action against rogue devel­op­ers or moves to ban them from Face­book were “extreme­ly rare”, he said, adding: “In the time I was there, I didn’t see them con­duct a sin­gle audit of a developer’s sys­tems.”

Face­book announced on Mon­day that it had hired a dig­i­tal foren­sics firm to con­duct an audit of Cam­bridge Ana­lyt­i­ca. The deci­sion comes more than two years after Face­book was made aware of the report­ed data breach.

Dur­ing the time he was at Face­book, Parak­i­las said the com­pa­ny was keen to encour­age more devel­op­ers to build apps for its plat­form and “one of the main ways to get devel­op­ers inter­est­ed in build­ing apps was through offer­ing them access to this data”. Short­ly after arriv­ing at the company’s Sil­i­con Val­ley head­quar­ters he was told that any deci­sion to ban an app required the per­son­al approval of the chief exec­u­tive, Mark Zucker­berg, although the pol­i­cy was lat­er relaxed to make it eas­i­er to deal with rogue devel­op­ers.

While the pre­vi­ous pol­i­cy of giv­ing devel­op­ers access to Face­book users’ friends’ data was sanc­tioned in the small print in Facebook’s terms and con­di­tions, and users could block such data shar­ing by chang­ing their set­tings, Parak­i­las said he believed the pol­i­cy was prob­lem­at­ic.

“It was well under­stood in the com­pa­ny that that pre­sent­ed a risk,” he said. “Face­book was giv­ing data of peo­ple who had not autho­rised the app them­selves, and was rely­ing on terms of ser­vice and set­tings that peo­ple didn’t read or under­stand.”

It was this fea­ture that was exploit­ed by Glob­al Sci­ence Research, and the data pro­vid­ed to Cam­bridge Ana­lyt­i­ca in 2014. GSR was run by the Cam­bridge Uni­ver­si­ty psy­chol­o­gist Alek­san­dr Kogan, who built an app that was a per­son­al­i­ty test for Face­book users.

The test auto­mat­i­cal­ly down­loaded the data of friends of peo­ple who took the quiz, osten­si­bly for aca­d­e­m­ic pur­pos­es. Cam­bridge Ana­lyt­i­ca has denied know­ing the data was obtained improp­er­ly, and Kogan main­tains he did noth­ing ille­gal and had a “close work­ing rela­tion­ship” with Face­book.

While Kogan’s app only attract­ed around 270,000 users (most of whom were paid to take the quiz), the com­pa­ny was then able to exploit the friends per­mis­sion fea­ture to quick­ly amass data per­tain­ing to more than 50 mil­lion Face­book users.

“Kogan’s app was one of the very last to have access to friend per­mis­sions,” Parak­i­las said, adding that many oth­er sim­i­lar apps had been har­vest­ing sim­i­lar quan­ti­ties of data for years for com­mer­cial pur­pos­es. Aca­d­e­m­ic research from 2010, based on an analy­sis of 1,800 Face­books apps [32], con­clud­ed that around 11% of third-par­ty devel­op­ers request­ed data belong­ing to friends of users.

If those fig­ures were extrap­o­lat­ed, tens of thou­sands of apps, if not more, were like­ly to have sys­tem­at­i­cal­ly culled “pri­vate and per­son­al­ly iden­ti­fi­able” data belong­ing to hun­dreds of mil­lions of users, Parak­i­las said.

The ease with which it was pos­si­ble for any­one with rel­a­tive­ly basic cod­ing skills to cre­ate apps and start trawl­ing for data was a par­tic­u­lar con­cern, he added.

Parak­i­las said he was unsure why Face­book stopped allow­ing devel­op­ers to access friends data around mid-2014, rough­ly two years after he left the com­pa­ny. How­ev­er, he said he believed one rea­son may have been that Face­book exec­u­tives were becom­ing aware that some of the largest apps were acquir­ing enor­mous troves of valu­able data.

He recalled con­ver­sa­tions with exec­u­tives who were ner­vous about the com­mer­cial val­ue of data being passed to oth­er com­pa­nies.

“They were wor­ried that the large app devel­op­ers were build­ing their own social graphs, mean­ing they could see all the con­nec­tions between these peo­ple,” he said. “They were wor­ried that they were going to build their own social net­works.”

‘They treat­ed it like a PR exer­cise’

Parak­i­las said he lob­bied inter­nal­ly at Face­book for “a more rig­or­ous approach” to enforc­ing data pro­tec­tion, but was offered lit­tle sup­port. His warn­ings includ­ed a Pow­er­Point pre­sen­ta­tion he said he deliv­ered to senior exec­u­tives in mid-2012 “that includ­ed a map of the vul­ner­a­bil­i­ties for user data on Facebook’s plat­form”.

“I includ­ed the pro­tec­tive mea­sures that we had tried to put in place, where we were exposed, and the kinds of bad actors who might do mali­cious things with the data,” he said. “On the list of bad actors I includ­ed for­eign state actors and data bro­kers.”

Frus­trat­ed at the lack of action, Parak­i­las left Face­book in late 2012. “I didn’t feel that the com­pa­ny treat­ed my con­cerns seri­ous­ly. I didn’t speak out pub­licly for years out of self-inter­est, to be frank.”

That changed, Parak­i­las said, when he heard the con­gres­sion­al tes­ti­mo­ny giv­en by Face­book lawyers to Sen­ate and House inves­ti­ga­tors in late 2017 about Russia’s attempt to sway the pres­i­den­tial elec­tion. “They treat­ed it like a PR exer­cise,” he said. “They seemed to be entire­ly focused on lim­it­ing their lia­bil­i­ty and expo­sure rather than help­ing the coun­try address a nation­al secu­ri­ty issue.”

It was at that point that Parak­i­las decid­ed to go pub­lic with his con­cerns, writ­ing an opin­ion arti­cle in the New York Times [33] that said Face­book could not be trust­ed to reg­u­late itself. Since then, Parak­i­las has become an advis­er to the Cen­ter for Humane Tech­nol­o­gy [34], which is run by Tris­tan Har­ris, a for­mer Google employ­ee turned whistle­blow­er on the indus­try.

———-

“ ‘Utter­ly hor­ri­fy­ing’: ex-Face­book insid­er says covert data har­vest­ing was rou­tine” by Paul Lewis; The Guardian; 03/20/2018 [30]

“Sandy Parak­i­las, the plat­form oper­a­tions man­ag­er at Face­book respon­si­ble for polic­ing data breach­es by third-par­ty soft­ware devel­op­ers between 2011 and 2012, told the Guardian he warned senior exec­u­tives at the com­pa­ny that its lax approach to data pro­tec­tion risked a major breach.”

The plat­form oper­a­tions man­ag­er at Face­book respon­si­ble for polic­ing data breach­es by third-par­ty soft­ware devel­op­ers between 2011 and 2012: That’s who is mak­ing these claims. In oth­er words, Sandy Parak­i­las is indeed some­one who should be inti­mate­ly famil­iar with Face­book’s poli­cies of hand­ing user data over to app devel­op­ers because it was his job to ensure that data was­n’t breached.

And as Parak­i­las makes clear, he was­n’t actu­al­ly able to do his job. When the data left Face­book’s servers after get­ting hand­ed over to app devel­op­er Face­book had no idea what devel­op­ers were doing with the data and appar­ent­ly no inter­est in learn­ing:

...
“My con­cerns were that all of the data that left Face­book servers to devel­op­ers could not be mon­i­tored by Face­book, so we had no idea what devel­op­ers were doing with the data,” he said.

Parak­i­las said Face­book had terms of ser­vice and set­tings that “peo­ple didn’t read or under­stand” and the com­pa­ny did not use its enforce­ment mech­a­nisms, includ­ing audits of exter­nal devel­op­ers, to ensure data was not being mis­used.

Parak­i­las, whose job was to inves­ti­gate data breach­es by devel­op­ers sim­i­lar to the one lat­er sus­pect­ed of Glob­al Sci­ence Research, which har­vest­ed tens of mil­lions of Face­book pro­files and pro­vid­ed the data to Cam­bridge Ana­lyt­i­ca, said the slew of recent dis­clo­sures had left him dis­ap­point­ed with his supe­ri­ors for not heed­ing his warn­ings.

“It has been painful watch­ing,” he said, “because I know that they could have pre­vent­ed it.”

Asked what kind of con­trol Face­book had over the data giv­en to out­side devel­op­ers, he replied: “Zero. Absolute­ly none. Once the data left Face­book servers there was not any con­trol, and there was no insight into what was going on.”
...

And this com­plete­ly lack of over­sight by Face­book led Parak­i­las to assume there was “some­thing of a black mar­ket” for that Face­book data. But when he expressed these con­cerns with fel­low exec­u­tives he was warned not to look. Not know­ing how this data was being used was iron­i­cal­ly part of Face­book’s legal strat­e­gy, it seems:

...
Parak­i­las said he “always assumed there was some­thing of a black mar­ket” for Face­book data that had been passed to exter­nal devel­op­ers. How­ev­er, he said that when he told oth­er exec­u­tives the com­pa­ny should proac­tive­ly “audit devel­op­ers direct­ly and see what’s going on with the data” he was dis­cour­aged from the approach.

He said one Face­book exec­u­tive advised him against look­ing too deeply at how the data was being used, warn­ing him: “Do you real­ly want to see what you’ll find?” Parak­i­las said he inter­pret­ed the com­ment to mean that “Face­book was in a stronger legal posi­tion if it didn’t know about the abuse that was hap­pen­ing”.

He added: “They felt that it was bet­ter not to know. I found that utter­ly shock­ing and hor­ri­fy­ing.”
...

“They felt that it was bet­ter not to know. I found that utter­ly shock­ing and hor­ri­fy­ing.”

Well, at least one exec­u­tive at Face­book was utter­ly shocked and hor­ri­fied by the “bet­ter not to know” pol­i­cy towards hand­ing per­son­al pri­vate infor­ma­tion over to devel­op­ers. And that one exec­u­tive, Parak­i­las, left the com­pa­ny and is now a whis­tle-blow­er.

And one of the things that made Parak­i­las par­tic­u­lar­ly con­cerned that this was wide­spread among app was the fact that it was so easy to cre­ate apps that could then just be released onto Face­book to trawl for Face­book pro­file data from users and their unwit­ting friends:

...
The ease with which it was pos­si­ble for any­one with rel­a­tive­ly basic cod­ing skills to cre­ate apps and start trawl­ing for data was a par­tic­u­lar con­cern, he added.
...

And while rogue app devel­op­ers were at times dealt with, it was exceed­ing­ly rare with Parak­i­las not wit­ness­ing a sin­gle audit of a devel­op­er’s sys­tems dur­ing his time there.

Even more alarm­ing is that Face­book was appar­ent­ly quite on encour­ag­ing app devel­op­ers to grab this Face­book pro­file data as an incen­tive to encour­age even more app devel­op. Apps were seen as so impor­tant to Face­book that Mark Zucker­berg him­self had to give his per­son­al approval to ban on app. And while that pol­i­cy was lat­er relaxed to not require Zucker­berg’s approval, it does­n’t sound like that pol­i­cy change actu­al­ly result­ed in more apps get­ting banned:

...
Parak­i­las said that when he worked at Face­book it failed to take full advan­tage of its enforce­ment mech­a­nisms, such as a clause that enables the social media giant to audit exter­nal devel­op­ers who mis­use its data.

Legal action against rogue devel­op­ers or moves to ban them from Face­book were “extreme­ly rare”, he said, adding: “In the time I was there, I didn’t see them con­duct a sin­gle audit of a developer’s sys­tems.”

Dur­ing the time he was at Face­book, Parak­i­las said the com­pa­ny was keen to encour­age more devel­op­ers to build apps for its plat­form and “one of the main ways to get devel­op­ers inter­est­ed in build­ing apps was through offer­ing them access to this data”. Short­ly after arriv­ing at the company’s Sil­i­con Val­ley head­quar­ters he was told that any deci­sion to ban an app required the per­son­al approval of the chief exec­u­tive, Mark Zucker­berg, although the pol­i­cy was lat­er relaxed to make it eas­i­er to deal with rogue devel­op­ers.
...

So how many Face­book users had their pri­vate pro­file infor­ma­tion like­ly via this ‘fine print’ fea­ture that allowed app devel­op­ers to scrape the pro­files of app users and their friends? Accord­ing to Parak­i­las, prob­a­bly a major­i­ty of Face­book users. So that black mar­ket of Face­book pro­files prob­a­bly includes a major­i­ty of Face­book users. But even more amaz­ing is that Face­book hand­ed out this per­son­al user infor­ma­tion to app devel­op­ers in exchange for a 30 share of the mon­ey they made through the app. Face­book was basi­cal­ly direct­ly sell­ing pri­vate user data to devel­op­ers, which is a big rea­son why Parak­i­las’s esti­mate that a major­i­ty of Face­book users were impact­ed by this is like­ly true. Espe­cial­ly if, as Parak­i­las hints, the num­ber of devel­op­ers grab­bing user pro­file infor­ma­tion via these apps might be in the hun­dreds of thou­sands. That’s a lot of devel­op­ers poten­tial­ly feed­ing into that black mar­ket:

...
‘A major­i­ty of Face­book users’

Parak­i­las, 38, who now works as a prod­uct man­ag­er for Uber, is par­tic­u­lar­ly crit­i­cal of Facebook’s pre­vi­ous pol­i­cy of allow­ing devel­op­ers to access the per­son­al data of friends of peo­ple who used apps on the plat­form, with­out the knowl­edge or express con­sent of those friends.

That fea­ture, called friends per­mis­sion, was a boon to out­side soft­ware devel­op­ers who, from 2007 onwards, were giv­en per­mis­sion by Face­book to build quizzes and games – like the wide­ly pop­u­lar Far­mVille – that were host­ed on the plat­form.

The apps pro­lif­er­at­ed on Face­book in the years lead­ing up to the company’s 2012 ini­tial pub­lic offer­ing, an era when most users were still access­ing the plat­form via lap­tops and com­put­ers rather than smart­phones.

Face­book took a 30% cut of pay­ments made through apps, but in return enabled their cre­ators to have access to Face­book user data.

Parak­i­las does not know how many com­pa­nies sought friends per­mis­sion data before such access was ter­mi­nat­ed around mid-2014. How­ev­er, he said he believes tens or maybe even hun­dreds of thou­sands of devel­op­ers may have done so.

Parak­i­las esti­mates that “a major­i­ty of Face­book users” could have had their data har­vest­ed by app devel­op­ers with­out their knowl­edge. The com­pa­ny now has stricter pro­to­cols around the degree of access third par­ties have to data.

...

Dur­ing the time he was at Face­book, Parak­i­las said the com­pa­ny was keen to encour­age more devel­op­ers to build apps for its plat­form and “one of the main ways to get devel­op­ers inter­est­ed in build­ing apps was through offer­ing them access to this data”. Short­ly after arriv­ing at the company’s Sil­i­con Val­ley head­quar­ters he was told that any deci­sion to ban an app required the per­son­al approval of the chief exec­u­tive, Mark Zucker­berg, although the pol­i­cy was lat­er relaxed to make it eas­i­er to deal with rogue devel­op­ers.
...

“Face­book took a 30% cut of pay­ments made through apps, but in return enabled their cre­ators to have access to Face­book user data.”

And that, right there, is per­haps the biggest scan­dal here: Face­book just hand­ed user data away in exchange for rev­enue streams from app devel­op­ers. And this was a key ele­ment of its busi­ness mod­el dur­ing this 2007–2014 peri­od. “Read the fine print” in the terms of ser­vice was the excuse they use:

...
“It was well under­stood in the com­pa­ny that that pre­sent­ed a risk,” he said. “Face­book was giv­ing data of peo­ple who had not autho­rised the app them­selves, and was rely­ing on terms of ser­vice and set­tings that peo­ple didn’t read or under­stand.”

It was this fea­ture that was exploit­ed by Glob­al Sci­ence Research, and the data pro­vid­ed to Cam­bridge Ana­lyt­i­ca in 2014. GSR was run by the Cam­bridge Uni­ver­si­ty psy­chol­o­gist Alek­san­dr Kogan, who built an app that was a per­son­al­i­ty test for Face­book users.
...

And this is all why Alek­san­dr Kogan’s asser­tions that he had a close work­ing rela­tion­ship with Face­book and did noth­ing tech­ni­cal­ly wrong do actu­al­ly seem to be backed up by Parak­i­las’s whis­tle-blow­ing. Both because it’s hard to see what Kogan did that was­n’t part of Face­book’s busi­ness mod­el and also because it’s hard to ignore that Kogan’s GSR shell com­pa­ny was one of the very last apps to have per­mis­sion to exploit their “friends’ per­mis­sion” app loop­hole. That sure does sug­gest that Kogan real­ly did have a “close work­ing rela­tion­ship” with Face­book. So close he got seem­ing­ly favored treat­ment, and that’s com­pared to the seem­ing­ly vast num­ber of apps that were appar­ent­ly using this “friends per­mis­sions” fea­ture: 1 in 10 Face­book apps, accord­ing to a 2010 study:

...
The test auto­mat­i­cal­ly down­loaded the data of friends of peo­ple who took the quiz, osten­si­bly for aca­d­e­m­ic pur­pos­es. Cam­bridge Ana­lyt­i­ca has denied know­ing the data was obtained improp­er­ly, and Kogan main­tains he did noth­ing ille­gal and had a “close work­ing rela­tion­ship” with Face­book.

While Kogan’s app only attract­ed around 270,000 users (most of whom were paid to take the quiz), the com­pa­ny was then able to exploit the friends per­mis­sion fea­ture to quick­ly amass data per­tain­ing to more than 50 mil­lion Face­book users.

“Kogan’s app was one of the very last to have access to friend per­mis­sions,” Parak­i­las said, adding that many oth­er sim­i­lar apps had been har­vest­ing sim­i­lar quan­ti­ties of data for years for com­mer­cial pur­pos­es. Aca­d­e­m­ic research from 2010, based on an analy­sis of 1,800 Face­books apps [32], con­clud­ed that around 11% of third-par­ty devel­op­ers request­ed data belong­ing to friends of users.

If those fig­ures were extrap­o­lat­ed, tens of thou­sands of apps, if not more, were like­ly to have sys­tem­at­i­cal­ly culled “pri­vate and per­son­al­ly iden­ti­fi­able” data belong­ing to hun­dreds of mil­lions of users, Parak­i­las said.
...

““Kogan’s app was one of the very last to have access to friend per­mis­sions,” Parak­i­las said, adding that many oth­er sim­i­lar apps had been har­vest­ing sim­i­lar quan­ti­ties of data for years for com­mer­cial pur­pos­es. Aca­d­e­m­ic research from 2010, based on an analy­sis of 1,800 Face­books apps [32], con­clud­ed that around 11% of third-par­ty devel­op­ers request­ed data belong­ing to friends of users.”

As of 2010, around 11 per­cent of app devel­op­ers request­ed data belong­ing to friends of users. Keep that in mind when Face­book claims that Alek­san­dr Kogan improp­er­ly obtained data from the friends of the peo­ple who down­loaded Kogan’s app.

So what made Face­book even­tu­al­ly end this “friends per­mis­sions” pol­i­cy in mid-2014? While Parak­i­las has already left the com­pa­ny by then, he does recall con­ver­sa­tions with exec­u­tive who were ner­vous about com­peti­tors build­ing their own social net­works from all the data Face­book was giv­ing away:

...
Parak­i­las said he was unsure why Face­book stopped allow­ing devel­op­ers to access friends data around mid-2014, rough­ly two years after he left the com­pa­ny. How­ev­er, he said he believed one rea­son may have been that Face­book exec­u­tives were becom­ing aware that some of the largest apps were acquir­ing enor­mous troves of valu­able data.

He recalled con­ver­sa­tions with exec­u­tives who were ner­vous about the com­mer­cial val­ue of data being passed to oth­er com­pa­nies.

“They were wor­ried that the large app devel­op­ers were build­ing their own social graphs, mean­ing they could see all the con­nec­tions between these peo­ple,” he said. “They were wor­ried that they were going to build their own social net­works.”
...

That’s how much data Face­book was hand­ing out to encour­age new app devel­op­ment: so much data that they were con­cerned about cre­at­ing com­peti­tors.

Final­ly, it’s impor­tant to note that the pic­ture paint­ed by Parak­i­las only goes until the end of 2012, when he left in frus­tra­tion. So we don’t actu­al­ly have tes­ti­mo­ny of Face­book insid­ers who were involved with app data breach­es like Parak­i­las dur­ing the peri­od when Cam­bridge Ana­lyt­i­ca was engaged in its mass data col­lec­tion scheme:

...
Frus­trat­ed at the lack of action, Parak­i­las left Face­book in late 2012. “I didn’t feel that the com­pa­ny treat­ed my con­cerns seri­ous­ly. I didn’t speak out pub­licly for years out of self-inter­est, to be frank.”
...

Now, it seems like a safe bet that the prob­lem only got worse after Parak­i­las left giv­en how the Cam­bridge Ana­lyt­i­ca sit­u­a­tion played out, but we don’t know yet just had bad it was at this point.

Alek­san­dr Kogan: Face­book’s Close Friend (Until He Belat­ed­ly Was­n’t)

So, fac­tor­ing in what we just saw with Parak­i­las’s claims about extent to which Face­book was hand­ing out pri­vate Face­book pro­file data — the inter­nal pro­file that Face­book builds up about you — to app devel­op­ers for wide­spread com­mer­cial appli­ca­tions, let’s take a look at the some of the claims Alek­san­dr Kogan has made about his rela­tion­ship with Face­book. Because while Kogan makes some extra­or­di­nary claims, they are also con­sis­tent with Parak­i­las’s claims, although in some cas­es Kogan’s descrip­tion actu­al­ly goes much fur­ther than Parak­i­las.

For instance, accord­ing to the fol­low­ing Observ­er arti­cle ...

1. In an email to col­leagues at the Uni­ver­si­ty of Cam­bridge, Alek­san­dr Kogan said that he had cre­at­ed the Face­book app in 2013 for aca­d­e­m­ic pur­pos­es, and used it for “a num­ber of stud­ies”. After he found­ed GSR, Kogan wrote, he trans­ferred the app to the com­pa­ny and changed its name, logo, descrip­tion, and terms and con­di­tions.

2. Kogan also claims in that email that the con­tract his GSR com­pa­ny signed with Face­book in 2014 made it absolute­ly clear the data was going to be used for com­mer­cial appli­ca­tions and that app users were grant­i­ng Kogan’s com­pa­ny the right to license or resell the data. “We made clear the app was for com­mer­cial use – we nev­er men­tioned aca­d­e­m­ic research nor the Uni­ver­si­ty of Cam­bridge,” Kogan wrote.We clear­ly stat­ed that the users were grant­i­ng us the right to use the data in broad scope, includ­ing sell­ing and licens­ing the data. These changes were all made on the Face­book app plat­form and thus they had full abil­i­ty to review the nature of the app and raise issues. Face­book at no point raised any con­cerns at all about any of these changes.” So Kogan says he made it clear to Face­book and user the app was for com­mer­cial pur­pos­es and that the data might be resold which sounds like the kind of sit­u­a­tion Sandy Parak­i­las said he wit­nessed except even more open (which should be eas­i­ly ver­i­fi­able if the app code still exists).

3. Face­book did­n’t actu­al­ly kick Kogan off of its plat­form until March 16th of this year, just days before this sto­ry broke. Which con­sis­tent with Kogan’s claims that he had a good work­ing rela­tion­ship with Face­book.

4. When Kogan found­ed Glob­al Sci­ence Research (GSR) in May 2014, he co-found­ed it with anoth­er Cam­bridge researcher, Joseph Chan­cel­lor. Chan­cel­lor is cur­rent­ly employed by Face­book.

5. Face­book gave Kogan’s Uni­ver­si­ty of Cam­bridge lab pro­vid­ed the dataset of “every friend­ship formed in 2011 in every coun­try in the world at the nation­al aggre­gate lev­el”. 57 bil­lion Face­book rela­tion­ships in all. The data was anonymized and aggre­gat­ed, so it did­n’t lit­er­al­ly include details on indi­vid­ual Face­book friends and was instead the aggre­gate “friend” counts at a nation­al. The data was used to pub­lish a study in Per­son­al­i­ty and Indi­vid­ual Dif­fer­ences in 2015 and two Face­book employ­ees were named as co-authors of the study, along­side researchers from Cam­bridge, Har­vard and the Uni­ver­si­ty of Cal­i­for­nia, Berke­ley. But it’s still a sign that Kogan is indeed being hon­est when he says he had a close work­ing rela­tion­ship with Face­book. It’s also a reminder that when Face­book claims that it was just hand­ing out data for “research pur­pos­es” only, if that was true it would have hand­ed out anonymized aggre­gat­ed data like they did in this sit­u­a­tion with Kogan.

6. That study co-authored by Kogan’s team and Face­book did­n’t just use the anonymized aggre­gat­ed friend­ship data. The study also used non-anonymized Face­book ata col­lect­ed through Face­book apps using exact­ly the same tech­niques Kogan’s app for Cam­bridge Ana­lyt­i­ca used. This study was pub­lished in August of 2015. Again, it was a study co-authored by Face­book. GSR co-founder Joseph Chan­cel­lor left GSR a month lat­er and joined Face­book as a user expe­ri­ence research in Novem­ber 2015. Recall that it was a month lat­er, Decem­ber 2015, when we saw the first news reports of Ted Cruz’s cam­paign using Face­book data. Also recall that Face­book respond­ed to that Decem­ber 2015 report by say­ing it would look into the mat­ter. Face­book final­ly sent Cam­bridge Ana­lyt­i­ca a let­ter in August of 2016, days before Steve Ban­non became Trump’s cam­paign man­ag­er, ask­ing that Cam­bridge Ana­lyt­i­ca delete the data. So the fact that Face­book co-authored a paper with Kogan and Chan­cel­lor in August of the 2015 and then Chan­cel­lor joined Face­book in 2015 is a pret­ty sig­nif­i­cant bit of con­text for look­ing into Face­book’s behav­ior. Because Face­book did­n’t just know it was guilty of work­ing close­ly with Kogan. They also knew they just co-authored an aca­d­e­m­ic paper using data gath­ered with the same tech­nique Cam­bridge Ana­lyt­i­ca was charged with using.

7. Kogan does chal­lenge one of the claims by Christo­pher Wylie. Specif­i­cal­ly, Wylie claimed that Face­book became alarmed over the vol­ume of data Kogan’s app was scoop­ing up (50 mil­lion pro­files) but Kogan assuaged those con­cerns by say­ing it was all for research. Kogan says this is a fab­ri­ca­tion and Face­book nev­er actu­al­ly con­tact­ed him express­ing alarm.

So, accord­ing to Alek­san­dr Kogan, Face­book real­ly did have an excep­tion­al­ly close rela­tion­ship with Kogan and Face­book real­ly was total­ly on board with what Kogan and Cam­bridge Ana­lyt­i­ca were doing [35]:

The Guardian

Face­book gave data about 57bn friend­ships to aca­d­e­m­ic
Vol­ume of data sug­gests trust­ed part­ner­ship with Alek­san­dr Kogan, says ana­lyst

Julia Car­rie Wong and Paul Lewis in San Fran­cis­co
Thu 22 Mar 2018 10.56 EDT
Last mod­i­fied on Sat 24 Mar 2018 22.56 EDT

Before Face­book sus­pend­ed Alek­san­dr Kogan [8] from its plat­form for the data har­vest­ing “scam [3]” at the cen­tre of the unfold­ing Cam­bridge Ana­lyt­i­ca scan­dal, the social media com­pa­ny enjoyed a close enough rela­tion­ship with the researcher that it pro­vid­ed him with an anonymised, aggre­gate dataset of 57bn Face­book friend­ships.

Face­book pro­vid­ed the dataset of “every friend­ship formed in 2011 in every coun­try in the world at the nation­al aggre­gate lev­el” to Kogan’s Uni­ver­si­ty of Cam­bridge lab­o­ra­to­ry for a study on inter­na­tion­al friend­ships [36] pub­lished in Per­son­al­i­ty and Indi­vid­ual Dif­fer­ences in 2015. Two Face­book employ­ees were named as co-authors of the study, along­side researchers from Cam­bridge, Har­vard and the Uni­ver­si­ty of Cal­i­for­nia, Berke­ley. Kogan was pub­lish­ing under the name Alek­san­dr Spec­tre at the time.

A Uni­ver­si­ty of Cam­bridge press release [37] on the study’s pub­li­ca­tion not­ed that the paper was “the first out­put of ongo­ing research col­lab­o­ra­tions between Spectre’s lab in Cam­bridge and Face­book”. Face­book did not respond to queries about whether any oth­er col­lab­o­ra­tions occurred.

“The sheer vol­ume of the 57bn friend pairs implies a pre-exist­ing rela­tion­ship,” said Jonathan Albright, research direc­tor at the Tow Cen­ter for Dig­i­tal Jour­nal­ism at Colum­bia Uni­ver­si­ty. “It’s not com­mon for Face­book to share that kind of data. It sug­gests a trust­ed part­ner­ship between Alek­san­dr Kogan/Spectre and Face­book.”

Face­book down­played the sig­nif­i­cance of the dataset, which it said was shared with Kogan in 2013. “The data that was shared was lit­er­al­ly num­bers – num­bers of how many friend­ships were made between pairs of coun­tries – ie x num­ber of friend­ships made between the US and UK,” Face­book spokes­woman Chris­tine Chen said by email. “There was no per­son­al­ly iden­ti­fi­able infor­ma­tion includ­ed in this data.”

Facebook’s rela­tion­ship with Kogan has since soured.

“We end­ed our work­ing rela­tion­ship with Kogan alto­geth­er after we learned that he vio­lat­ed Facebook’s terms of ser­vice for his unre­lat­ed work as a Face­book app devel­op­er,” Chen said. Face­book has said that it learned of Kogan’s mis­use of the data in Decem­ber 2015, when the Guardian first report­ed [2] that the data had been obtained by Cam­bridge Ana­lyt­i­ca.

“We start­ed to take steps to end the rela­tion­ship right after the Guardian report, and after inves­ti­ga­tion we end­ed the rela­tion­ship soon after, in 2016,” Chen said.

On Fri­day 16 March, in antic­i­pa­tion of the Observ­er [38]’s report­ing that Kogan had improp­er­ly har­vest­ed and shared the data of more than 50 mil­lion Amer­i­cans [15], Face­book sus­pend­ed Kogan from the plat­form, issued a state­ment [8] say­ing that he “lied” to the com­pa­ny, and char­ac­terised [3] his activ­i­ties as “a scam – and a fraud”.

On Tues­day, Face­book went fur­ther, say­ing [39] in a state­ment: “The entire com­pa­ny is out­raged we were deceived.” And on Wednes­day, in his first pub­lic state­ment [40] on the scan­dal, its chief exec­u­tive, Mark Zucker­berg, called Kogan’s actions a “breach of trust”.

But Face­book has not explained how it came to have such a close rela­tion­ship with Kogan that it was co-author­ing research papers with him, nor why it took until this week – more than two years after the Guardian ini­tial­ly report­ed on Kogan’s data har­vest­ing [2] activ­i­ties – for it to inform the users whose per­son­al infor­ma­tion was improp­er­ly shared.

And Kogan has offered a defence of his actions in an inter­view [41] with the BBC and an email to his Cam­bridge col­leagues obtained by the Guardian. “My view is that I’m being basi­cal­ly used as a scape­goat by both Face­book and Cam­bridge Ana­lyt­i­ca,” Kogan said on Radio 4 on Wednes­day.

The data col­lec­tion that result­ed in Kogan’s sus­pen­sion by Face­book was under­tak­en by Glob­al Sci­ence Research (GSR), a com­pa­ny he found­ed in May 2014 with anoth­er Cam­bridge researcher, Joseph Chan­cel­lor. Chan­cel­lor is cur­rent­ly employed by Face­book.

Between June and August of that year, GSR paid approx­i­mate­ly 270,000 indi­vid­u­als to use a Face­book ques­tion­naire app that har­vest­ed data from their own Face­book pro­files, as well as from their friends, result­ing in a dataset of more than 50 mil­lion users. The data was sub­se­quent­ly giv­en to Cam­bridge Ana­lyt­i­ca, in what Face­book has said was a vio­la­tion of Kogan’s agree­ment to use the data sole­ly for aca­d­e­m­ic pur­pos­es.

In his email to col­leagues at Cam­bridge, Kogan said that he had cre­at­ed the Face­book app in 2013 for aca­d­e­m­ic pur­pos­es, and used it for “a num­ber of stud­ies”. After he found­ed GSR, Kogan wrote, he trans­ferred the app to the com­pa­ny and changed its name, logo, descrip­tion, and terms and con­di­tions. CNN first report­ed on the Cam­bridge email. Kogan did not respond to the Guardian’s request for com­ment on this arti­cle.

“We made clear the app was for com­mer­cial use – we nev­er men­tioned aca­d­e­m­ic research nor the Uni­ver­si­ty of Cam­bridge,” Kogan wrote. “We clear­ly stat­ed that the users were grant­i­ng us the right to use the data in broad scope, includ­ing sell­ing and licens­ing the data. These changes were all made on the Face­book app plat­form and thus they had full abil­i­ty to review the nature of the app and raise issues. Face­book at no point raised any con­cerns at all about any of these changes.”

Kogan is not alone in crit­i­cis­ing Facebook’s appar­ent efforts to place the blame on him.

“In my view, it’s Face­book that did most of the shar­ing,” said Albright, who ques­tioned why Face­book cre­at­ed a sys­tem for third par­ties to access so much per­son­al infor­ma­tion in the first place. That sys­tem “was designed to share their users’ data in mean­ing­ful ways in exchange for stock val­ue”, he added.

Whistle­blow­er Christo­pher Wylie told the Observ­er [42] that Face­book was aware of the vol­ume of data being pulled by Kogan’s app. “Their secu­ri­ty pro­to­cols were trig­gered because Kogan’s apps were pulling this enor­mous amount of data, but appar­ent­ly Kogan told them it was for aca­d­e­m­ic use,” Wylie said. “So they were like: ‘Fine.’”

In the Cam­bridge email, Kogan char­ac­terised this claim as a “fab­ri­ca­tion”, writ­ing: “There was no exchange with Face­book about it, and ... we nev­er claimed dur­ing the project that it was for aca­d­e­m­ic research. In fact, we did our absolute best not to have the project have any entan­gle­ments with the uni­ver­si­ty.”

The col­lab­o­ra­tion between Kogan and Face­book researchers which result­ed in the report pub­lished in 2015 also used data har­vest­ed by a Face­book app. The study analysed two datasets, the anony­mous macro-lev­el nation­al set of 57bn friend pairs pro­vid­ed by Face­book and a small­er dataset col­lect­ed by the Cam­bridge aca­d­e­mics.

For the small­er dataset, the research team used the same method of pay­ing peo­ple to use a Face­book app that har­vest­ed data about the indi­vid­u­als and their friends. Face­book was not involved in this part of the study. The study notes that the users signed a con­sent form about the research and that “no decep­tion was used”.

The paper was pub­lished in late August 2015. In Sep­tem­ber 2015, Chan­cel­lor left GSR, accord­ing to com­pa­ny records. In Novem­ber 2015, Chan­cel­lor was hired to work at Face­book as a user expe­ri­ence researcher.

...

———-

“Face­book gave data about 57bn friend­ships to aca­d­e­m­ic” by Julia Car­rie Wong and Paul Lewis; The Guardian; 03/22/2018 [35]

“Before Face­book sus­pend­ed Alek­san­dr Kogan [8] from its plat­form for the data har­vest­ing “scam [3]” at the cen­tre of the unfold­ing Cam­bridge Ana­lyt­i­ca scan­dal, the social media com­pa­ny enjoyed a close enough rela­tion­ship with the researcher that it pro­vid­ed him with an anonymised, aggre­gate dataset of 57bn Face­book friend­ships.

An anonymized, aggre­gate dataset of 57bn Face­book friend­ships sure makes it a lot eas­i­er to take Kogan at his word when he claims a close work­ing rela­tion­ship with Face­book.

Now, keep in mind that the aggre­gate anonymized data was aggre­gate at the nation­al lev­el, so it’s not as if Face­book gave Kogan a list of 57 bil­lion Face­book friend­ships. And when you think about it, that aggre­gat­ed anonymized data is far less sen­si­tive than the per­son­al Face­book pro­file data Kogan and oth­er app devel­op­ers were rou­tine­ly grab­bing dur­ing this peri­od. It’s the fact that Face­book gave this data to Kogan in the first place that lends cre­dence to his claims.

But the biggest fac­tor lend­ing cre­dence to Kogan’s claims is the fact that Face­book co-authored a study with Kogan and oth­er at the Uni­ver­si­ty of Cam­bridge using that anonymized aggre­gat­ed data. Two Face­book employ­ees were named as co-authors of the study. That is def­i­nite­ly a sign of close work­ing rela­tion­ship:

...
Face­book pro­vid­ed the dataset of “every friend­ship formed in 2011 in every coun­try in the world at the nation­al aggre­gate lev­el” to Kogan’s Uni­ver­si­ty of Cam­bridge lab­o­ra­to­ry for a study on inter­na­tion­al friend­ships [36] pub­lished in Per­son­al­i­ty and Indi­vid­ual Dif­fer­ences in 2015. Two Face­book employ­ees were named as co-authors of the study, along­side researchers from Cam­bridge, Har­vard and the Uni­ver­si­ty of Cal­i­for­nia, Berke­ley. Kogan was pub­lish­ing under the name Alek­san­dr Spec­tre at the time.

A Uni­ver­si­ty of Cam­bridge press release [37] on the study’s pub­li­ca­tion not­ed that the paper was “the first out­put of ongo­ing research col­lab­o­ra­tions between Spectre’s lab in Cam­bridge and Face­book”. Face­book did not respond to queries about whether any oth­er col­lab­o­ra­tions occurred.

“The sheer vol­ume of the 57bn friend pairs implies a pre-exist­ing rela­tion­ship,” said Jonathan Albright, research direc­tor at the Tow Cen­ter for Dig­i­tal Jour­nal­ism at Colum­bia Uni­ver­si­ty. “It’s not com­mon for Face­book to share that kind of data. It sug­gests a trust­ed part­ner­ship between Alek­san­dr Kogan/Spectre and Face­book.”
...

Even more damn­ing for Face­book is that the research co-authored by Kogan, Face­book, and oth­er researchers did­n’t just includ­ed the anonymized aggre­gat­ed data. It also includ­ed a sec­ond data set of non-anonymized data that was har­vest­ed in exact­ly the same way Kogan’s GSR app worked. And while Face­book appar­ent­ly was­n’t involved in that part of the study, that’s beside the point. Face­book clear­ly knew about it if they co-authored the study:

...
The col­lab­o­ra­tion between Kogan and Face­book researchers which result­ed in the report pub­lished in 2015 also used data har­vest­ed by a Face­book app. The study analysed two datasets, the anony­mous macro-lev­el nation­al set of 57bn friend pairs pro­vid­ed by Face­book and a small­er dataset col­lect­ed by the Cam­bridge aca­d­e­mics.

For the small­er dataset, the research team used the same method of pay­ing peo­ple to use a Face­book app that har­vest­ed data about the indi­vid­u­als and their friends. Face­book was not involved in this part of the study. The study notes that the users signed a con­sent form about the research and that “no decep­tion was used”.

The paper was pub­lished in late August 2015. In Sep­tem­ber 2015, Chan­cel­lor left GSR, accord­ing to com­pa­ny records. In Novem­ber 2015, Chan­cel­lor was hired to work at Face­book as a user expe­ri­ence researcher.
...

But, alas, Kogan’s rela­tion­ship with Face­book as since soured, with Face­book now act­ing as if Kogan had total­ly vio­lat­ed their trust. And yet it’s hard to ignore the fact that Kogan was­n’t for­mal­ly kicked off Face­book’s plat­form until March 16th of this year, just a few days before all these sto­ries about Kogan and Face­book were about to go pub­lic:

...
Facebook’s rela­tion­ship with Kogan has since soured.

“We end­ed our work­ing rela­tion­ship with Kogan alto­geth­er after we learned that he vio­lat­ed Facebook’s terms of ser­vice for his unre­lat­ed work as a Face­book app devel­op­er,” Chen said. Face­book has said that it learned of Kogan’s mis­use of the data in Decem­ber 2015, when the Guardian first report­ed [2] that the data had been obtained by Cam­bridge Ana­lyt­i­ca.

“We start­ed to take steps to end the rela­tion­ship right after the Guardian report, and after inves­ti­ga­tion we end­ed the rela­tion­ship soon after, in 2016,” Chen said.

On Fri­day 16 March, in antic­i­pa­tion of the Observ­er [38]’s report­ing that Kogan had improp­er­ly har­vest­ed and shared the data of more than 50 mil­lion Amer­i­cans [15], Face­book sus­pend­ed Kogan from the plat­form, issued a state­ment [8] say­ing that he “lied” to the com­pa­ny, and char­ac­terised [3] his activ­i­ties as “a scam – and a fraud”.

On Tues­day, Face­book went fur­ther, say­ing [39] in a state­ment: “The entire com­pa­ny is out­raged we were deceived.” And on Wednes­day, in his first pub­lic state­ment [40] on the scan­dal, its chief exec­u­tive, Mark Zucker­berg, called Kogan’s actions a “breach of trust”.
...

““The entire com­pa­ny is out­raged we were deceived.” And on Wednes­day, in his first pub­lic state­ment [40] on the scan­dal, its chief exec­u­tive, Mark Zucker­berg, called Kogan’s actions a “breach of trust”.”

Mark Zucker­berg is com­plain­ing about a “breach of trust.” LOL!

And yet Face­book has yet to explain the nature of its rela­tion­ship with Kogan or why it was that they did­n’t kick him off the plat­form until only recent­ly. But Kogan has an expla­na­tion: He’s a scape­goat and he was­n’t doing any­thing Face­book did­n’t know he was doing. And when you notice that Kogan’s co-founder of GSR, Joseph Chan­cel­lor, is now a Face­book employ­ee, it’s hard not to take his claims seri­ous­ly:

...
But Face­book has not explained how it came to have such a close rela­tion­ship with Kogan that it was co-author­ing research papers with him, nor why it took until this week – more than two years after the Guardian ini­tial­ly report­ed on Kogan’s data har­vest­ing [2] activ­i­ties – for it to inform the users whose per­son­al infor­ma­tion was improp­er­ly shared.

And Kogan has offered a defence of his actions in an inter­view [41] with the BBC and an email to his Cam­bridge col­leagues obtained by the Guardian. “My view is that I’m being basi­cal­ly used as a scape­goat by both Face­book and Cam­bridge Ana­lyt­i­ca,” Kogan said on Radio 4 on Wednes­day.

The data col­lec­tion that result­ed in Kogan’s sus­pen­sion by Face­book was under­tak­en by Glob­al Sci­ence Research (GSR), a com­pa­ny he found­ed in May 2014 with anoth­er Cam­bridge researcher, Joseph Chan­cel­lor. Chan­cel­lor is cur­rent­ly employed by Face­book.
...

But if Kogan’s claims are to be tak­en seri­ous­ly, we have a pret­ty seri­ous scan­dal on our hands. Because Kogan claims that not only did he make it clear to Face­book and his app users that the data they were col­lect­ing was for com­mer­cial use — with no men­tion of aca­d­e­m­ic or research pur­pos­es of the Uni­ver­si­ty of Cam­bridge — but he also claims that he made it clear the data GSR was col­lect­ing could be licensed and resold. And Face­book at no point raised any con­cerns at all about any of this:

...
“We made clear the app was for com­mer­cial use – we nev­er men­tioned aca­d­e­m­ic research nor the Uni­ver­si­ty of Cam­bridge,” Kogan wrote. “We clear­ly stat­ed that the users were grant­i­ng us the right to use the data in broad scope, includ­ing sell­ing and licens­ing the data. These changes were all made on the Face­book app plat­form and thus they had full abil­i­ty to review the nature of the app and raise issues. Face­book at no point raised any con­cerns at all about any of these changes.”

Kogan is not alone in crit­i­cis­ing Facebook’s appar­ent efforts to place the blame on him.

“In my view, it’s Face­book that did most of the shar­ing,” said Albright, who ques­tioned why Face­book cre­at­ed a sys­tem for third par­ties to access so much per­son­al infor­ma­tion in the first place. That sys­tem “was designed to share their users’ data in mean­ing­ful ways in exchange for stock val­ue”, he added.
...

Now, it’s worth not­ing that the casu­al accep­tance of the com­mer­cial use of the data col­lect­ed over these Face­book apps and the poten­tial licens­ing and reselling of that data is actu­al­ly a far more seri­ous­ly sit­u­a­tion than the one Sandy Parak­i­las described dur­ing his time at Face­book. Recall that, accord­ing to Parak­i­las, app devel­op­ers sim­ply had to tell Face­book was that they were going to use the pro­file data on app users and their friends to ‘improve the user expe­ri­ence.’ It was fine if they were com­mer­cial apps from Face­book’s per­spec­tive. But Parak­i­las did­n’t describe a sit­u­a­tion where app devel­op­ers open­ly made it clear they might license or resell the data. So Kogan’s claim that it was clear his app had com­mer­cial appli­ca­tions and might involve reselling the data is even more egre­gious than the sit­u­a­tion Parak­i­las described. But don’t for­get that Parak­i­las left Face­book in late 2012 and Kogan’s app would have been approved in 2014 so it’s entire­ly pos­si­ble Face­book’s poli­cies got even more egre­gious after Parak­i­las left.

And it’s worth not­ing how Kogan’s claims dif­fer from Christo­pher Wylie’s. Wylie asserts that Face­book grew alarmed by the vol­ume of data GSR’s app was pulling from Face­book users and Kogan assured them it was for research pur­pos­es. Where­as Kogan says Face­book nev­er expressed any alarm at all:

...
Whistle­blow­er Christo­pher Wylie told the Observ­er [42] that Face­book was aware of the vol­ume of data being pulled by Kogan’s app. “Their secu­ri­ty pro­to­cols were trig­gered because Kogan’s apps were pulling this enor­mous amount of data, but appar­ent­ly Kogan told them it was for aca­d­e­m­ic use,” Wylie said. “So they were like: ‘Fine.’”

In the Cam­bridge email, Kogan char­ac­terised this claim as a “fab­ri­ca­tion”, writ­ing: “There was no exchange with Face­book about it, and ... we nev­er claimed dur­ing the project that it was for aca­d­e­m­ic research. In fact, we did our absolute best not to have the project have any entan­gle­ments with the uni­ver­si­ty.”
...

So as we can see, when it comes to Face­book’s “friends per­mis­sions” data shar­ing pol­i­cy, its arrange­ment with Alek­san­dr Kogan was prob­a­bly one of the more respon­si­ble ones it engaged in because, hey, at least Kogan’s work was osten­si­bly for research pur­pos­es and involved at least some anonymized data.

Cam­bridge Ana­lyt­i­ca’s Infor­mal Friend: Palan­tir

And as we can also see, the more we learn about this sit­u­a­tion, the hard­er it gets to dis­miss Kogan’s claims that Face­book is mak­ing in a scape­goat in order to cov­er up not just the rela­tion­ship Face­book had with Kogan but the fact that what Kogan was doing was rou­tine for app devel­op­ers for years.

But as the fol­low­ing New York Times arti­cle makes clear, Face­book’s rela­tion­ship with Alek­san­dr Kogan isn’t the only work­ing rela­tion­ship Face­book needs to wor­ry about that might lead back to Cam­bridge Ana­lyt­i­ca. Because it turns out there’s anoth­er Face­book con­nec­tion to Cam­bridge Ana­lyt­i­ca and it’s poten­tial­ly far, far more scan­dalous than Face­book’s rela­tion­ship with Kogan: It turns out Palan­tir might be the orig­i­na­tor of the idea to cre­ate Kogan’s app for the pur­pose of col­lect­ing psy­cho­log­i­cal pro­files. That’s right, accord­ing to doc­u­ments the New York Times has seen, Palan­tir, the pri­vate intel­li­gence firm with a close rela­tion­ship with the US nation­al secu­ri­ty state, was in talks with Cam­bridge Ana­lyt­i­ca from 2013–2014 about psy­cho­log­i­cal­ly pro­fil­ing vot­ers and it was an employ­ee of Palan­tir who raised the idea of cre­at­ing that app in the first place.

And this is of course wild­ly scan­dalous if true because Palan­tir was found­ed by the Face­book exec­u­tive Peter Thiel who also hap­pens to be a far right polit­i­cal activist and a close ally of Pres­i­dent Trump.

But it gets worse. And weird­er. Because it sounds like one of the peo­ple encour­ag­ing SCL (Cam­bridge Ana­lyt­i­ca’s par­ent com­pa­ny) to work with Palan­tir was none oth­er than Sophie Schmidt, daugh­ter of Google CEO Eric Schmidt.

Keep in mind that this isn’t the first time we’ve heard about Palan­tir’s ties to Cam­bridge Ana­lyt­i­ca and Sophie Schmidt’s role in this. It was report­ed by the Observ­er last May [43]. Accord­ing to that May 2017 arti­cle in the Observ­er, Schmidt was pass­ing through Lon­don in June of 2013 when she decid­ed to called up her for­mer boss at SCL and rec­om­mend that they con­tact Palan­tir. Also if inter­est is that if you look at the cur­rent ver­sion of that Observ­er arti­cle, all men­tion of Sophie Schmidt has been removed and there’s a note that the arti­cle is the sub­ject of legal com­plaints on behalf of Cam­bridge Ana­lyt­i­ca LLC and SCL Elec­tions Lim­it­ed [17]. But in the orig­i­nal arti­cle [44] she’s men­tioned quite exten­sive­ly. It would appear that some­one is very upset about the Sophie Schmidt angle to this sto­ry.

So this Palantir/Sophie Schmidt side of this sto­ry isn’t a new. But we’re learn­ing a lot more infor­ma­tion about that rela­tion­ship now. For instance:

1. In ear­ly 2013, Cam­bridge Ana­lyt­i­ca CEO Alexan­der Nix, an SCL direc­tor at the time, and a Palan­tir exec­u­tive dis­cussed work­ing togeth­er on elec­tion cam­paigns.

2. And SCL employ­ee wrote to a col­league in a June 2013 email that Schmidt is push­ing them to work with Palan­tir. “Ever come across Palan­tir. Amus­ing­ly Eric Schmidt’s daugh­ter was an intern with us and is try­ing to push us towards them?” .

3. Accord­ing to Christo­pher Wylie’s tes­ti­mo­ny to law­mak­ers, “There were Palan­tir staff who would come into the office and work on the data...And we would go and meet with Palan­tir staff at Palan­tir.” Wylie said that Palan­tir employ­ees were eager to learn more about using Face­book data and psy­cho­graph­ics. Those dis­cus­sions con­tin­ued through spring 2014.

4. The Palan­tir employ­ee who float­ed the idea of cre­ate the app ulti­mate­ly built by Alek­san­dr Kogan is Alfredas Chmieli­auskas. Chmieli­auskas works on busi­ness devel­op­ment for Palan­tire accord­ing to his LinkedIn page.

5. Palan­tir and Cam­bridge Ana­lyt­i­ca nev­er for­mal­ly start­ed work­ing togeth­er. A Palan­tir spokes­woman acknowl­edged that the com­pa­nies had briefly con­sid­ered work­ing togeth­er but said that Palan­tir declined a part­ner­ship, in part because exec­u­tives there want­ed to steer clear of elec­tion work. Emails indi­cate that Mr. Nix and Mr. Chmieli­auskas sought to revive talks about a for­mal part­ner­ship through ear­ly 2014, but Palan­tir exec­u­tives again declined. Wylie acknowl­edges that Palan­tir and Cam­bridge Ana­lyt­i­ca nev­er signed a con­tract or entered into a for­mal busi­ness rela­tion­ship. But he said some Palan­tir employ­ees helped engi­neer Cam­bridge Analytica’s psy­cho­graph­ic mod­els. In oth­er words, while there was nev­er a for­mal rela­tion­ship, there was an pret­ty sig­nif­i­cant infor­mal rela­tion­ship.

6. Mr. Chmieli­auskas was in com­mu­ni­ca­tion with Wylie’s team in 2014 dur­ing the peri­od when Cam­bridge Ana­lyt­i­ca was ini­tial­ly try­ing to con­vince the Uni­ver­si­ty of Cam­bridge team to work with them. Recall that Cam­bridge Ana­lyt­i­ca ini­tial­ly dis­cov­ered that the Uni­ver­si­ty of Cam­bridge team had exact­ly the kind of data they were inter­est­ed in col­lect­ed via a Face­book app, but the nego­ti­a­tions ulti­mate­ly failed and it was then that Cam­bridge Ana­lyt­i­ca found Alek­san­dr Kogan who agreed to cre­ate his own app. Well, accord­ing to this report, it was Chmieli­auskas who ini­tial­ly sug­gest­ed to Cam­bridge Ana­lyt­i­ca that the firm cre­ate its own ver­sion of the Uni­ver­si­ty of Cam­bridge team’s app as lever­age in those nego­ti­a­tions. In essence, Chmieli­auskas want­ed Cam­bridge Ana­lyt­i­ca to show the Uni­ver­si­ty of Cam­bridge team that they could col­lect the infor­ma­tion them­selves, pre­sum­ably to dri­ve a hard­er bar­gain. And when those nego­ti­a­tions failed Cam­bridge Ana­lyt­i­ca did indeed cre­ate their own app after team­ing up with Kogan.

7. Palan­tir asserts that Chmieli­auskas was act­ing in his own capac­i­ty when he con­tin­ued com­mu­ni­cat­ing with Wylie and made the sug­ges­tion to cre­ate their own app. Palan­tir ini­tial­ly told the New York Times that it had “nev­er had a rela­tion­ship with Cam­bridge Ana­lyt­i­ca, nor have we ever worked on any Cam­bridge Ana­lyt­i­ca data.” Palan­tir lat­er revised this, say­ing that Mr. Chmieli­auskas was not act­ing on the company’s behalf when he advised Mr. Wylie on the Face­book data.

And, again, do not for­get that Palan­tir is own by Peter Thiel, the far right [45] bil­lion­aire [46] ear­ly investor in Face­book and one of Face­book’s board mem­bers to this day [47]. He was also a Trump del­e­gate in 2016 [48] and was in dis­cus­sions with the Trump admin­is­tra­tion to lead the pow­er­ful Pres­i­den­t’s Intel­li­gence Advi­so­ry Board, although he ulti­mate­ly turned that offer down [49]. Oh, and he’s an advo­cate of the Dark Enlight­en­ment [50].

Basi­cal­ly, Peter Thiel was a mem­ber of the ‘Alt Right’ before that term was ever coined. And he’s a very pow­er­ful influ­ence at Face­book. So learn­ing that Palan­tir and Cam­bridge Ana­lyt­i­ca were in dis­cus­sion to work togeth­er on elec­tion projects in 2013 and 2014, a Palan­tir employ­ee was advis­ing Cam­bridge Ana­lyt­i­ca dur­ing the nego­ti­a­tions with the Uni­ver­si­ty of Cam­bridge team, and that Palan­tir employ­ees helped engi­neer Cam­bridge Ana­lyt­i­ca’s psy­cho­graph­ic mod­el based on Face­book is the kind of rev­e­la­tion that just might qual­i­fy as the most scan­dalous rev­e­la­tion in this entire mess [51]:

“Spy Contractor’s Idea Helped Cam­bridge Ana­lyt­i­ca Har­vest Face­book Data” by NICHOLAS CONFESSORE and MATTHEW ROSENBERG; The New York Times; 03/27/2018 [51]

As a start-up called Cam­bridge Ana­lyt­i­ca [3] sought to har­vest the Face­book data of tens of mil­lions of Amer­i­cans in sum­mer 2014, the com­pa­ny received help from at least one employ­ee at Palan­tir Tech­nolo­gies, a top Sil­i­con Val­ley con­trac­tor to Amer­i­can spy agen­cies and the Pen­ta­gon.

It was a Palan­tir employ­ee in Lon­don, work­ing close­ly with the data sci­en­tists build­ing Cambridge’s psy­cho­log­i­cal pro­fil­ing tech­nol­o­gy, who sug­gest­ed the sci­en­tists cre­ate their own app — a mobile-phone-based per­son­al­i­ty quiz — to gain access to Face­book users’ friend net­works, accord­ing to doc­u­ments obtained by The New York Times.

Cam­bridge ulti­mate­ly took a sim­i­lar approach. By ear­ly sum­mer, the com­pa­ny found a uni­ver­si­ty researcher to har­vest data using a per­son­al­i­ty ques­tion­naire and Face­book app. The researcher scraped pri­vate data from over 50 mil­lion Face­book users — and Cam­bridge Ana­lyt­i­ca [52] went into busi­ness sell­ing so-called psy­cho­me­t­ric pro­files of Amer­i­can vot­ers, set­ting itself on a col­li­sion course with reg­u­la­tors and law­mak­ers in the Unit­ed States and Britain.

The rev­e­la­tions pulled Palan­tir — co-found­ed by the wealthy lib­er­tar­i­an Peter Thiel [53] — into the furor sur­round­ing Cam­bridge, which improp­er­ly obtained Face­book data to build ana­lyt­i­cal tools it deployed on behalf of Don­ald J. Trump and oth­er Repub­li­can can­di­dates in 2016. Mr. Thiel, a sup­port­er of Pres­i­dent Trump, serves on the board at Face­book.

“There were senior Palan­tir employ­ees that were also work­ing on the Face­book data,” said Christo­pher Wylie [54], a data expert and Cam­bridge Ana­lyt­i­ca co-founder, in tes­ti­mo­ny before British law­mak­ers on Tues­day.

...

The con­nec­tions between Palan­tir and Cam­bridge Ana­lyt­i­ca were thrust into the spot­light by Mr. Wylie’s tes­ti­mo­ny on Tues­day. Both com­pa­nies are linked to tech-dri­ven bil­lion­aires who backed Mr. Trump’s cam­paign: Cam­bridge is chiefly owned by Robert Mer­cer, the com­put­er sci­en­tist and hedge fund mag­nate, while Palan­tir was co-found­ed in 2003 by Mr. Thiel, who was an ini­tial investor in Face­book.

The Palan­tir employ­ee, Alfredas Chmieli­auskas, works on busi­ness devel­op­ment for the com­pa­ny, accord­ing to his LinkedIn page. In an ini­tial state­ment, Palan­tir said it had “nev­er had a rela­tion­ship with Cam­bridge Ana­lyt­i­ca, nor have we ever worked on any Cam­bridge Ana­lyt­i­ca data.” Lat­er on Tues­day, Palan­tir revised its account, say­ing that Mr. Chmieli­auskas was not act­ing on the company’s behalf when he advised Mr. Wylie on the Face­book data.

“We learned today that an employ­ee, in 2013–2014, engaged in an entire­ly per­son­al capac­i­ty with peo­ple asso­ci­at­ed with Cam­bridge Ana­lyt­i­ca,” the com­pa­ny said. “We are look­ing into this and will take the appro­pri­ate action.”

The com­pa­ny said it was con­tin­u­ing to inves­ti­gate but knew of no oth­er employ­ees who took part in the effort. Mr. Wylie told law­mak­ers that mul­ti­ple Palan­tir employ­ees played a role.

Doc­u­ments and inter­views indi­cate that start­ing in 2013, Mr. Chmieli­auskas began cor­re­spond­ing with Mr. Wylie and a col­league from his Gmail account. At the time, Mr. Wylie and the col­league worked for the British defense and intel­li­gence con­trac­tor SCL Group, which formed Cam­bridge Ana­lyt­i­ca with Mr. Mer­cer the next year. The three shared Google doc­u­ments to brain­storm ideas about using big data to cre­ate sophis­ti­cat­ed behav­ioral pro­files, a prod­uct code-named “Big Dad­dy.”

A for­mer intern at SCL — Sophie Schmidt, the daugh­ter of Eric Schmidt, then Google’s exec­u­tive chair­man — urged the com­pa­ny to link up with Palan­tir, accord­ing to Mr. Wylie’s tes­ti­mo­ny and a June 2013 email viewed by The Times.

“Ever come across Palan­tir. Amus­ing­ly Eric Schmidt’s daugh­ter was an intern with us and is try­ing to push us towards them?” one SCL employ­ee wrote to a col­league in the email.

Ms. Schmidt did not respond to requests for com­ment, nor did a spokesman for Cam­bridge Ana­lyt­i­ca.

In ear­ly 2013, Alexan­der Nix, an SCL direc­tor who became chief exec­u­tive of Cam­bridge Ana­lyt­i­ca, and a Palan­tir exec­u­tive dis­cussed work­ing togeth­er on elec­tion cam­paigns.

A Palan­tir spokes­woman acknowl­edged that the com­pa­nies had briefly con­sid­ered work­ing togeth­er but said that Palan­tir declined a part­ner­ship, in part because exec­u­tives there want­ed to steer clear of elec­tion work. Emails reviewed by The Times indi­cate that Mr. Nix and Mr. Chmieli­auskas sought to revive talks about a for­mal part­ner­ship through ear­ly 2014, but Palan­tir exec­u­tives again declined.

In his tes­ti­mo­ny, Mr. Wylie acknowl­edged that Palan­tir and Cam­bridge Ana­lyt­i­ca nev­er signed a con­tract or entered into a for­mal busi­ness rela­tion­ship. But he said some Palan­tir employ­ees helped engi­neer Cambridge’s psy­cho­graph­ic mod­els.

“There were Palan­tir staff who would come into the office and work on the data,” Mr. Wylie told law­mak­ers. “And we would go and meet with Palan­tir staff at Palan­tir.” He did not pro­vide an exact num­ber for the employ­ees or iden­ti­fy them.

Palan­tir employ­ees were impressed with Cambridge’s back­ing from Mr. Mer­cer, one of the world’s rich­est men, accord­ing to mes­sages viewed by The Times. And Cam­bridge Ana­lyt­i­ca viewed Palantir’s Sil­i­con Val­ley ties as a valu­able resource for launch­ing and expand­ing its own busi­ness.

In an inter­view this month with The Times, Mr. Wylie said that Palan­tir employ­ees were eager to learn more about using Face­book data and psy­cho­graph­ics. Those dis­cus­sions con­tin­ued through spring 2014, accord­ing to Mr. Wylie.

Mr. Wylie said that he and Mr. Nix vis­it­ed Palantir’s Lon­don office on Soho Square. One side was set up like a high-secu­ri­ty office, Mr. Wylie said, with sep­a­rate rooms that could be entered only with par­tic­u­lar codes. The oth­er side, he said, was like a tech start-up — “weird inspi­ra­tional quotes and stuff on the wall and free beer, and there’s a Ping-Pong table.”

Mr. Chmieli­auskas con­tin­ued to com­mu­ni­cate with Mr. Wylie’s team in 2014, as the Cam­bridge employ­ees were locked in pro­tract­ed nego­ti­a­tions with a researcher at Cam­bridge Uni­ver­si­ty, Michal Kosin­s­ki, to obtain Face­book data through an app Mr. Kosin­s­ki had built. The data was cru­cial to effi­cient­ly scale up Cambridge’s psy­cho­met­rics prod­ucts so they could be used in elec­tions and for cor­po­rate clients.

“I had left field idea,” Mr. Chmieli­auskas wrote in May 2014. “What about repli­cat­ing the work of the cam­bridge prof as a mobile app that con­nects to face­book?” Repro­duc­ing the app, Mr. Chmieli­auskas wrote, “could be a valu­able lever­age nego­ti­at­ing with the guy.”

Those nego­ti­a­tions failed. But Mr. Wylie struck gold with anoth­er Cam­bridge researcher, the Russ­ian-Amer­i­can psy­chol­o­gist Alek­san­dr Kogan, who built his own per­son­al­i­ty quiz app for Face­book. Over sub­se­quent months, Dr. Kogan’s work helped Cam­bridge devel­op psy­cho­log­i­cal pro­files of mil­lions of Amer­i­can vot­ers.

———-

“Spy Contractor’s Idea Helped Cam­bridge Ana­lyt­i­ca Har­vest Face­book Data” by NICHOLAS CONFESSORE and MATTHEW ROSENBERG; The New York Times; 03/27/2018 [51]

“The rev­e­la­tions pulled Palan­tir — co-found­ed by the wealthy lib­er­tar­i­an Peter Thiel [53] — into the furor sur­round­ing Cam­bridge, which improp­er­ly obtained Face­book data to build ana­lyt­i­cal tools it deployed on behalf of Don­ald J. Trump and oth­er Repub­li­can can­di­dates in 2016. Mr. Thiel, a sup­port­er of Pres­i­dent Trump, serves on the board at Face­book.

Yep, a Face­book board mem­ber’s pri­vate intel­li­gence firm was work­ing close­ly with Cam­brige Ana­lyt­i­ca as they devel­oped their psy­cho­log­i­cal pro­fil­ing tech­nol­o­gy. It’s quite a rev­e­la­tion. The kind of explo­sive rev­e­la­tion that has Palan­tir first deny­ing that there was any rela­tion­ship at all, fol­lowed with acknowledgement/denial that, yes, a Palan­tir employ­ee, Alfredas Chmieli­auskas, was indeed work­ing with Cam­bridge Ana­lyt­i­ca but not on behalf of Palan­tir:

...
It was a Palan­tir employ­ee in Lon­don, work­ing close­ly with the data sci­en­tists build­ing Cambridge’s psy­cho­log­i­cal pro­fil­ing tech­nol­o­gy, who sug­gest­ed the sci­en­tists cre­ate their own app — a mobile-phone-based per­son­al­i­ty quiz — to gain access to Face­book users’ friend net­works, accord­ing to doc­u­ments obtained by The New York Times.

...

The Palan­tir employ­ee, Alfredas Chmieli­auskas, works on busi­ness devel­op­ment for the com­pa­ny, accord­ing to his LinkedIn page. In an ini­tial state­ment, Palan­tir said it had “nev­er had a rela­tion­ship with Cam­bridge Ana­lyt­i­ca, nor have we ever worked on any Cam­bridge Ana­lyt­i­ca data.” Lat­er on Tues­day, Palan­tir revised its account, say­ing that Mr. Chmieli­auskas was not act­ing on the company’s behalf when he advised Mr. Wylie on the Face­book data.
...

Adding the scan­dalous nature of it all is that Google CEO Eric Schmidt’s daugh­ter sud­den­ly appeared in June of 2013 to also pro­mote to her old boss at SCL a rela­tion­ship with Palan­tir:

...
Doc­u­ments and inter­views indi­cate that start­ing in 2013, Mr. Chmieli­auskas began cor­re­spond­ing with Mr. Wylie and a col­league from his Gmail account. At the time, Mr. Wylie and the col­league worked for the British defense and intel­li­gence con­trac­tor SCL Group, which formed Cam­bridge Ana­lyt­i­ca with Mr. Mer­cer the next year. The three shared Google doc­u­ments to brain­storm ideas about using big data to cre­ate sophis­ti­cat­ed behav­ioral pro­files, a prod­uct code-named “Big Dad­dy.”

A for­mer intern at SCL — Sophie Schmidt, the daugh­ter of Eric Schmidt, then Google’s exec­u­tive chair­man — urged the com­pa­ny to link up with Palan­tir, accord­ing to Mr. Wylie’s tes­ti­mo­ny and a June 2013 email viewed by The Times.

“Ever come across Palan­tir. Amus­ing­ly Eric Schmidt’s daugh­ter was an intern with us and is try­ing to push us towards them?” one SCL employ­ee wrote to a col­league in the email.

Ms. Schmidt did not respond to requests for com­ment, nor did a spokesman for Cam­bridge Ana­lyt­i­ca.
...

But this June 2013 pro­pos­al by Sophie Schmidt was­n’t what start­ed Cam­bridge Ana­lyt­i­ca’s rela­tion­ship with Palan­tir. Because that report­ed­ly start­ed in ear­ly 2013, when Alexan­der Nix and a Palan­tir exec­u­tive dis­cussed work­ing togeth­er on elec­tion cam­paigns:

...
In ear­ly 2013, Alexan­der Nix, an SCL direc­tor who became chief exec­u­tive of Cam­bridge Ana­lyt­i­ca, and a Palan­tir exec­u­tive dis­cussed work­ing togeth­er on elec­tion cam­paigns.
...

So Sophie Schmidt swooped in to pro­mote Palan­tir to Cam­bridge Ana­lyt­i­ca months after the nego­ti­a­tions began. It rais­es the ques­tion of who encour­aged her to do that.

Palan­tir now admits these nego­ti­a­tions hap­pened, but claims that they chose not to work with Cam­bridge Ana­lyt­i­ca because they “want­ed to steer clear of elec­tion work.” And emails indi­cate that Palan­tir did indeed for­mal­ly turn down the idea of work­ing with Cam­bridge Ana­lyt­i­ca since the emails show that Nix and Chmieli­auskas sought to revive talks about a for­mal part­ner­ship through ear­ly 2014, but Palan­tir exec­u­tives again declined. And yet, accord­ing to Christo­pher Wylie, some Palan­tir employ­ees helped engi­neer their psy­chogroph­ic mod­els. And that sug­gests Palan­tir turned down a for­mal rela­tion­ship in favor of an infor­mal one:

...
A Palan­tir spokes­woman acknowl­edged that the com­pa­nies had briefly con­sid­ered work­ing togeth­er but said that Palan­tir declined a part­ner­ship, in part because exec­u­tives there want­ed to steer clear of elec­tion work. Emails reviewed by The Times indi­cate that Mr. Nix and Mr. Chmieli­auskas sought to revive talks about a for­mal part­ner­ship through ear­ly 2014, but Palan­tir exec­u­tives again declined.

In his tes­ti­mo­ny, Mr. Wylie acknowl­edged that Palan­tir and Cam­bridge Ana­lyt­i­ca nev­er signed a con­tract or entered into a for­mal busi­ness rela­tion­ship. But he said some Palan­tir employ­ees helped engi­neer Cambridge’s psy­cho­graph­ic mod­els.

“There were Palan­tir staff who would come into the office and work on the data,” Mr. Wylie told law­mak­ers. “And we would go and meet with Palan­tir staff at Palan­tir.” He did not pro­vide an exact num­ber for the employ­ees or iden­ti­fy them.
...

“There were Palan­tir staff who would come into the office and work on the data...And we would go and meet with Palan­tir staff at Palan­tir.”

That sure sounds like a rela­tion­ship! For­mal or not.

And that infor­mal rela­tion­ship con­tin­ued dur­ing the peri­od when Cam­bridge Ana­lyt­i­ca was in nego­ti­a­tion with the ini­tial Uni­ver­si­ty of Cam­bridge Psy­cho­met­rics Cen­tre in 2014:

...
In an inter­view this month with The Times, Mr. Wylie said that Palan­tir employ­ees were eager to learn more about using Face­book data and psy­cho­graph­ics. Those dis­cus­sions con­tin­ued through spring 2014, accord­ing to Mr. Wylie.

Mr. Wylie said that he and Mr. Nix vis­it­ed Palantir’s Lon­don office on Soho Square. One side was set up like a high-secu­ri­ty office, Mr. Wylie said, with sep­a­rate rooms that could be entered only with par­tic­u­lar codes. The oth­er side, he said, was like a tech start-up — “weird inspi­ra­tional quotes and stuff on the wall and free beer, and there’s a Ping-Pong table.”

Mr. Chmieli­auskas con­tin­ued to com­mu­ni­cate with Mr. Wylie’s team in 2014, as the Cam­bridge employ­ees were locked in pro­tract­ed nego­ti­a­tions with a researcher at Cam­bridge Uni­ver­si­ty, Michal Kosin­s­ki, to obtain Face­book data through an app Mr. Kosin­s­ki had built. The data was cru­cial to effi­cient­ly scale up Cambridge’s psy­cho­met­rics prod­ucts so they could be used in elec­tions and for cor­po­rate clients.
...

And it was dur­ing those nego­ti­a­tions, in May of 2014, when Chmieli­auskas first pro­posed the idea of just repli­cat­ing what the Uni­ver­si­ty of Cam­bridge Psy­cho­met­rics Cen­tre was doing for lever­age in the nego­ti­a­tions. When those nego­ti­a­tions ulti­mate­ly failed, Cam­bridge Ana­lyt­i­ca found anoth­er Cam­bridge Uni­ver­si­ty psy­chol­o­gist, Alek­san­dr Kogan, to build the app for them:

...
“I had left field idea,” Mr. Chmieli­auskas wrote in May 2014. “What about repli­cat­ing the work of the cam­bridge prof as a mobile app that con­nects to face­book?” Repro­duc­ing the app, Mr. Chmieli­auskas wrote, “could be a valu­able lever­age nego­ti­at­ing with the guy.”

Those nego­ti­a­tions failed. But Mr. Wylie struck gold with anoth­er Cam­bridge researcher, the Russ­ian-Amer­i­can psy­chol­o­gist Alek­san­dr Kogan, who built his own per­son­al­i­ty quiz app for Face­book. Over sub­se­quent months, Dr. Kogan’s work helped Cam­bridge devel­op psy­cho­log­i­cal pro­files of mil­lions of Amer­i­can vot­ers.
...

And that’s what we know so far about the rela­tion­ship between Cam­bridge Ana­lyt­i­ca and Palan­tir. Which rais­es a num­ber of ques­tions. Like whether or not this infor­mal rela­tion­ship con­tin­ued well after Cam­bridge Ana­lyt­i­ca start­ed har­vest­ing all that Face­book infor­ma­tion. Let’s look at sev­en key the facts about we know Palan­tir’s involve­ment in this so far:

1. Palan­tir employ­ees helped build the psy­cho­graph­ic pro­files.

2. Mr. Chmieli­auskas was in con­tact with Wylie at least as late as May of 2014 as Cam­bridge Ana­lyt­i­ca was nego­ti­at­ing with the Uni­ver­si­ty of Cam­bridge’s Psy­cho­met­rics Cen­tre.

3. We don’t know when this infor­mal rela­tion­ship between Palan­tir and Cam­bridge Ana­lyt­i­ca end­ed.

4. We don’t know if the infor­mal rela­tion­ship between Palan­tir and Cam­bridge Ana­lyt­i­ca — which large­ly appears to cen­ter around Mr. Chmieli­auskas — real­ly was large­ly Chmieli­auskas’s ini­tia­tive alone after Palan­tir ini­tial­ly reject­ed a for­mal rela­tion­ship (it’s pos­si­ble) or if Chmieli­auskas was direct­ed to pur­sue this rela­tion­ship infor­mal­ly but on behalf of Palan­tir to main­tain deni­a­bil­i­ty in the case of awk­ward sit­u­a­tions like the present one (also very pos­si­ble, and savvy giv­en the cur­rent sit­u­a­tion).

5. We don’t know if the Palan­tir employ­ees who helped build those psy­cho­graph­ic pro­files were work­ing with the data Cam­bridge Ana­lyt­i­ca har­vest­ed from Face­book or were they work­ing with the ear­li­er, inad­e­quate sets of data that did­n’t include the Face­book data? Because if the Palan­tir employ­ees helped build the psy­cho­graph­ic pro­files based on the Face­book data that implies this infor­mal rela­tion­ship went on a lot longer than May of 2014 since that’s when it first start­ed get­ting col­lect­ed via Kogan’s app. How long? We don’t yet know.

6. Nei­ther do we know how much of this data ulti­mate­ly fell into the hands of Palan­tir. As Wylie described it, “There were Palan­tir staff who would come into the office and work on the data...And we would go and meet with Palan­tir staff at Palan­tir.” So did those Palan­tir employ­ees who were work­ing on “the data” take any of that data back to Palan­tir?

7. For that mat­ter, giv­en that Peter Thiel sits on the board of Face­book, and giv­en how freely Face­book hands out this kind of data, we have to ask the ques­tion of whether or not Palan­tir already has direct access to exact­ly the kind of data Cam­bridge Ana­lyt­i­ca was har­vest­ing. Did Palan­tir even need Cam­bridge Ana­lyt­i­ca’s data? Per­haps Palan­tir was already using apps of their own to har­vest this kind of data? We don’t know. At the same time, don’t for­get that even if Palan­tir had ready access to the same Face­book pro­file data gath­ered by Kogan’s app, it’s still pos­si­ble Palan­tir would have had an inter­est in the com­pa­ny pure­ly to see how the data was ana­lyzed and learn from that. In oth­er words, the inter­est in Cam­bridge Ana­lyt­i­ca may be been more relat­ed to the algo­rithms, and not the data, for Peter Thiel’s Palan­tir. Don’t for­get that if any­one is the real pow­er behind the throne at Face­book it’s prob­a­bly Thiel.

8. What on earth is going on with Sophie Schmidt, daugh­ter of Google CEO Eric Schmidt, push­ing Cam­bridge Ana­lyt­i­ca to work with Palan­tir in June of 2013, months after Cam­bridge Ana­lyt­ic and Palan­tir began talk­ing with each oth­er? That seems poten­tial­ly sig­nif­i­cant.

Those are just some of the ques­tions raised about Palan­tir’s ambigu­ous­ly omi­nous rela­tion­ship with Cam­bridge Ana­lyt­i­ca. Bad don’t for­get that it’s not just Palan­tir that we need to ask these kinds of ques­tions. For instance, what about Steve Ban­non’s Bre­it­bart? Does Bre­it­bart, home the neo-Nazi ‘Alt Right’, also have access to all that har­vest­ed Cam­bridge Ana­lyt­i­ca data? Not just the raw Face­book data but also the processed psy­cho­log­i­cal pro­file data on 50 mil­lion Amer­i­cans that Cam­bridge Ana­lyt­i­ca gen­er­at­ed. Does Bre­it­bart have the processed pro­files too? And what about the Repub­li­can Par­ty? And all the oth­er enti­ties out there who gained access to this Face­book pro­file data. Just how many dif­fer­ent enti­ties around the globe pos­sess that Cam­bridge Ana­lyt­i­ca data set?

It’s Not Just Cam­bridge Ana­lyt­i­ca. Or Face­book. Or Google. It’s Soci­ety.

Of course, as we saw with Sandy Parak­i­las’s whis­tle-blow­er claims, when it comes to the ques­tion of who might pos­sess Face­book pro­file data har­vest­ed dur­ing the 2007–2014 peri­od when Face­book had “friends per­mis­sions” pol­i­cy, the list of sus­pects includes poten­tial­ly hun­dreds of thou­sands of devel­op­ers and any­one who has pur­chased this infor­ma­tion on the black mar­ket.

Don’t for­get one of the oth­er amaz­ing aspects of this whole sit­u­a­tion: if hun­dreds of thou­sands of devel­op­ers were using this fea­ture to scrape user pro­files, that means this real­ly was an open secret. Lots and lots of peo­ple were doing this. For years. So, like many scan­dals, per­haps the most scan­dalous part of it is that we’re learn­ing about some­thing we should have known all along and many of did know all along. It’s not like it’s a secret that peo­ple are being sur­veilled in detail in the inter­net age and this data is being stored and aggre­gat­ed in pub­lic and pri­vate data­bas­es and put up for sale. We’ve col­lec­tive­ly known this all along. At least on some lev­el.

And yet this sur­veil­lance is so per­va­sive that it’s almost nev­er thought about on a moment by moment basis at an indi­vid­ual lev­el. When peo­ple browse the web they pre­sum­ably aren’t think­ing about the vol­ume of track­ing cook­ies and oth­er per­son­al infor­ma­tion slurped up as a result of that mouse click. Nor are they think­ing about how that click con­tributes to the numer­ous per­son­al pro­files of them float­ing around the com­mer­cial data bro­ker­age mar­ket­place. So in a more fun­da­men­tal sense we don’t actu­al­ly know we’re being sur­veilled because we’re not think­ing about it.

It’s one exam­ple of how humans aren’t wired to nat­u­ral­ly think about the macro forces impact­ing their lives in day to day deci­sions, which was fine when we were cave men but becomes a prob­lem­at­ic instinct when we’re lit­er­al­ly mas­ter­ing the laws of physics and shap­ing our world and envi­ron­ment. From physics and nature to his­to­ry and con­tem­po­rary trends, the vast major­i­ty of human­i­ty spends very lit­tle time study­ing these top­ics. Which is com­plete­ly under­stand­able giv­en the lack of time or resources to do so, but that under­stand­able instinct cre­ates world per­fect­ly set up for abuse by sur­veil­lance states, both pub­lic and pri­vate, which makes it less under­stand­able and much more prob­lem­at­ic.

So, in the inter­est of gain­ing per­spec­tive on how we got to this point where the Face­book emerged as an ever-grow­ing Panop­ti­con in just a few short years after its con­cep­tion, let’s take a look at one last arti­cle. It’s an arti­cle by inves­tiga­tive jour­nal­ist Yasha Levine, who recent­ly pub­lished the must-read book Sur­veil­lance Val­ley: The Secret Mil­i­tary His­to­ry of the Inter­net [55]. It’s a book filled with vital his­tor­i­cal fun fact about the inter­net. Fun facts like...

1. How the inter­net began as a sys­tem built for nation­al secu­ri­ty pur­pos­es with a focus on mil­i­tary hard­ware and com­mand and con­trol com­mu­ni­ca­tion pur­pos­es in gen­er­al. But there was also a focus on build­ing a sys­tem that could col­lect, store, process, and dis­trib­ute of mas­sive vol­umes of infor­ma­tion used to wage the Viet­nam war. Beyond that, these ear­ly com­put­er net­works also act­ed as a col­lec­tion and shar­ing sys­tem for deal­ing with domes­tic nation­al secu­ri­ty con­cerns (con­cerns that cen­tered around track­ing anti-war pro­test­ers, civ­il rights activists, etc). That’s what the inter­net start­ed out as. A sys­tem for stor­ing data about peo­ple and con­flict for US nation­al secu­ri­ty pur­pos­es.

2. Build­ing data­bas­es of pro­files on peo­ple (for­eign and domes­tic) was one of the very first goals of these inter­net pre­de­ces­sors. In fact, one of the key vision­ar­ies behind the devel­op­ment of the inter­net, Ithiel de Sola Pool, both helped shape the devel­op­ment of the ear­ly inter­net as a sur­veil­lance and coun­terin­sur­gency tech­nol­o­gy and also pio­neered data-dri­ven elec­tion cam­paigns. He even start­ed a pri­vate firm to do this: Simul­mat­ics. Pool’s vision was a world where the sur­veil­lance state act­ed as a benign mas­ter that the kept the peace peace­ful­ly by using supe­ri­or knowl­edge to nudge peo­ple in the ‘right’ direc­tion.

3. This vision of vast data­base of per­son­al pro­files for the pur­pose was large­ly a secret at first, but it did­n’t remain that way. And there was actu­al­ly quite a bit of pub­lic para­noia in the US about these inter­net-pre­de­ces­sors, espe­cial­ly with­in the anti-Viet­nam war activist com­mu­ni­ties. Flash for­ward a cou­ple decades and that para­noia has fad­ed almost entirely...until scan­dals like the cur­rent one erupt and we tem­porar­i­ly grow con­cerned.

4. What Cam­bridge Ana­lyt­i­ca is accused of doing is what the data giants like Face­book and Google do every day and have been going for years. And it’s not just the giants. Small­er firms are scoop­ing up fast amounts of infor­ma­tion too...it’s just not as vast as what the giants are col­lect­ing. Even cute apps, like the wild­ly pop­u­lar Angry Birds, has been found to col­lect all sorts of data about users.

5. While it’s great that pub­lic atten­tion is being direct­ed at the kind of sleazy manip­u­la­tive activ­i­ties Cam­bridge Ana­lyt­i­ca was engag­ing in, decep­tive­ly wield­ing real pow­er over real unwit­ting peo­ple, it is a wild mis­char­ac­ter­i­za­tion to act like Cam­bridge Ana­lyt­i­ca was exert­ing mass mind-con­trol over the mass­es using inter­net mar­ket­ing voodoo. What Cam­bridge Ana­lyt­i­ca, or any of the oth­er sleazy manip­u­la­tors, were doing was indeed influ­en­tial, but it needs to be viewed in the con­text of a polit­i­cal state of affairs where mas­sive num­bers of Amer­i­cans, includ­ing Trump vot­ers, real­ly have been col­lec­tive­ly failed by the Amer­i­can pow­er estab­lish­ment for decades. The col­lapse of the Amer­i­can mid­dle class and rise of the plu­toc­ra­cy is what cre­at­ed the kind of macro envi­ron­ment where car­ni­val bark­er like Don­ald Trump could use firms like Cam­bridge Ana­lyt­i­ca to ‘nudge’ peo­ple in the direc­tion of vot­ing for him. In oth­er words, the focus on Cam­bridge Ana­lyt­i­ca’s manip­u­la­tion of peo­ple’s psy­cho­log­i­cal pro­files in the absence of the recog­ni­tion of the mas­sive polit­i­cal fail­ures of last sev­er­al decades in Amer­i­ca — the mass socioe­co­nom­ic fail­ures of the Amer­i­can embrace of ‘Reaganon­ics’ and right-wing eco­nom­ic gospel cou­pled with the Amer­i­can Left­’s fail­ure to effec­tive­ly repu­di­ate these doc­trines — is pro­found­ly ahis­tor­i­cal. The sto­ry of the rise of the pow­er of firms like Face­book, Google, and Cam­bridge Ana­lyt­i­ca is a sto­ry the implic­it­ly includes the sto­ry of that entire his­to­ry of political/socioeconomic fail­ures tied to fail­ure to effec­tive­ly respond to the rise of the Amer­i­can right-wing over the last sev­er­al decades. And we are mak­ing a mas­sive mis­take if we for­get that. Cam­bridge Ana­lyt­i­ca would­n’t have been near­ly as effec­tive in nudg­ing peo­ple towards vot­ing for some­one like Trump if so many peo­ple weren’t already so ready to burn the cur­rent sys­tem down.

These are the kinds of his­tor­i­cal chap­ters that can’t be left out of any analy­sis of Cam­bridge Ana­lyt­i­ca. Because Cam­bridge Ana­lyt­i­ca isn’t the excep­tion. It’s an excep­tion­al­ly sleazy exam­ple of the rules we’ve been play­ing by for a while, whether we real­ized it or not [56]:

The Baf­fler

The Cam­bridge Ana­lyt­i­ca Con

Yasha Levine,
March 21, 2018

“The man with the prop­er imag­i­na­tion is able to con­ceive of any com­mod­i­ty in such a way that it becomes an object of emo­tion to him and to those to whom he imparts his pic­ture, and hence cre­ates desire rather than a mere feel­ing of ought.”

Wal­ter Dill Scott, Influ­enc­ing Men in Busi­ness: Psy­chol­o­gy of Argu­ment and Sug­ges­tion (1911)

This week, Cam­bridge Ana­lyt­i­ca, the British elec­tion data out­fit fund­ed by bil­lion­aire Robert Mer­cer and linked to Steven Ban­non and Pres­i­dent Don­ald Trump, blew up the news cycle. The charge, as report­ed by twin exposés in the New York Times [3] and the Guardian [42], is that the firm inap­pro­pri­ate­ly accessed Face­book pro­file infor­ma­tion belong­ing to 50 mil­lion peo­ple and then used that data to con­struct a pow­er­ful inter­net-based psy­cho­log­i­cal influ­ence weapon. This new­fan­gled con­struct was then used to brain­wash-car­pet-bomb the Amer­i­can elec­torate, shred­ding our democ­ra­cy and turn­ing peo­ple into pli­able zom­bie sup­port­ers of Don­ald Trump.

In the words of a pink-haired Cam­bridge Ana­lyt­i­ca data-war­rior-turned-whistle­blow­er, the com­pa­ny served as a dig­i­tal armory that turned “Likes” into weapons and pro­duced “Steve Bannon’s psy­cho­log­i­cal war­fare mind­fuck tool.”

Scary, right? Makes me won­der if I’m still not under Cam­bridge Analytica’s influ­ence right now.

Nat­u­ral­ly, there are also rumors of a nefar­i­ous Russ­ian con­nec­tion. And appar­ent­ly there’s more dirt com­ing. Chan­nel 4 News in Britain just pub­lished an inves­ti­ga­tion show­ing top Cam­bridge Ana­lyt­i­ca execs brag­ging to an under­cov­er reporter that their team uses high-tech psy­cho­me­t­ric voodoo to win elec­tions for clients all over the world, but also dab­bles in tra­di­tion­al meat­space tech­niques as well: bribes, kom­pro­mat, black­mail, Ukrain­ian escort honeypots—you know, the works.

It’s good that the main­stream news media are final­ly start­ing to pay atten­tion to this dark cor­ner of the inter­net —and pro­duc­ing exposés of shady sub rosa polit­i­cal cam­paigns and their eager exploita­tion of our online dig­i­tal trails in order to con­t­a­m­i­nate our infor­ma­tion streams and influ­ence our deci­sions. It’s about time.

But this sto­ry is being cov­ered and framed in a mis­lead­ing way. So far, much of the main­stream cov­er­age, dri­ven by the Times and Guardian reports, looks at Cam­bridge Ana­lyt­i­ca in isolation—almost entire­ly out­side of any his­tor­i­cal or polit­i­cal con­text. This makes it seem to read­ers unfa­mil­iar with the long his­to­ry of the strug­gle for con­trol of the dig­i­tal sphere as if the main prob­lem is that the bad actors at Cam­bridge Ana­lyt­i­ca crossed the trans­mis­sion wires of Face­book in the Promethean man­ner of Vic­tor Frankenstein—taking what were nor­mal­ly respectable, sci­en­tif­ic data pro­to­cols and per­vert­ing them to serve the dia­bol­i­cal aim of rean­i­mat­ing the decom­pos­ing lump of polit­i­cal flesh known as Don­ald Trump.

So if we’re going to view the actions of Cam­bridge Ana­lyt­i­ca in their prop­er light, we need first to start with an admis­sion. We must con­cede that covert influ­ence is not some­thing unusu­al or for­eign to our soci­ety, but is as Amer­i­can as apple pie and free­dom fries. The use of manip­u­la­tive, psy­cho­log­i­cal­ly dri­ven adver­tis­ing and mar­ket­ing tech­niques to sell us prod­ucts, lifestyles, and ideas has been the foun­da­tion of mod­ern Amer­i­can soci­ety [57], going back to the days of the self-styled inven­tor of pub­lic rela­tions, Edward Bernays. It oozes out of every pore on our body politic. It’s what holds our ail­ing con­sumer soci­ety togeth­er. And when it comes to mar­ket­ing can­di­dates and polit­i­cal mes­sages, using data to influ­ence peo­ple and shape their deci­sions has been the holy grail of the com­put­er age, going back half a cen­tu­ry.

Let’s start with the basics: What Cam­bridge Ana­lyt­i­ca is accused of doing—siphoning people’s data, com­pil­ing pro­files, and then deploy­ing that infor­ma­tion to influ­ence them to vote a cer­tain way—Facebook and Sil­i­con Val­ley giants like Google do every day, indeed, every minute we’re logged on, on a far greater and more inva­sive scale.

Today’s inter­net busi­ness ecosys­tem is built on for-prof­it sur­veil­lance, behav­ioral pro­fil­ing, manip­u­la­tion and influ­ence. That’s the name of the game. It isn’t just Face­book or Cam­bridge Ana­lyt­i­ca or even Google. It’s Ama­zon. It’s eBay. It’s Palan­tir. It’s Angry Birds. It’s MoviePass [58]. It’s Lock­heed Mar­tin [59]. It’s every app you’ve ever down­loaded. Every phone you bought. Every pro­gram you watched on your on-demand cable TV pack­age.

All of these games, apps, and plat­forms prof­it from the con­cert­ed siphon­ing up of all data trails to pro­duce pro­files for all sorts of micro-tar­get­ed influ­ence ops in the pri­vate sec­tor. This com­merce in user data per­mit­ted Face­book to earn $40 bil­lion last year, while Google raked in $110 bil­lion.

What do these com­pa­nies know about us, their users? Well, just about every­thing.

Sil­i­con Val­ley of course keeps a tight lid on this infor­ma­tion, but you can get a glimpse of the kinds of data our pri­vate dig­i­tal dossiers con­tain by trawl­ing through their patents. Take, for instance, a series of patents Google filed in the mid-2000s for its Gmail-tar­get­ed adver­tis­ing tech­nol­o­gy. The lan­guage, stripped of opaque tech jar­gon, revealed that just about every­thing we enter into Google’s many prod­ucts and platforms—from email cor­re­spon­dence to Web search­es and inter­net browsing—is ana­lyzed and used to pro­file users in an extreme­ly inva­sive and per­son­al way. Email cor­re­spon­dence is parsed for mean­ing and sub­ject mat­ter. Names are matched to real iden­ti­ties and address­es. Email attachments—say, bank state­ments or test­ing results from a med­ical lab—are scraped for infor­ma­tion. Demo­graph­ic and psy­cho­graph­ic data, includ­ing social class, per­son­al­i­ty type, age, sex, polit­i­cal affil­i­a­tion, cul­tur­al inter­ests, social ties, per­son­al income, and mar­i­tal sta­tus is extract­ed. In one patent, I dis­cov­ered that Google appar­ent­ly had the abil­i­ty to deter­mine if a per­son was a legal U.S. res­i­dent or not. It also turned out you didn’t have to be a reg­is­tered Google user to be snared in this pro­fil­ing appa­ra­tus. All you had to do was com­mu­ni­cate with some­one who had a Gmail address.

On the whole, Google’s pro­fil­ing phi­los­o­phy was no dif­fer­ent than Facebook’s, which also con­structs “shad­ow pro­files” to col­lect and mon­e­tize data, even if you nev­er had a reg­is­tered Face­book or Gmail account.

It’s not just the big plat­form monop­o­lies that do this, but all the small­er com­pa­nies that run their busi­ness­es on ser­vices oper­at­ed by Google and Face­book. It even includes cute games [60] like Angry Birds, devel­oped by Finland’s Rovio Enter­tain­ment, that’s been down­loaded more than a bil­lion times. The Android ver­sion of Angry Birds was found to pull per­son­al data [61] on its play­ers, includ­ing eth­nic­i­ty, mar­i­tal sta­tus, and sex­u­al orientation—including options for the “sin­gle,” “mar­ried,” “divorced,” “engaged,” and “swinger” cat­e­gories. Pulling per­son­al data like this didn’t con­tra­dict Google’s terms of ser­vices for its Android plat­form. Indeed, for-prof­it sur­veil­lance was the whole point of why Google start­ed plan­ning to launch an iPhone rival as far back as 2004.

In launch­ing Android, Google made a gam­ble [60] that by releas­ing its pro­pri­etary oper­at­ing sys­tem to man­u­fac­tur­ers free of charge, it wouldn’t be rel­e­gat­ed to run­ning apps on Apple iPhone or Microsoft Mobile Win­dows like some kind of dig­i­tal sec­ond-class cit­i­zen. If it played its cards right and Android suc­ceed­ed, Google would be able to con­trol the envi­ron­ment that under­pins the entire mobile expe­ri­ence, mak­ing it the ulti­mate gate­keep­er of the many mon­e­tized inter­ac­tions among users, apps, and adver­tis­ers. And that’s exact­ly what hap­pened. Today, Google monop­o­lizes the smart phone mar­ket and dom­i­nates the mobile for-prof­it sur­veil­lance busi­ness [62].

These detailed psy­cho­log­i­cal pro­files, togeth­er with the direct access to users that plat­forms like Google and Face­book deliv­er, make both com­pa­nies cat­nip to adver­tis­ers, PR flacks—and dark-mon­ey polit­i­cal out­fits like Cam­bridge Ana­lyt­i­ca.

Indeed, polit­i­cal cam­paigns showed an ear­ly and pro­nounced affin­i­ty for the idea of tar­get­ed access and influ­ence on plat­forms like Face­book. Instead of blan­ket­ing air­waves with a sin­gle polit­i­cal ad, they could show peo­ple ads that appealed specif­i­cal­ly to the issues they held dear. They could also ensure that any such mes­sage spread through a tar­get­ed person’s larg­er social net­work through repost­ing and shar­ing.

The enor­mous com­mer­cial inter­est that polit­i­cal cam­paigns have shown in social media has earned them priv­i­leged atten­tion from Sil­i­con Val­ley plat­forms in return. Face­book runs a sep­a­rate polit­i­cal divi­sion specif­i­cal­ly geared to help its cus­tomers tar­get and influ­ence vot­ers.

The com­pa­ny even allows polit­i­cal cam­paigns to upload their own lists of poten­tial vot­ers and sup­port­ers direct­ly into Facebook’s data sys­tem. So armed, dig­i­tal polit­i­cal oper­a­tives can then use those people’s social net­works to iden­ti­fy oth­er prospec­tive vot­ers who might be sup­port­ive of their candidate—and then tar­get them with a whole new tidal wave of ads. “There’s a lev­el of pre­ci­sion that doesn’t exist in any oth­er medi­um,” Crys­tal Pat­ter­son, a Face­book employ­ee who works with gov­ern­ment and pol­i­tics cus­tomers, told the New York Times back in 2015. “It’s get­ting the right mes­sage to the right peo­ple at the right time.”

Nat­u­ral­ly, a whole slew of com­pa­nies and oper­a­tives in our increas­ing­ly data-dri­ven elec­tion scene have cropped up over the last decade to plug in to these amaz­ing influ­ence machines. There is a whole con­stel­la­tion of them work­ing all sorts of strate­gies: tra­di­tion­al vot­er tar­get­ing, polit­i­cal pro­pa­gan­da mills, troll armies [63], and bots [64].

Some of these firms are polit­i­cal­ly agnos­tic; they’ll work for any­one with cash. Oth­ers are par­ti­san. The Demo­c­ra­t­ic Par­ty Data Death Star [65] is NGP VAN [66]. The Repub­li­cans have a few of their own—including i360, a data mon­ster gen­er­ous­ly fund­ed by Charles Koch. Nat­u­ral­ly, i360 part­ners with Face­book to deliv­er tar­get vot­ers. It also claims to have 700 per­son­al data points cross-tab­u­lat­ed on 199 mil­lion vot­ers and near­ly 300 mil­lion con­sumers, with the abil­i­ty to pro­file and tar­get them with pin-point accu­ra­cy based on their beliefs and views.

Here’s how The Nation­al Jour­nal’s Andrew Rice described i360 in 2015:

Like Google, the Nation­al Secu­ri­ty Agency, or the Demo­c­ra­t­ic data machine, i360 has a vora­cious appetite for per­son­al infor­ma­tion. It is con­stant­ly ingest­ing new data into its tar­get­ing sys­tems, which pre­dict not only par­ti­san iden­ti­fi­ca­tion but also sen­ti­ments about issues such as abor­tion, tax­es, and health care. When I vis­it­ed the i360 office, an employ­ee gave me a demon­stra­tion, zoom­ing in on a map to focus on a par­tic­u­lar 66-year-old high school teacher who lives in an apart­ment com­plex in Alexan­dria, Vir­ginia. . . . Though the adver­tis­ing indus­try typ­i­cal­ly eschews address­ing any sin­gle individual—it’s not just inva­sive, it’s also inefficient—it is becom­ing com­mon­place to tar­get extreme­ly nar­row audi­ences. So the school­teacher, along with a few look-alikes, might see a tai­lored ad the next time she clicks on YouTube.

Sil­i­con Val­ley doesn’t just offer cam­paigns a neu­tral plat­form; it also works close­ly along­side polit­i­cal can­di­dates to the point that the biggest inter­net com­pa­nies have become an exten­sion of the Amer­i­can polit­i­cal sys­tem. As one recent study showed, tech com­pa­nies rou­tine­ly embed their employ­ees inside major polit­i­cal cam­paigns: “Face­book, Twit­ter, and Google go beyond pro­mot­ing their ser­vices and facil­i­tat­ing dig­i­tal adver­tis­ing buys, active­ly shap­ing cam­paign com­mu­ni­ca­tion through their close col­lab­o­ra­tion with polit­i­cal staffers . . . these firms serve as qua­si-dig­i­tal con­sul­tants to cam­paigns, shap­ing dig­i­tal strat­e­gy, con­tent, and exe­cu­tion.”

In 2008, the hip young Black­ber­ry-tot­ing Barack Oba­ma was the first major-par­ty can­di­date on the nation­al scene to tru­ly lever­age the pow­er of inter­net-tar­get­ed agit­prop. With help from Face­book cofounder Chris Hugh­es, who built and ran Obama’s inter­net cam­paign divi­sion, the first Oba­ma cam­paign built an inno­v­a­tive micro-tar­get­ing ini­tia­tive to raise huge amounts of mon­ey in small chunks direct­ly from Obama’s sup­port­ers and sell his mes­sage with a hith­er­to unprece­dent­ed laser-guid­ed pre­ci­sion in the gen­er­al elec­tion cam­paign.

...

Now, of course, every elec­tion is a Face­book Elec­tion. And why not? As Bloomberg News has not­ed [67], Sil­i­con Val­ley ranks elec­tions “along­side the Super Bowl and the Olympics in terms of events that draw block­buster ad dol­lars and boost engage­ment.” In 2016, $1 bil­lion was spent on dig­i­tal advertising—with the bulk going to Face­book, Twit­ter, and Google.

What’s inter­est­ing here is that because so much mon­ey is at stake, there are absolute­ly no rules that would restrict any­thing an unsa­vory polit­i­cal appa­ratchik or a Sil­i­con Val­ley oli­garch might want to foist on the unsus­pect­ing dig­i­tal pub­lic. Creep­i­ly, Facebook’s own inter­nal research divi­sion car­ried out exper­i­ments show­ing that the plat­form could influ­ence people’s emo­tion­al state in con­nec­tion to a cer­tain top­ic or event. Com­pa­ny engi­neers call this fea­ture “emo­tion­al con­ta­gion [68]”—i.e., the abil­i­ty to viral­ly influ­ence people’s emo­tions and ideas just through the con­tent of sta­tus updates. In the twist­ed econ­o­my of emo­tion­al con­ta­gion, a neg­a­tive post by a user sup­press­es pos­i­tive posts by their friends, while a pos­i­tive post sup­press­es neg­a­tive posts. “When a Face­book user posts, the words they choose influ­ence the words cho­sen lat­er by their friends,” explained [69] the company’s lead sci­en­tist on this study.

On a very basic lev­el, Facebook’s opaque con­trol of its feed algo­rithm means the plat­form has real pow­er over people’s ideas and actions dur­ing an elec­tion. This can be done by a data shift as sim­ple and sub­tle as imper­cep­ti­bly tweak­ing a person’s feed to show more posts from friends who are, say, sup­port­ers of a par­tic­u­lar polit­i­cal can­di­date or a spe­cif­ic polit­i­cal idea or event. As far as I know, there is no law pre­vent­ing Face­book from doing just that: it’s plain­ly able and will­ing to influ­ence a user’s feed based on polit­i­cal aims—whether done for inter­nal cor­po­rate objec­tives, or due to pay­ments from polit­i­cal groups, or by the per­son­al pref­er­ences of Mark Zucker­berg.

So our present-day freak­out over Cam­bridge Ana­lyt­i­ca needs to be put in the broad­er his­tor­i­cal con­text of our decades-long com­pla­cen­cy over Sil­i­con Valley’s busi­ness mod­el. The fact is that com­pa­nies like Face­book and Google are the real mali­cious actors here—they are vital pub­lic com­mu­ni­ca­tions sys­tems that run on pro­fil­ing and manip­u­la­tion for pri­vate prof­it with­out any reg­u­la­tion or demo­c­ra­t­ic over­sight from the soci­ety in which it oper­ates. But, hey, let’s blame Cam­bridge Ana­lyt­i­ca. Or bet­ter yet, take a cue from the Times and blame the Rus­sians [9] along with Cam­bridge Ana­lyt­i­ca.

***

There’s anoth­er, big­ger cul­tur­al issue with the way we’ve begun to exam­ine and dis­cuss Cam­bridge Analytica’s bat­tery of inter­net-based influ­ence ops. Peo­ple are still daz­zled by the idea that the inter­net, in its pure, untaint­ed form, is some kind of mag­ic machine dis­trib­ut­ing democ­ra­cy and egal­i­tar­i­an­ism across the globe with the touch of a few key­strokes. This is the gospel preached by a stal­wart cho­rus of Net prophets, from Jeff Jarvis and the late John Per­ry Bar­low to Clay Shirky and Kevin Kel­ly. These char­la­tans all feed on an hon­or­able demo­c­ra­t­ic impulse: peo­ple still want to des­per­ate­ly believe in the utopi­an promise of this technology—its abil­i­ty to equal­ize pow­er, end cor­rup­tion, top­ple cor­po­rate media monop­o­lies, and empow­er the indi­vid­ual.

This mythology—which is of course aggres­sive­ly con­fect­ed for mass con­sump­tion by Sil­i­con Val­ley mar­ket­ing and PR outfits—is deeply root­ed in our cul­ture; it helps explain why oth­er­wise seri­ous jour­nal­ists work­ing for main­stream news out­lets can uniron­i­cal­ly employ phras­es such as “infor­ma­tion wants to be free” and “Facebook’s engine of democ­ra­cy” and get away with it.

The truth is that the inter­net has nev­er been about egal­i­tar­i­an­ism or democ­ra­cy.

The ear­ly inter­net came out of a series of Viet­nam War coun­terin­sur­gency projects aimed at devel­op­ing com­put­er tech­nol­o­gy that would give the gov­ern­ment a way to man­age a com­plex series of glob­al com­mit­ments and to mon­i­tor and pre­vent polit­i­cal strife—both at home and abroad. The inter­net, going back to its first incar­na­tion as the ARPANET mil­i­tary net­work, was always about sur­veil­lance, pro­fil­ing, and tar­get­ing [70].

The influ­ence of U.S. coun­terin­sur­gency doc­trine on the devel­op­ment of mod­ern com­put­ers and the inter­net is not some­thing that many peo­ple know about. But it is a sub­ject that I explore at length in my book, Sur­veil­lance Val­ley. So what jumps out at me is how seam­less­ly the report­ed activ­i­ties of Cam­bridge Ana­lyt­i­ca fit into this his­tor­i­cal nar­ra­tive.

Cam­bridge Ana­lyt­i­ca is a sub­sidiary of the SCL Group, a mil­i­tary con­trac­tor set up by a spooky huck­ster named Nigel Oakes that sells itself as a high-pow­ered con­clave of experts spe­cial­iz­ing in data-dri­ven coun­terin­sur­gency. It’s done work for the Pen­ta­gon, NATO, and the UK Min­istry of Defense in places like Afghanistan and Nepal [71], where it says it ran a “cam­paign to reduce and ulti­mate­ly stop the large num­bers of Maoist insur­gents in Nepal from break­ing into hous­es in remote areas to steal food, harass the home­own­ers and cause dis­rup­tion.”

In the grander scheme of high-tech coun­terin­sur­gency boon­dog­gles, which fea­tures such sto­ried psy-ops out­fits as Peter Thiel’s Palan­tir and Cold War dinosaurs like Lock­heed Mar­tin, the SCL Group appears to be a com­par­a­tive­ly minor play­er. Nev­er­the­less, its ambi­tious claims to recon­fig­ure the world order with some well-placed algo­rithms recalls one of the first major play­ers in the field: Simul­mat­ics, a 1960s coun­terin­sur­gency mil­i­tary con­trac­tor that pio­neered data-dri­ven elec­tion cam­paigns and whose founder, Ithiel de Sola Pool, helped shape the devel­op­ment of the ear­ly inter­net as a sur­veil­lance and coun­terin­sur­gency tech­nol­o­gy.

Ithiel de Sola Pool descend­ed from a promi­nent rab­bini­cal fam­i­ly that traced its roots to medieval Spain. Vir­u­lent­ly anti­com­mu­nist and tech-obsessed, he got his start in polit­i­cal work in 1950s work­ing on project [72] at the Hoover Insti­tu­tion at Stan­ford Uni­ver­si­ty that sought to under­stand the nature and caus­es of left-wing rev­o­lu­tions and reduce their like­ly course down to a math­e­mat­i­cal for­mu­la.

He then moved to MIT and made a name for him­self help­ing cal­i­brate the mes­sag­ing of John F. Kennedy’s 1960 pres­i­den­tial cam­paign. His idea was to mod­el the Amer­i­can elec­torate by decon­struct­ing each vot­er into 480 data points that defined every­thing from their reli­gious views to racial atti­tudes to socio-eco­nom­ic sta­tus. He would then use that data to run sim­u­la­tions on how they would respond to a par­tic­u­lar message—and those tri­al runs would per­mit major cam­paigns to fine-tune their mes­sages accord­ing­ly.

These new tar­get­ed mes­sag­ing tac­tics, enabled by rudi­men­ta­ry com­put­ers, had many fans in the per­ma­nent polit­i­cal class of Wash­ing­ton; their liveli­hoods, after all, were large­ly root­ed in their claims to ana­lyze and pre­dict polit­i­cal behav­ior. And so Pool lever­aged his research to launch Simul­mat­ics, a data ana­lyt­ics start­up that offered com­put­er sim­u­la­tion ser­vices to major Amer­i­can cor­po­ra­tions, help­ing them pre-test prod­ucts and con­struct adver­tis­ing cam­paigns.

Simul­mat­ics also did a brisk busi­ness as a mil­i­tary and intel­li­gence con­trac­tor. It ran sim­u­la­tions for Radio Lib­er­ty, the CIA’s covert anti-com­mu­nist radio sta­tion, help­ing the agency mod­el the Sovi­et Union’s inter­nal com­mu­ni­ca­tion sys­tem in order to pre­dict the effect that for­eign news broad­casts would have on the country’s polit­i­cal sys­tem. At the same time, Simul­mat­ics ana­lysts were doing coun­terin­sur­gency work under an ARPA con­tract in Viet­nam, con­duct­ing inter­views and gath­er­ing data to help mil­i­tary plan­ners under­stand why Viet­namese peas­ants rebelled and resist­ed Amer­i­can paci­fi­ca­tion efforts. Simulmatic’s work in Viet­nam was just one piece of a bru­tal Amer­i­can coun­terin­sur­gency pol­i­cy that involved covert pro­grams of assas­si­na­tions, ter­ror, and tor­ture that col­lec­tive­ly came to be known as the Phoenix Pro­gram.

At the same time, Pool was also per­son­al­ly involved in an ear­ly ARPANET-con­nect­ed ver­sion of Thiel’s Palan­tir effort—a pio­neer­ing sys­tem that would allow mil­i­tary plan­ners and intel­li­gence to ingest and work with large and com­plex data sets. Pool’s pio­neer­ing work won him a devot­ed fol­low­ing among a group of tech­nocrats who shared a utopi­an belief in the pow­er of com­put­er sys­tems to run soci­ety from the top down in a har­mo­nious man­ner. They saw the left-wing upheavals of the 1960s not as a polit­i­cal or ide­o­log­i­cal prob­lem but as a chal­lenge of man­age­ment and engi­neer­ing. Pool fed these rever­ies by set­ting out to build com­put­er­ized sys­tems that could mon­i­tor the world in real time and ren­der people’s lives trans­par­ent. He saw these sur­veil­lance and man­age­ment regimes in utopi­an terms—as a vital tool to man­age away social strife and con­flict. “Secre­cy in the modem world is gen­er­al­ly a desta­bi­liz­ing fac­tor,” he wrote in a 1969 essay. “Noth­ing con­tributes more to peace and sta­bil­i­ty than those activ­i­ties of elec­tron­ic and pho­to­graph­ic eaves­drop­ping, of con­tent analy­sis and tex­tu­al inter­pre­ta­tion.”

With the advent of cheap­er com­put­er tech­nol­o­gy in the 1960s, cor­po­rate and gov­ern­ment data­bas­es were already mak­ing a good deal of Pool’s prophe­cy come to pass, via sophis­ti­cat­ed new modes of con­sumer track­ing and pre­dic­tive mod­el­ing. But rather than greet­ing such advances as the augurs of a new demo­c­ra­t­ic mir­a­cle, peo­ple at the time saw it as a threat. Crit­ics across the polit­i­cal spec­trum warned that the pro­lif­er­a­tion of these tech­nolo­gies would lead to cor­po­ra­tions and gov­ern­ments con­spir­ing to sur­veil, manip­u­late, and con­trol soci­ety.

This fear res­onat­ed with every part of the culture—from the new left to prag­mat­ic cen­trists and reac­tionary South­ern Democ­rats. It prompt­ed some high-pro­file exposés in papers like the New York Times and Wash­ing­ton Post. It was report­ed on in trade mag­a­zines of the nascent com­put­er indus­try like Com­put­er­World. And it com­mand­ed prime real estate in estab­lish­ment rags like The Atlantic.

Pool per­son­i­fied the prob­lem. His belief in the pow­er of com­put­ers to bend people’s will and man­age soci­ety was seen as a dan­ger. He was attacked and demo­nized by the anti­war left. He was also reviled by main­stream anti-com­mu­nist lib­er­als.

A prime exam­ple: The 480, a 1964 best-sell­ing polit­i­cal thriller whose plot revolved around the dan­ger that com­put­er polling and sim­u­la­tion posed for demo­c­ra­t­ic pol­i­tics—a plot direct­ly inspired by the activ­i­ties of Ithiel de Sola Pool’s Simul­mat­ics. This new­fan­gled infor­ma­tion tech­nol­o­gy was seen a weapon of manip­u­la­tion and coer­cion, wield­ed by cyn­i­cal tech­nocrats who did not care about win­ning peo­ple over with real ideas, gen­uine states­man­ship or polit­i­cal plat­forms but sim­ply sold can­di­dates just like they would a car or a bar of soap.

***

Simul­mat­ics and its first-gen­er­a­tion imi­ta­tions are now ancient history—dating back from the long-ago time when com­put­ers took up entire rooms. But now we live in Ithiel de Sola Pool’s world. The inter­net sur­rounds us, engulf­ing and mon­i­tor­ing every­thing we do. We are tracked and watched and pro­filed every minute of every day by count­less companies—from giant plat­form monop­o­lies like Face­book and Google to bou­tique data-dri­ven elec­tion firms like i360 and Cam­bridge Ana­lyt­i­ca.

Yet the fear that Ithiel de Sola Pool and his tech­no­crat­ic world view inspired half a cen­tu­ry ago has been wiped from our cul­ture. For decades, we’ve been told that a cap­i­tal­ist soci­ety where no secrets could be kept from our benev­o­lent elite is not some­thing to fear—but some­thing to cheer and pro­mote.

Now, only after Don­ald Trump shocked the lib­er­al polit­i­cal class is this fear start­ing to resur­face. But it’s doing so in a twist­ed, nar­row way.

***

And that’s the big­ger issue with the Cam­bridge Ana­lyt­i­ca freak­out: it’s not just anti-his­tor­i­cal, it’s also pro­found­ly anti-polit­i­cal. Peo­ple are still try­ing to blame Don­ald Trump’s sur­prise 2016 elec­toral vic­to­ry on some­thing, anything—other than America’s degen­er­ate pol­i­tics and a polit­i­cal class that has presided over a stun­ning nation­al decline. The keep­ers of con­ven­tion­al wis­dom all insist in one way or anoth­er that Trump won because some­thing nov­el and unique hap­pened; that some­thing had to have gone hor­ri­bly wrong. And if you’re able to iden­ti­fy and iso­late this some­thing and get rid of it, every­thing will go back to normal—back to sta­tus quo, when every­thing was good.

Cam­bridge Ana­lyt­i­ca has been one of the less­er bogey­man [73] used to explain Trump’s vic­to­ry for quite a while [74], going back more than year. Back in March 2017, the New York Times, which now trum­pets the saga of Cam­bridge Analytica’s Face­book heist, was skep­ti­cal­ly ques­tion­ing the company’s tech­nol­o­gy and its role [7] in help­ing bring about a Trump vic­to­ry. With con­sid­er­able jus­ti­fi­ca­tion, Times reporters then chalked up the company’s over­heat­ed rhetoric to the com­pe­ti­tion for clients in a crowd­ed field of data-dri­ven elec­tion influ­ence ops.

Yet now, with Robert Meuller’s Rus­sia inves­ti­ga­tion drag­ging on and pro­duc­ing no smok­ing gun point­ing to defin­i­tive col­lu­sion, it seems that Cam­bridge Ana­lyt­i­ca has been upgrad­ed to Class A supervil­lain. Now the idea that Steve Ban­non and Robert Mer­cer con­coct­ed a secret psy­cho­log­i­cal weapon to bewitch the Amer­i­can elec­torate isn’t just a far-fetched mar­ket­ing ploy [75]—it’s a real and present dan­ger to a vir­tu­ous info-media sta­tus quo. And it’s most cer­tain­ly not the exten­sion of a lav­ish­ly fund­ed ini­tia­tive that Amer­i­can firms have been pur­su­ing for half a cen­tu­ry. No, like the Trump upris­ing it has alleged­ly mid­wifed into being, it is an oppor­tunis­tic per­ver­sion of the Amer­i­can way. Employ­ing pow­er­ful tech­nol­o­gy that rewires the inner work­ings of our body politic, Cam­bridge Ana­lyt­i­ca and its back­ers duped the Amer­i­can peo­ple into vot­ing for Trump and destroy­ing Amer­i­can democ­ra­cy.

It’s a com­fort­ing idea for our polit­i­cal elite, but it’s not true. Alexan­der Nix, Cam­bridge Analytica’s well-groomed CEO, is not a cun­ning mas­ter­mind but a gar­den-vari­ety dig­i­tal hack. Nix’s busi­ness plan is but an updat­ed ver­sion of Ithiel de Sola Pool’s vision of per­ma­nent peace and pros­per­i­ty won through a placid regime of behav­ioral­ly man­aged social con­trol. And while Nix has been sus­pend­ed fol­low­ing the blus­ter-filled video footage of his cyber-brag­ging aired on Chan­nel 4, we’re kid­ding our­selves if we think his pun­ish­ment will serve as any sort of deter­rent for the thou­sands upon thou­sands of Big Data oper­a­tors nail­ing down bil­lions in cam­paign, mil­i­tary, and cor­po­rate con­tracts to con­tin­ue mon­e­tiz­ing user data into the void. Cam­bridge Ana­lyt­i­ca is unde­ni­ably a rogue’s gallery of bad polit­i­cal actors, but to fin­ger the real cul­prits behind Don­ald Trump’s takeover Amer­i­ca, the self-appoint­ed watch­dogs of our country’s imper­iled polit­i­cal virtue had best take a long and sober­ing look in the mir­ror.

———-

“The Cam­bridge Ana­lyt­i­ca Con” by Yasha Levine; The Baf­fler; 03/21/2018 [56]

“It’s good that the main­stream news media are final­ly start­ing to pay atten­tion to this dark cor­ner of the inter­net —and pro­duc­ing exposés of shady sub rosa polit­i­cal cam­paigns and their eager exploita­tion of our online dig­i­tal trails in order to con­t­a­m­i­nate our infor­ma­tion streams and influ­ence our deci­sions. It’s about time.”

Yes indeed, it is great to see that this top­ic is final­ly get­ting the atten­tion it has long deserved. But it’s not great to see the top­ic lim­it­ed to Cam­bridge Ana­lyt­i­ca and Face­book. As Levine puts it, “We must con­cede that covert influ­ence is not some­thing unusu­al or for­eign to our soci­ety, but is as Amer­i­can as apple pie and free­dom fries.” Soci­eties in gen­er­al are held togeth­er via overt and covert influ­ence, but we’ve got­ten real­ly, real­ly good at that over the last half cen­tu­ry in Amer­i­ca and the sto­ry of Cam­bridge Ana­lyt­i­ca, and the larg­er sto­ry of Sandy Parak­i­las’s whis­tle-blow­ing about mass data col­lec­tion, can’t real­ly be under­stood out­side that his­tor­i­cal con­text:

...
But this sto­ry is being cov­ered and framed in a mis­lead­ing way. So far, much of the main­stream cov­er­age, dri­ven by the Times and Guardian reports, looks at Cam­bridge Ana­lyt­i­ca in isolation—almost entire­ly out­side of any his­tor­i­cal or polit­i­cal con­text. This makes it seem to read­ers unfa­mil­iar with the long his­to­ry of the strug­gle for con­trol of the dig­i­tal sphere as if the main prob­lem is that the bad actors at Cam­bridge Ana­lyt­i­ca crossed the trans­mis­sion wires of Face­book in the Promethean man­ner of Vic­tor Frankenstein—taking what were nor­mal­ly respectable, sci­en­tif­ic data pro­to­cols and per­vert­ing them to serve the dia­bol­i­cal aim of rean­i­mat­ing the decom­pos­ing lump of polit­i­cal flesh known as Don­ald Trump.

So if we’re going to view the actions of Cam­bridge Ana­lyt­i­ca in their prop­er light, we need first to start with an admis­sion. We must con­cede that covert influ­ence is not some­thing unusu­al or for­eign to our soci­ety, but is as Amer­i­can as apple pie and free­dom fries. The use of manip­u­la­tive, psy­cho­log­i­cal­ly dri­ven adver­tis­ing and mar­ket­ing tech­niques to sell us prod­ucts, lifestyles, and ideas has been the foun­da­tion of mod­ern Amer­i­can soci­ety [57], going back to the days of the self-styled inven­tor of pub­lic rela­tions, Edward Bernays. It oozes out of every pore on our body politic. It’s what holds our ail­ing con­sumer soci­ety togeth­er. And when it comes to mar­ket­ing can­di­dates and polit­i­cal mes­sages, using data to influ­ence peo­ple and shape their deci­sions has been the holy grail of the com­put­er age, going back half a cen­tu­ry.
...

And the first step in putting the Cam­bridge Ana­lyt­i­ca sto­ry in prop­er per­spec­tive is rec­og­niz­ing that what it is accused of doing — grab­bing per­son­al data and build­ing pro­files for the pur­pose of influ­enc­ing vot­ers — is done every day by enti­ties like Face­book and Google. It’s a reg­u­lar part of our lives. And you don’t even need to use Face­book or Google to become part of this vast com­mer­cial sur­veil­lance sys­tem. You just need to com­mu­ni­cate with some­one who does use those plat­forms:

...
Let’s start with the basics: What Cam­bridge Ana­lyt­i­ca is accused of doing—siphoning people’s data, com­pil­ing pro­files, and then deploy­ing that infor­ma­tion to influ­ence them to vote a cer­tain way—Facebook and Sil­i­con Val­ley giants like Google do every day, indeed, every minute we’re logged on, on a far greater and more inva­sive scale.

Today’s inter­net busi­ness ecosys­tem is built on for-prof­it sur­veil­lance, behav­ioral pro­fil­ing, manip­u­la­tion and influ­ence. That’s the name of the game. It isn’t just Face­book or Cam­bridge Ana­lyt­i­ca or even Google. It’s Ama­zon. It’s eBay. It’s Palan­tir. It’s Angry Birds. It’s MoviePass [58]. It’s Lock­heed Mar­tin [59]. It’s every app you’ve ever down­loaded. Every phone you bought. Every pro­gram you watched on your on-demand cable TV pack­age.

All of these games, apps, and plat­forms prof­it from the con­cert­ed siphon­ing up of all data trails to pro­duce pro­files for all sorts of micro-tar­get­ed influ­ence ops in the pri­vate sec­tor. This com­merce in user data per­mit­ted Face­book to earn $40 bil­lion last year, while Google raked in $110 bil­lion.

What do these com­pa­nies know about us, their users? Well, just about every­thing.

Sil­i­con Val­ley of course keeps a tight lid on this infor­ma­tion, but you can get a glimpse of the kinds of data our pri­vate dig­i­tal dossiers con­tain by trawl­ing through their patents. Take, for instance, a series of patents Google filed in the mid-2000s for its Gmail-tar­get­ed adver­tis­ing tech­nol­o­gy. The lan­guage, stripped of opaque tech jar­gon, revealed that just about every­thing we enter into Google’s many prod­ucts and platforms—from email cor­re­spon­dence to Web search­es and inter­net browsing—is ana­lyzed and used to pro­file users in an extreme­ly inva­sive and per­son­al way. Email cor­re­spon­dence is parsed for mean­ing and sub­ject mat­ter. Names are matched to real iden­ti­ties and address­es. Email attachments—say, bank state­ments or test­ing results from a med­ical lab—are scraped for infor­ma­tion. Demo­graph­ic and psy­cho­graph­ic data, includ­ing social class, per­son­al­i­ty type, age, sex, polit­i­cal affil­i­a­tion, cul­tur­al inter­ests, social ties, per­son­al income, and mar­i­tal sta­tus is extract­ed. In one patent, I dis­cov­ered that Google appar­ent­ly had the abil­i­ty to deter­mine if a per­son was a legal U.S. res­i­dent or not. It also turned out you didn’t have to be a reg­is­tered Google user to be snared in this pro­fil­ing appa­ra­tus. All you had to do was com­mu­ni­cate with some­one who had a Gmail address.

On the whole, Google’s pro­fil­ing phi­los­o­phy was no dif­fer­ent than Facebook’s, which also con­structs “shad­ow pro­files” to col­lect and mon­e­tize data, even if you nev­er had a reg­is­tered Face­book or Gmail account.
...

The next step in con­tex­tu­al­iz­ing this is rec­og­niz­ing that Face­book and Google are mere­ly the biggest fish in an ocean of data bro­ker­age mar­kets that has many small­er inhab­i­tants try­ing to do the same thing. This is part of what makes Face­book’s hand­ing over of pro­file data to app devel­op­ers so scandalous...Facebook clear­ly new there was a vora­cious mar­ket for this infor­ma­tion and made a lot of mon­ey sell­ing into that mar­ket:

...
It’s not just the big plat­form monop­o­lies that do this, but all the small­er com­pa­nies that run their busi­ness­es on ser­vices oper­at­ed by Google and Face­book. It even includes cute games [60] like Angry Birds, devel­oped by Finland’s Rovio Enter­tain­ment, that’s been down­loaded more than a bil­lion times. The Android ver­sion of Angry Birds was found to pull per­son­al data [61] on its play­ers, includ­ing eth­nic­i­ty, mar­i­tal sta­tus, and sex­u­al orientation—including options for the “sin­gle,” “mar­ried,” “divorced,” “engaged,” and “swinger” cat­e­gories. Pulling per­son­al data like this didn’t con­tra­dict Google’s terms of ser­vices for its Android plat­form. Indeed, for-prof­it sur­veil­lance was the whole point of why Google start­ed plan­ning to launch an iPhone rival as far back as 2004.

In launch­ing Android, Google made a gam­ble [60] that by releas­ing its pro­pri­etary oper­at­ing sys­tem to man­u­fac­tur­ers free of charge, it wouldn’t be rel­e­gat­ed to run­ning apps on Apple iPhone or Microsoft Mobile Win­dows like some kind of dig­i­tal sec­ond-class cit­i­zen. If it played its cards right and Android suc­ceed­ed, Google would be able to con­trol the envi­ron­ment that under­pins the entire mobile expe­ri­ence, mak­ing it the ulti­mate gate­keep­er of the many mon­e­tized inter­ac­tions among users, apps, and adver­tis­ers. And that’s exact­ly what hap­pened. Today, Google monop­o­lizes the smart phone mar­ket and dom­i­nates the mobile for-prof­it sur­veil­lance busi­ness [62].

These detailed psy­cho­log­i­cal pro­files, togeth­er with the direct access to users that plat­forms like Google and Face­book deliv­er, make both com­pa­nies cat­nip to adver­tis­ers, PR flacks—and dark-mon­ey polit­i­cal out­fits like Cam­bridge Ana­lyt­i­ca.
...

And when it comes to polit­i­cal cam­paigns, the dig­i­tal giants like Face­book and Google already have spe­cial elec­tion units set up to give priv­i­leged access to polit­i­cal cam­paigns so they can influ­ence vot­ers even more effec­tive­ly. The sto­ries about the Trump cam­paign’s use of Face­book “embeds” to run a mas­sive sys­tem­at­ic adver­tis­ing cam­paign of “A/B test­ing on steroids” to sys­tem­at­i­cal­ly exper­i­ment on vot­er ad respons­es [21] is part of that larg­er sto­ry of how these giants have already made the manip­u­la­tion of vot­ers big busi­ness:

...
Indeed, polit­i­cal cam­paigns showed an ear­ly and pro­nounced affin­i­ty for the idea of tar­get­ed access and influ­ence on plat­forms like Face­book. Instead of blan­ket­ing air­waves with a sin­gle polit­i­cal ad, they could show peo­ple ads that appealed specif­i­cal­ly to the issues they held dear. They could also ensure that any such mes­sage spread through a tar­get­ed person’s larg­er social net­work through repost­ing and shar­ing.

The enor­mous com­mer­cial inter­est that polit­i­cal cam­paigns have shown in social media has earned them priv­i­leged atten­tion from Sil­i­con Val­ley plat­forms in return. Face­book runs a sep­a­rate polit­i­cal divi­sion specif­i­cal­ly geared to help its cus­tomers tar­get and influ­ence vot­ers.

The com­pa­ny even allows polit­i­cal cam­paigns to upload their own lists of poten­tial vot­ers and sup­port­ers direct­ly into Facebook’s data sys­tem. So armed, dig­i­tal polit­i­cal oper­a­tives can then use those people’s social net­works to iden­ti­fy oth­er prospec­tive vot­ers who might be sup­port­ive of their candidate—and then tar­get them with a whole new tidal wave of ads. “There’s a lev­el of pre­ci­sion that doesn’t exist in any oth­er medi­um,” Crys­tal Pat­ter­son, a Face­book employ­ee who works with gov­ern­ment and pol­i­tics cus­tomers, told the New York Times back in 2015. “It’s get­ting the right mes­sage to the right peo­ple at the right time.”

Nat­u­ral­ly, a whole slew of com­pa­nies and oper­a­tives in our increas­ing­ly data-dri­ven elec­tion scene have cropped up over the last decade to plug in to these amaz­ing influ­ence machines. There is a whole con­stel­la­tion of them work­ing all sorts of strate­gies: tra­di­tion­al vot­er tar­get­ing, polit­i­cal pro­pa­gan­da mills, troll armies [63], and bots [64].

Some of these firms are polit­i­cal­ly agnos­tic; they’ll work for any­one with cash. Oth­ers are par­ti­san. The Demo­c­ra­t­ic Par­ty Data Death Star [65] is NGP VAN [66]. The Repub­li­cans have a few of their own—including i360, a data mon­ster gen­er­ous­ly fund­ed by Charles Koch. Nat­u­ral­ly, i360 part­ners with Face­book to deliv­er tar­get vot­ers. It also claims to have 700 per­son­al data points cross-tab­u­lat­ed on 199 mil­lion vot­ers and near­ly 300 mil­lion con­sumers, with the abil­i­ty to pro­file and tar­get them with pin-point accu­ra­cy based on their beliefs and views.

Here’s how The Nation­al Jour­nal’s Andrew Rice described i360 in 2015:

Like Google, the Nation­al Secu­ri­ty Agency, or the Demo­c­ra­t­ic data machine, i360 has a vora­cious appetite for per­son­al infor­ma­tion. It is con­stant­ly ingest­ing new data into its tar­get­ing sys­tems, which pre­dict not only par­ti­san iden­ti­fi­ca­tion but also sen­ti­ments about issues such as abor­tion, tax­es, and health care. When I vis­it­ed the i360 office, an employ­ee gave me a demon­stra­tion, zoom­ing in on a map to focus on a par­tic­u­lar 66-year-old high school teacher who lives in an apart­ment com­plex in Alexan­dria, Vir­ginia. . . . Though the adver­tis­ing indus­try typ­i­cal­ly eschews address­ing any sin­gle individual—it’s not just inva­sive, it’s also inefficient—it is becom­ing com­mon­place to tar­get extreme­ly nar­row audi­ences. So the school­teacher, along with a few look-alikes, might see a tai­lored ad the next time she clicks on YouTube.

Sil­i­con Val­ley doesn’t just offer cam­paigns a neu­tral plat­form; it also works close­ly along­side polit­i­cal can­di­dates to the point that the biggest inter­net com­pa­nies have become an exten­sion of the Amer­i­can polit­i­cal sys­tem. As one recent study showed, tech com­pa­nies rou­tine­ly embed their employ­ees inside major polit­i­cal cam­paigns: “Face­book, Twit­ter, and Google go beyond pro­mot­ing their ser­vices and facil­i­tat­ing dig­i­tal adver­tis­ing buys, active­ly shap­ing cam­paign com­mu­ni­ca­tion through their close col­lab­o­ra­tion with polit­i­cal staffers . . . these firms serve as qua­si-dig­i­tal con­sul­tants to cam­paigns, shap­ing dig­i­tal strat­e­gy, con­tent, and exe­cu­tion.”
...

And offer­ing spe­cial ser­vices to cam­paign manip­u­late vot­ers isn’t just big busi­ness. It’s a large­ly unreg­u­lat­ed busi­ness. If Face­book decides to covert­ly manip­u­late you by alter­ing its news­feed algo­rithms so it shows you news arti­cles more from your con­ser­v­a­tive-lean­ing friends (or lib­er­al-lean­ing friends), that’s total­ly legal. Because, again, sub­tly manip­u­lat­ing peo­ple is as Amer­i­can as apple pie:

...
Now, of course, every elec­tion is a Face­book Elec­tion. And why not? As Bloomberg News has not­ed [67], Sil­i­con Val­ley ranks elec­tions “along­side the Super Bowl and the Olympics in terms of events that draw block­buster ad dol­lars and boost engage­ment.” In 2016, $1 bil­lion was spent on dig­i­tal advertising—with the bulk going to Face­book, Twit­ter, and Google.

What’s inter­est­ing here is that because so much mon­ey is at stake, there are absolute­ly no rules that would restrict any­thing an unsa­vory polit­i­cal appa­ratchik or a Sil­i­con Val­ley oli­garch might want to foist on the unsus­pect­ing dig­i­tal pub­lic. Creep­i­ly, Facebook’s own inter­nal research divi­sion car­ried out exper­i­ments show­ing that the plat­form could influ­ence people’s emo­tion­al state in con­nec­tion to a cer­tain top­ic or event. Com­pa­ny engi­neers call this fea­ture “emo­tion­al con­ta­gion [68]”—i.e., the abil­i­ty to viral­ly influ­ence people’s emo­tions and ideas just through the con­tent of sta­tus updates. In the twist­ed econ­o­my of emo­tion­al con­ta­gion, a neg­a­tive post by a user sup­press­es pos­i­tive posts by their friends, while a pos­i­tive post sup­press­es neg­a­tive posts. “When a Face­book user posts, the words they choose influ­ence the words cho­sen lat­er by their friends,” explained [69] the company’s lead sci­en­tist on this study.

On a very basic lev­el, Facebook’s opaque con­trol of its feed algo­rithm means the plat­form has real pow­er over people’s ideas and actions dur­ing an elec­tion. This can be done by a data shift as sim­ple and sub­tle as imper­cep­ti­bly tweak­ing a person’s feed to show more posts from friends who are, say, sup­port­ers of a par­tic­u­lar polit­i­cal can­di­date or a spe­cif­ic polit­i­cal idea or event. As far as I know, there is no law pre­vent­ing Face­book from doing just that: it’s plain­ly able and will­ing to influ­ence a user’s feed based on polit­i­cal aims—whether done for inter­nal cor­po­rate objec­tives, or due to pay­ments from polit­i­cal groups, or by the per­son­al pref­er­ences of Mark Zucker­berg.
...

And this con­tem­po­rary state of affairs did­n’t emerge spon­ta­neous­ly. As Levine cov­ers in Sur­veil­lance Val­ley, this is what the inter­net — back when it was the ARPANET mil­i­tary net­work — was all about from its very con­cep­tion:

...
There’s anoth­er, big­ger cul­tur­al issue with the way we’ve begun to exam­ine and dis­cuss Cam­bridge Analytica’s bat­tery of inter­net-based influ­ence ops. Peo­ple are still daz­zled by the idea that the inter­net, in its pure, untaint­ed form, is some kind of mag­ic machine dis­trib­ut­ing democ­ra­cy and egal­i­tar­i­an­ism across the globe with the touch of a few key­strokes. This is the gospel preached by a stal­wart cho­rus of Net prophets, from Jeff Jarvis and the late John Per­ry Bar­low to Clay Shirky and Kevin Kel­ly. These char­la­tans all feed on an hon­or­able demo­c­ra­t­ic impulse: peo­ple still want to des­per­ate­ly believe in the utopi­an promise of this technology—its abil­i­ty to equal­ize pow­er, end cor­rup­tion, top­ple cor­po­rate media monop­o­lies, and empow­er the indi­vid­ual.

This mythology—which is of course aggres­sive­ly con­fect­ed for mass con­sump­tion by Sil­i­con Val­ley mar­ket­ing and PR outfits—is deeply root­ed in our cul­ture; it helps explain why oth­er­wise seri­ous jour­nal­ists work­ing for main­stream news out­lets can uniron­i­cal­ly employ phras­es such as “infor­ma­tion wants to be free” and “Facebook’s engine of democ­ra­cy” and get away with it.

The truth is that the inter­net has nev­er been about egal­i­tar­i­an­ism or democ­ra­cy.

The ear­ly inter­net came out of a series of Viet­nam War coun­terin­sur­gency projects aimed at devel­op­ing com­put­er tech­nol­o­gy that would give the gov­ern­ment a way to man­age a com­plex series of glob­al com­mit­ments and to mon­i­tor and pre­vent polit­i­cal strife—both at home and abroad. The inter­net, going back to its first incar­na­tion as the ARPANET mil­i­tary net­work, was always about sur­veil­lance, pro­fil­ing, and tar­get­ing [70].

The influ­ence of U.S. coun­terin­sur­gency doc­trine on the devel­op­ment of mod­ern com­put­ers and the inter­net is not some­thing that many peo­ple know about. But it is a sub­ject that I explore at length in my book, Sur­veil­lance Val­ley. So what jumps out at me is how seam­less­ly the report­ed activ­i­ties of Cam­bridge Ana­lyt­i­ca fit into this his­tor­i­cal nar­ra­tive.
...

“The ear­ly inter­net came out of a series of Viet­nam War coun­terin­sur­gency projects aimed at devel­op­ing com­put­er tech­nol­o­gy that would give the gov­ern­ment a way to man­age a com­plex series of glob­al com­mit­ments and to mon­i­tor and pre­vent polit­i­cal strife—both at home and abroad. The inter­net, going back to its first incar­na­tion as the ARPANET mil­i­tary net­work, was always about sur­veil­lance, pro­fil­ing, and tar­get­ing [70]

And one of the key fig­ures behind this ear­ly ARPANET ver­sion of the inter­net, Ithiel de Sola Pool, got his start in this area in the 1950’s work­ing at the Hoover Insti­tu­tion at Stan­ford Uni­ver­si­ty to under­stand the nature and caus­es of left-wing rev­o­lu­tions and dis­till this down to a math­e­mat­i­cal for­mu­la. Pool, an vir­u­lent anti-Com­mu­nist, also worked for JFK’s 1960 cam­paign and went on to start a pri­vate com­pa­ny, Simul­mat­ics, offer­ing ser­vices in mod­el­ing and manip­u­lat­ing human behav­ior based on large data sets on peo­ple:

...
Cam­bridge Ana­lyt­i­ca is a sub­sidiary of the SCL Group, a mil­i­tary con­trac­tor set up by a spooky huck­ster named Nigel Oakes that sells itself as a high-pow­ered con­clave of experts spe­cial­iz­ing in data-dri­ven coun­terin­sur­gency. It’s done work for the Pen­ta­gon, NATO, and the UK Min­istry of Defense in places like Afghanistan and Nepal [71], where it says it ran a “cam­paign to reduce and ulti­mate­ly stop the large num­bers of Maoist insur­gents in Nepal from break­ing into hous­es in remote areas to steal food, harass the home­own­ers and cause dis­rup­tion.”

In the grander scheme of high-tech coun­terin­sur­gency boon­dog­gles, which fea­tures such sto­ried psy-ops out­fits as Peter Thiel’s Palan­tir and Cold War dinosaurs like Lock­heed Mar­tin, the SCL Group appears to be a com­par­a­tive­ly minor play­er. Nev­er­the­less, its ambi­tious claims to recon­fig­ure the world order with some well-placed algo­rithms recalls one of the first major play­ers in the field: Simul­mat­ics, a 1960s coun­terin­sur­gency mil­i­tary con­trac­tor that pio­neered data-dri­ven elec­tion cam­paigns and whose founder, Ithiel de Sola Pool, helped shape the devel­op­ment of the ear­ly inter­net as a sur­veil­lance and coun­terin­sur­gency tech­nol­o­gy.

Ithiel de Sola Pool descend­ed from a promi­nent rab­bini­cal fam­i­ly that traced its roots to medieval Spain. Vir­u­lent­ly anti­com­mu­nist and tech-obsessed, he got his start in polit­i­cal work in 1950s work­ing on project [72] at the Hoover Insti­tu­tion at Stan­ford Uni­ver­si­ty that sought to under­stand the nature and caus­es of left-wing rev­o­lu­tions and reduce their like­ly course down to a math­e­mat­i­cal for­mu­la.

He then moved to MIT and made a name for him­self help­ing cal­i­brate the mes­sag­ing of John F. Kennedy’s 1960 pres­i­den­tial cam­paign. His idea was to mod­el the Amer­i­can elec­torate by decon­struct­ing each vot­er into 480 data points that defined every­thing from their reli­gious views to racial atti­tudes to socio-eco­nom­ic sta­tus. He would then use that data to run sim­u­la­tions on how they would respond to a par­tic­u­lar message—and those tri­al runs would per­mit major cam­paigns to fine-tune their mes­sages accord­ing­ly.

These new tar­get­ed mes­sag­ing tac­tics, enabled by rudi­men­ta­ry com­put­ers, had many fans in the per­ma­nent polit­i­cal class of Wash­ing­ton; their liveli­hoods, after all, were large­ly root­ed in their claims to ana­lyze and pre­dict polit­i­cal behav­ior. And so Pool lever­aged his research to launch Simul­mat­ics, a data ana­lyt­ics start­up that offered com­put­er sim­u­la­tion ser­vices to major Amer­i­can cor­po­ra­tions, help­ing them pre-test prod­ucts and con­struct adver­tis­ing cam­paigns.

Simul­mat­ics also did a brisk busi­ness as a mil­i­tary and intel­li­gence con­trac­tor. It ran sim­u­la­tions for Radio Lib­er­ty, the CIA’s covert anti-com­mu­nist radio sta­tion, help­ing the agency mod­el the Sovi­et Union’s inter­nal com­mu­ni­ca­tion sys­tem in order to pre­dict the effect that for­eign news broad­casts would have on the country’s polit­i­cal sys­tem. At the same time, Simul­mat­ics ana­lysts were doing coun­terin­sur­gency work under an ARPA con­tract in Viet­nam, con­duct­ing inter­views and gath­er­ing data to help mil­i­tary plan­ners under­stand why Viet­namese peas­ants rebelled and resist­ed Amer­i­can paci­fi­ca­tion efforts. Simulmatic’s work in Viet­nam was just one piece of a bru­tal Amer­i­can coun­terin­sur­gency pol­i­cy that involved covert pro­grams of assas­si­na­tions, ter­ror, and tor­ture that col­lec­tive­ly came to be known as the Phoenix Pro­gram.
...

And part of what drove Pool’s was a utopi­an belief that com­put­ers and mas­sive amounts of data could be used to run soci­ety har­mo­nious­ly. Left-wing rev­o­lu­tions were prob­lems to be man­aged with Big Data. It’s a pret­ty impor­tant his­tor­i­cal con­text when think­ing about the role Cam­bridge Ana­lyt­i­ca played in elect­ing Don­ald Trump:

...
At the same time, Pool was also per­son­al­ly involved in an ear­ly ARPANET-con­nect­ed ver­sion of Thiel’s Palan­tir effort—a pio­neer­ing sys­tem that would allow mil­i­tary plan­ners and intel­li­gence to ingest and work with large and com­plex data sets. Pool’s pio­neer­ing work won him a devot­ed fol­low­ing among a group of tech­nocrats who shared a utopi­an belief in the pow­er of com­put­er sys­tems to run soci­ety from the top down in a har­mo­nious man­ner. They saw the left-wing upheavals of the 1960s not as a polit­i­cal or ide­o­log­i­cal prob­lem but as a chal­lenge of man­age­ment and engi­neer­ing. Pool fed these rever­ies by set­ting out to build com­put­er­ized sys­tems that could mon­i­tor the world in real time and ren­der people’s lives trans­par­ent. He saw these sur­veil­lance and man­age­ment regimes in utopi­an terms—as a vital tool to man­age away social strife and con­flict. “Secre­cy in the modem world is gen­er­al­ly a desta­bi­liz­ing fac­tor,” he wrote in a 1969 essay. “Noth­ing con­tributes more to peace and sta­bil­i­ty than those activ­i­ties of elec­tron­ic and pho­to­graph­ic eaves­drop­ping, of con­tent analy­sis and tex­tu­al inter­pre­ta­tion.”
...

And guess what: the Amer­i­can pub­lic was­n’t enam­ored with Pool’s vision of a world man­aged by com­put­ing tech­nol­o­gy and Big Data mod­els of soci­ety. When the pub­lic learned about these ear­ly ver­sion of the inter­net inspired by visions of a com­put­er-man­aged world in the 60’s and 70’s, the pub­lic got scared:

...
With the advent of cheap­er com­put­er tech­nol­o­gy in the 1960s, cor­po­rate and gov­ern­ment data­bas­es were already mak­ing a good deal of Pool’s prophe­cy come to pass, via sophis­ti­cat­ed new modes of con­sumer track­ing and pre­dic­tive mod­el­ing. But rather than greet­ing such advances as the augurs of a new demo­c­ra­t­ic mir­a­cle, peo­ple at the time saw it as a threat. Crit­ics across the polit­i­cal spec­trum warned that the pro­lif­er­a­tion of these tech­nolo­gies would lead to cor­po­ra­tions and gov­ern­ments con­spir­ing to sur­veil, manip­u­late, and con­trol soci­ety.

This fear res­onat­ed with every part of the culture—from the new left to prag­mat­ic cen­trists and reac­tionary South­ern Democ­rats. It prompt­ed some high-pro­file exposés in papers like the New York Times and Wash­ing­ton Post. It was report­ed on in trade mag­a­zines of the nascent com­put­er indus­try like Com­put­er­World. And it com­mand­ed prime real estate in estab­lish­ment rags like The Atlantic.

Pool per­son­i­fied the prob­lem. His belief in the pow­er of com­put­ers to bend people’s will and man­age soci­ety was seen as a dan­ger. He was attacked and demo­nized by the anti­war left. He was also reviled by main­stream anti-com­mu­nist lib­er­als.

A prime exam­ple: The 480, a 1964 best-sell­ing polit­i­cal thriller whose plot revolved around the dan­ger that com­put­er polling and sim­u­la­tion posed for demo­c­ra­t­ic pol­i­tics—a plot direct­ly inspired by the activ­i­ties of Ithiel de Sola Pool’s Simul­mat­ics. This new­fan­gled infor­ma­tion tech­nol­o­gy was seen a weapon of manip­u­la­tion and coer­cion, wield­ed by cyn­i­cal tech­nocrats who did not care about win­ning peo­ple over with real ideas, gen­uine states­man­ship or polit­i­cal plat­forms but sim­ply sold can­di­dates just like they would a car or a bar of soap.
...

But that fear some­how dis­ap­peared in sub­se­quent decades, only to be replaced with a faith in our benev­o­lent tech­no-elite. And a faith that this mass public/private sur­veil­lance sys­tem is actu­al­ly an empow­er­ing tool that will lead to a lim­it­less future. And that is per­haps the biggest scan­dal here: The pub­lic did­n’t just for­got to keep an eye on the pow­er­ful. The pub­lic for­got to keep an eye on the peo­ple whose pow­er is derived from keep­ing an eye on the pub­lic. We built a sur­veil­lance state at the same time we fell into a fog of civic and his­tor­i­cal amne­sia. And that has coin­cid­ed with the rise of a plu­toc­ra­cy, the dom­i­nance of right-wing anti-gov­ern­ment eco­nom­ic doc­trines, and the larg­er fail­ure of the Amer­i­can polit­i­cal and eco­nom­ic elites to deliv­er a soci­ety that actu­al­ly works for aver­age peo­ple. To put it anoth­er way, the rise of the mod­ern sur­veil­lance state is one ele­ment of a mas­sive, decades-long process of col­lec­tive­ly ‘drop­ping the ball’. We screwed up mas­sive­ly and Face­book and Google are just one of the con­se­quences of this. And yet we still don’t view the Trump phe­nom­e­na with­in the con­text of that mas­sive col­lec­tive screw up, which means we’re still screw­ing up mas­sive­ly:

...
Yet the fear that Ithiel de Sola Pool and his tech­no­crat­ic world view inspired half a cen­tu­ry ago has been wiped from our cul­ture. For decades, we’ve been told that a cap­i­tal­ist soci­ety where no secrets could be kept from our benev­o­lent elite is not some­thing to fear—but some­thing to cheer and pro­mote.

Now, only after Don­ald Trump shocked the lib­er­al polit­i­cal class is this fear start­ing to resur­face. But it’s doing so in a twist­ed, nar­row way.

***

And that’s the big­ger issue with the Cam­bridge Ana­lyt­i­ca freak­out: it’s not just anti-his­tor­i­cal, it’s also pro­found­ly anti-polit­i­cal. Peo­ple are still try­ing to blame Don­ald Trump’s sur­prise 2016 elec­toral vic­to­ry on some­thing, anything—other than America’s degen­er­ate pol­i­tics and a polit­i­cal class that has presided over a stun­ning nation­al decline. The keep­ers of con­ven­tion­al wis­dom all insist in one way or anoth­er that Trump won because some­thing nov­el and unique hap­pened; that some­thing had to have gone hor­ri­bly wrong. And if you’re able to iden­ti­fy and iso­late this some­thing and get rid of it, every­thing will go back to normal—back to sta­tus quo, when every­thing was good.
...

So the biggest sto­ry here isn’t that Cam­bridge Ana­lyt­i­ca was engaged in mass manip­u­la­tion cam­paign. And the biggest sto­ry isn’t even that Cam­bridge Ana­lyt­i­ca was engaged in a cut­ting-edge com­mer­cial mass manip­u­la­tion cam­paign. Because both of those sto­ries are eclipsed by the sto­ry that even if Cam­bridge Ana­lyt­i­ca real­ly was engaged in a com­mer­cial cut­ting edge cam­paign, it prob­a­bly was­n’t near­ly as cut­ting edge as what Face­book and Google and the oth­er data giants rou­tine­ly engage in. And this sit­u­a­tion has been build­ing for decades and with­in the con­text of the much larg­er scan­dal of the rise of a oli­garchy that more or less runs Amer­i­ca by and for pow­er­ful inter­ests. Pow­er­ful inter­ests that are over­whelm­ing­ly ded­i­cat­ed to right-wing elit­ist doc­trines that view the pub­lic as a resources to be con­trolled and exploit­ed for pri­vate prof­it.

It’s all a reminder that, like so many incred­i­bly com­plex issues, cre­at­ing very high qual­i­ty gov­ern­ment is the only fea­si­ble answer. A high qual­i­ty gov­ern­ment man­aged by a self-aware pub­lic. Some sort of ‘sur­veil­lance state’ is almost an inevitabil­i­ty as long as we have ubiq­ui­tous sur­veil­lance tech­nol­o­gy. Even the array of ‘cryp­to’ tools tout­ed in recent years have con­sis­tent­ly proven to be vul­ner­a­ble, which isn’t nec­es­sar­i­ly a bad thing since ubiq­ui­tous cryp­to-tech­nol­o­gy comes with its own suite of mega-col­lec­tive headaches [76]. Nation­al secu­ri­ty and per­son­al data inse­cu­ri­ty real­ly are inter­twined in both mutu­al­ly inclu­sive and exclu­sive ways. It’s not as if the nation­al secu­ri­ty hawk argu­ments that “you can’t be free if you’re dead from [insert war, ter­ror, ran­dom chaos things a nation­al secu­ri­ty state is sup­posed to deal with]” isn’t valid. But fears of Big Broth­er are also valid, as our present sit­u­a­tion amply demon­strates. The path isn’t clear, which is why a nation­al secu­ri­ty state with a sig­nif­i­cant pri­vate sec­tor com­po­nent and access to ample inti­mate details is like­ly for the fore­see­able future whether you like it or not. Peo­ple err on imme­di­ate safe­ty. So we bet­ter have very high qual­i­ty gov­ern­ment. Espe­cial­ly high qual­i­ty reg­u­la­tions for the pri­vate sec­tor com­po­nents of that nation­al secu­ri­ty state.

And while dig­i­tal giants like Google and Face­book will inevitably have access to a troves of per­son­al data that they need to offer the kinds of ser­vices peo­ple need, there’s no rea­son any sort of reg­u­lat­ing them heav­i­ly so they don’t become per­son­al data repos­i­to­ry for sale. Which is what they are now.

What do we do about ser­vices that peo­ple use to run their lives which, by def­i­n­i­tion, neces­si­tate the col­lec­tion of pri­vate data by a third-par­ty? How do we deal with these chal­lenges? Well, again, it starts with being aware of them and actu­al­ly try­ing to col­lec­tive­ly grap­ple with them so some sort of gen­er­al con­sen­sus can be arrive at. And that’s all why we need to rec­og­nize that it is imper­a­tive that the pub­lic sur­veils the sur­veil­lance state along with sur­veilling the rest of the world going on around us too. A self-aware sur­veil­lance state com­prised of a self-aware pop­u­lace of peo­ple who know what’s going on with their sur­veil­lance state and the world. In oth­er words, part of the solu­tion to ‘Big Data Big Broth­er’ real­ly is a soci­ety of ‘Lit­tle Broth­ers and Sis­ters’ who are col­lec­tive­ly very informed about what is going on in the world and polit­i­cal­ly capa­ble of effect­ing changes to that sur­veil­lance state — and the rest of gov­ern­ment or the pri­vate sec­tor — when nec­es­sary change is iden­ti­fied. In oth­er oth­er words, the one ‘utopi­an’ solu­tion we can’t afford to give up on is the utopia of a well-func­tion democ­ra­cy pop­u­lat­ed by a well-informed cit­i­zen­ry. A well-armed cit­i­zen­ry armed with rel­e­vant facts and wis­dom (and an exten­sive under­stand­ing of the his­to­ry and tech­nique of fas­cism and oth­er author­i­tar­i­an move­ments). Because a clue­less soci­ety will be an abu­sive­ly sur­veilled soci­ety.

But the fact that this Cam­bridge Ana­lyt­i­ca scan­dal is a sur­prise and is being cov­ered large­ly in iso­la­tion of this broad­er his­toric and con­tem­po­rary con­text is a reminder that we are no where near that demo­c­ra­t­ic ide­al of a well-informed cit­i­zen­ry. Well, guess what would be a real­ly valu­able tool for sur­veilling the sur­veil­lance state and the rest of the world around us and becom­ing that well-informed cit­i­zen­ry: the inter­net! Specif­i­cal­ly, we real­ly do need to read and digest grow­ing amounts of infor­ma­tion to make sense of an increas­ing­ly com­plex world. But the inter­net is just the start. The goal needs to be the kind of func­tion­al, self-aware democ­ra­cy were sit­u­a­tions like the cur­rent one don’t devel­op in a fog of col­lec­tive amne­sia and can be pro-active­ly man­aged. To put it anoth­er way, we need an inverse of Ithiel de Sola Pool’s vision of world with benev­o­lent elites use com­put­ers and Big Data to man­age the rab­ble and ward of polit­i­cal rev­o­lu­tions. Instead, we need a polit­i­cal rev­o­lu­tion of the rab­ble fueled by the knowl­edge of our his­to­ry and world the inter­net makes wide­ly acces­si­ble. And one of the key goals of the polit­i­cal rev­o­lu­tion needs to be to cre­ate a world with the knowl­edge the inter­net makes wide­ly avail­able is used to reign in our elites and build a world that works for every­one.

And yes, that implic­it­ly implies a left-wing rev­o­lu­tion since left-wing demo­c­ra­t­ic move­ments those are the only kind that have every­one in mind. And yes, this implies an eco­nom­ic rev­o­lu­tion that sys­tem­at­i­cal­ly frees up time for vir­tu­al­ly every­one one so peo­ple actu­al­ly have the time to inform them­selves. Eco­nom­ic secu­ri­ty and time secu­ri­ty. We need to build a world that pro­vide both to every­one.

So when we ask our­selves how we should respond to the grow­ing Cam­bridge Analytica/Facebook scan­dal, don’t for­get that one of the key lessons that the sto­ry of Cam­bridge Ana­lyt­i­ca teach­es us is that there is an immense amount of knowl­edge about our­selves — our his­to­ry and con­tem­po­rary con­text- that we need­ed to learn and did­n’t. And that includes envi­sion­ing what a func­tion­al demo­c­ra­t­ic soci­ety and econ­o­my that works for every­one would look like and build­ing it. Yes, the inter­net could be very help­ful in that process, just don’t for­get about every­thing else that will be required to build that func­tion­al democ­ra­cy [77].