The im­pact of Face­book and the tech gi­ants

The Monthly (Australia) - - THE MEDICINE - BY NICK FEIK

“To­day I want to fo­cus on the most im­por­tant ques­tion of all,” wrote Face­book CEO Mark Zucker­berg. “Are we build­ing the world we all want?” The “so­cial in­fra­struc­ture” built by the com­pany Zucker­berg founded is now reg­u­larly used by al­most two bil­lion peo­ple. His ‘Build­ing Global Com­mu­nity’ es­say, which he posted on Face­book in Fe­bru­ary, is an ode to the virtues of con­nec­tiv­ity. Join­ing up the world and em­pow­er­ing “us”, Face­book is con­nect­ing peo­ple more reg­u­larly and in­tri­cately than any­thing that’s ever come be­fore, and Zucker­berg in­tends it to be­come syn­ony­mous with hu­man progress.

“Progress now re­quires hu­man­ity com­ing to­gether not just as cities or na­tions, but also as a global com­mu­nity.” Rea­sons for op­ti­mism abound. “We had a good start to 2017,” Zucker­berg said in May, on the re­lease of Face­book’s lat­est fi­nan­cial fig­ures. To­tal rev­enue had soared by 49% in the past year and prof­its topped $1 bil­lion per month.

In April, Mel­bourne school­girl Ariella, 16, joined a Face­book group “16+ han­gouts” to chat with other teens. As Rachel Baxendale re­ported in the Aus­tralian, when an­other mem­ber re­alised Ariella was Jewish, he and more than a dozen other teenagers be­gan abus­ing her. “All aboard Jew ex­press next stop Auschwitz gassing chambers, I hear there is a lovely shower aboard, Ex­ter­mi­nate, Ex­ter­mi­nate,” wrote one. “I’ll make u proud,” wrote an­other. “I’ll f*** her in the gas chambers.”

By the time Ariella left the group 24 hours af­ter join­ing, she had com­piled al­most 50 pages of screen­shots of abu­sive mes­sages. She re­ported the abuse to Face­book.

“Thanks for your re­port,” Face­book replied. “You did the right thing by let­ting us know about this. We looked over the pro­file you re­ported, and though it doesn’t go against one of our spe­cific com­mu­nity stan­dards, we un­der­stand that the pro­file may still be of­fen­sive to you.” Face­book’s State­ment of Rights and Re­spon­si­bil­i­ties pro­hibits “hate speech”, and it has pre­vi­ously said that “while there is no uni­ver­sally ac­cepted def­i­ni­tion of hate speech, as a plat­form we de­fine the term to mean di­rect and se­ri­ous at­tacks on any pro­tected cat­e­gory of peo­ple based on their race, eth­nic­ity, na­tional ori­gin, re­li­gion, sex, gen­der, sex­ual ori­en­ta­tion, dis­abil­ity or dis­ease”.

Not that any of that was help­ful to Ariella. Was this the world Zucker­berg talked about build­ing?

In May, on the same day Face­book an­nounced its firstquar­ter earn­ings, Fair­fax Me­dia em­ploy­ees across Aus­tralia went on strike af­ter the com­pany de­cided to cut 125 editorial staff in a bid to save $30 million.

The staff cut was greeted with great pub­lic anger and frus­tra­tion. Most of it was di­rected at Fair­fax man­age­ment, par­tic­u­larly CEO Greg Hy­wood, who’d landed a $2.5 million share bonus a cou­ple of months ear­lier. (Some re­ports sug­gested he may have earned as much as $7.2 million in 2016.) Fair­fax’s man­age­ment has not ex­celled over the past decade, and Hy­wood was never go­ing to please many out­side his fam­ily for ac­cept­ing the bonus while shed­ding work­ers. But the prob­lems Fair­fax faced were much greater than those stem­ming from its man­age­ment, and have been grow­ing for years, just as they have at News Corp, the Guardian and al­most ev­ery other ma­jor news or­gan­i­sa­tion not funded by govern­ment.

It’s self-ev­i­dent that news doesn’t re­port it­self, but the eco­nomic model that has tra­di­tion­ally sup­ported qual­ity jour­nal­ism is mid-col­lapse. News­pa­per rev­enue has been fall­ing by 5% per year world­wide since 2009, ac­cord­ing to Bloomberg. Print cir­cu­la­tion has been fall­ing, as has print ad­ver­tis­ing.

In Aus­tralia, news­pa­per ad­ver­tis­ing rev­enue has dropped 40%, to $2.4 bil­lion, in just five years, ac­cord­ing to Price­wa­ter­house­Coop­ers. By con­trast, the on­line ad­ver­tis­ing mar­ket is grow­ing at 25% a year and on var­i­ous es­ti­mates will be worth $6 bil­lion this year. Ac­cord­ing to Mor­gan Stan­ley, Google and Face­book would gen­er­ate the lion’s share of this, be­tween $4 and $5 bil­lion – around 40% of our to­tal ad­ver­tis­ing mar­ket and ris­ing fast.

Glob­ally, these two tech com­pa­nies ac­count for ap­prox­i­mately half the en­tire dig­i­tal ad­ver­tis­ing mar­ket. Es­ti­mates vary, but it’s widely ac­cepted that they are pick­ing up 80–90% of all new dig­i­tal ad­ver­tis­ing.

By now these trends are re­ported with a de­gree of res­ig­na­tion. The leak of ad­ver­tis­ing to the tech gi­ants seems in­ex­orable. It’s not that read­ers are de­sert­ing the mast­heads: the num­ber of peo­ple who read them ei­ther in print or on­line has never been higher. It’s sim­ply that “print dol­lars turned into dig­i­tal cents”.

The New York Times re­cently added more dig­i­tal-only sub­scrip­tions than in any quar­ter in its his­tory: 300,000 for a to­tal of 2.2 million. Yet its ad­ver­tis­ing rev­enue in the same quar­ter fell by 7%, driven by an 18% drop in print ad­ver­tis­ing.

There hasn’t been a Trump bump in Aus­tralia. And ma­jor news out­lets here don’t have a po­ten­tial au­di­ence of a bil­lion peo­ple.

In May, Greg Hy­wood told the Se­nate se­lect com­mit­tee in­quiry into the fu­ture of pub­lic in­ter­est jour­nal­ism that in the “good old days” 85% of news­pa­per rev­enue was from ad­ver­tis­ing rev­enue and 15% from sub­scrip­tions. Now it’s more like 50/50 – and not be­cause of ris­ing sub­scrip­tions.

While Face­book, Google and “the in­ter­net” may be re­spon­si­ble for the col­lapse of the tra­di­tional me­dia busi­ness, blam­ing them is like hold­ing a shark re­spon­si­ble for bit­ing. Tech­nol­ogy was al­ways go­ing to re­veal mass-mar­ket ad­ver­tis­ing as a blunt in­stru­ment. Print­ing ev­ery sin­gle ad­ver­tise­ment for a sec­ond-hand car, and at­tempt­ing to dis­trib­ute this to ev­ery sin­gle per­son in the mar­ket, may have seemed great at the time, but time makes fools of all of us, es­pe­cially if we’re Fair­fax ex­ec­u­tives. Spray­ing ads for hol­i­days to Fiji across the me­dia was never go­ing to be as ef­fec­tive as sim­ply catch­ing those who googled “flights to Fiji”.

Face­book al­lows ad­ver­tis­ers to tar­get con­sumers by age range, gen­der, lo­ca­tion, ed­u­ca­tion level, po­lit­i­cal lean­ings, in­ter­ests, habits, be­liefs, dig­i­tal ac­tiv­i­ties and pur­chase be­hav­iour; by what they “like” and share, and who their friends are; by what de­vice they use. It can shoot an ad­ver­tise­ment

Its cor­po­rate aim now is to build an un­par­al­leled and ir­re­sistible ma­chine with which to know you and in­flu­ence you.

ad­ver­tise­ment di­rectly into your hand be­cause you’re a mid­dle-aged male who searched on­line for a hard­ware prod­uct and you’re near the new Bun­nings on a Satur­day af­ter­noon.

It knows when you’re hav­ing an­niver­saries, when you’re preg­nant, when you’re plan­ning a bar mitz­vah, and when you’re watch­ing a sad film and might feel like choco­late. What’s more, it’s get­ting smarter at peg­ging your in­ter­ests and vul­ner­a­bil­i­ties ev­ery time you log in. Its nat­u­ral-lan­guage pro­cess­ing and ma­chine-learn­ing al­go­rithms are build­ing a pro­file based on what you look at and for how long, what your friends shared and what you com­mented on. Its systems are gaug­ing why you chose to com­ment on this but not that, and are com­par­ing what you looked at ver­sus what you typed.

Ac­cord­ing to its own Data Pol­icy, Face­book re­ceives in­for­ma­tion about your ac­tiv­i­ties on and off Face­book (loy­alty cards, mail­ing lists, browser cook­ies, re­ceipts, apps, mo­bile phone per­mis­sions and the like) from hun­dreds of third-party part­ners. Ad­di­tion­ally, “we may share in­for­ma­tion about you within our fam­ily of com­pa­nies to fa­cil­i­tate, sup­port and in­te­grate their ac­tiv­i­ties and im­prove our ser­vices”. If you’re us­ing In­sta­gram, What­sApp or At­las, just to give a few ex­am­ples, the data be­longs to Face­book. Or any­one it chooses to share with.

It may have been set up with the best of in­ten­tions – to build com­mu­ni­ties – but its cor­po­rate aim now is to build an un­par­al­leled and ir­re­sistible ma­chine with which to know you and in­flu­ence you.

“It’s a com­mer­cial space; it’s like a shop­ping mall” is how Greens se­na­tor Scott Lud­lam, deputy chair of the in­quiry into the fu­ture of pub­lic in­ter­est jour­nal­ism, de­scribes it to me. “[And] the peo­ple who use Face­book are the com­modi­ties. It’s us that’s be­ing sold to ad­ver­tis­ers. I don’t know if that’s re­ally sunk in.”

We’re dis­cussing the im­pli­ca­tions of a cav­a­lier at­ti­tude towards users’ data and pri­vacy.

“I don’t know if I’d even say they’re cav­a­lier with pri­vacy. They’re min­ing our pri­vacy on a mas­sive scale and that’s the prod­uct: that’s what they sell.

“Their val­ues are some­what ar­bi­trary, and they’re not re­ally con­testable be­cause it’s a com­mer­cial space. ‘If you don’t like shop­ping in our shop­ping mall, you’re free to go sit in the car park.’”

Which might be a rea­son­able ar­gu­ment were it not for the ubiq­uity – the plat­form monopoly – of the on­line gi­ants, and the im­pact two com­pa­nies in par­tic­u­lar are hav­ing on the rest of so­ci­ety. When it comes to col­lect­ing and em­ploy­ing data, they have demon­strated an in­con­sis­tent re­gard for users’ rights.

In 2012, Face­book sup­ported an ex­per­i­ment in which re­searchers ma­nip­u­lated the News Feed of al­most 700,000 users to find out if they could al­ter peo­ple’s emo­tional states. By hid­ing cer­tain words, it was dis­cov­ered, un­sur­pris­ingly, that they could. “We show,” re­searchers an­nounced, “via a mas­sive ex­per­i­ment on Face­book, that emo­tional states can be trans­ferred to oth­ers via emo­tional con­ta­gion, lead­ing peo­ple to ex­pe­ri­ence the same emo­tions with­out their aware­ness.” The ex­per­i­ment was done with­out user knowl­edge or con­sent. When it be­came the sub­ject of con­tro­versy in 2014, the re­searchers first claimed that they did have peo­ple’s con­sent, be­cause it was “con­sis­tent with Face­book’s Data Use Pol­icy, to which all users agree prior to cre­at­ing an ac­count on Face­book, con­sti­tut­ing in­formed con­sent for this re­search”. The Face­book data sci­en­tist who led the re­search claimed it was car­ried out “be­cause we care about the emo­tional im­pact of Face­book and the peo­ple that use our prod­uct”. Fi­nally the com­pany’s chief tech­nol­ogy of­fi­cer apol­o­gised, adding that the com­pany had been “un­pre­pared” for the anger it stirred up, which sug­gests that per­haps it was the back­lash rather than the ex­per­i­ment it­self that caused re­morse.

In 2015, it was re­vealed that Face­book tracks the web brows­ing of every­one who vis­its a page on its site, even if the user doesn’t have an ac­count or has ex­plic­itly opted out of track­ing, and even af­ter a user has logged out.

The Guardian re­ported on the re­search com­mis­sioned by a Bel­gian data pro­tec­tion agency, which ar­gued that Face­book’s data col­lec­tion pro­cesses were un­law­ful.

“Euro­pean leg­is­la­tion is re­ally quite clear on this point. To be legally valid, an in­di­vid­ual’s con­sent towards on­line be­havioural ad­ver­tis­ing must be opt-in,” ex­plained Bren­dan Van Alsenoy, one of the re­port’s au­thors. “Face­book can­not rely on users’ in­ac­tion to in­fer con­sent. As far as non-users are con­cerned, Face­book re­ally has no le­gal ba­sis what­so­ever to jus­tify its cur­rent track­ing prac­tices.”

In May this year, Euro­pean reg­u­la­tors an­nounced that Face­book was break­ing data pri­vacy laws in France, Bel­gium and the Nether­lands, and faces in­ves­ti­ga­tions in Spain and Germany. French reg­u­la­tor CNIL an­nounced that that it was ap­ply­ing the max­i­mum fine that had been al­lowed un­der French pri­vacy law when its in­ves­ti­ga­tion be­gan: a grand to­tal of €150,000. CNIL had last year is­sued an or­der that Face­book stop track­ing non-users’ web ac­tiv­ity with­out their con­sent, and stop some trans­fers of per­sonal data to the US.

“We take note of the CNIL’s de­ci­sion with which we re­spect­fully dis­agree,” replied Face­book. It has ar­gued that it should only be sub­ject to rul­ings from the Ir­ish data pro­tec­tion author­ity be­cause its Euro­pean head­quar­ters are in Dublin. In Europe, though, new per­sonal data pro­tec­tion reg­u­la­tions will come into force mid next year, po­ten­tially al­low­ing reg­u­la­tors to im­pose fines of up to 4% of Face­book’s rev­enues.

Also in May, the Aus­tralian un­cov­ered a doc­u­ment out­lin­ing how the so­cial net­work can pin­point “mo­ments when young peo­ple need a con­fi­dence boost”. By mon­i­tor­ing posts, pic­tures, in­ter­ac­tions and in­ter­net ac­tiv­ity in real

time, Face­book can de­ter­mine when peo­ple as young as 14 feel “stressed”, “de­feated”, “over­whelmed”, “anx­ious”, “ner­vous”, “stupid”, “silly”, “use­less” and a “fail­ure”. The con­fi­den­tial pre­sen­ta­tion was in­tended to show how well Face­book knows its users, and by im­pli­ca­tion how will­ing it is to use this knowl­edge on be­half of ad­ver­tis­ers. Pri­vacy laws in Aus­tralia are nowhere near as strin­gent as in Europe, and not en­forced with any great vigour.

In the US, the Trump ad­min­is­tra­tion re­cently re­pealed data pro­tec­tion rules, mean­ing browser his­to­ries could be sold to ad­ver­tis­ers with­out user con­sent. Ac­cord­ing to re­search from Prince­ton Univer­sity pub­lished last year, Google and Face­book to­gether own all of the top ten third­party data col­lec­tors.

Not that any of this has so far caused any great pub­lic out­cry, ei­ther here or in the States, per­haps be­cause it all ap­pears to be in the ser­vice of giv­ing peo­ple ex­actly what they want. Noth­ing seems to in­ter­est the pub­lic less than de­bates about pri­vacy laws and meta­data col­lec­tion. Un­til re­cently, it didn’t seem to be a ma­jor is­sue.

In June 2007, David Stillwell, a PhD stu­dent at the Univer­sity of Cam­bridge, created a new Face­book app called myPer­son­al­ity. Vol­un­teer users filled out dif­fer­ent psy­cho­me­t­ric ques­tion­naires, in­clud­ing a hand­ful of psy­cho­log­i­cal ques­tions, and in re­turn re­ceived a “per­son­al­ity pro­file”. They could also opt to share their Face­book pro­file data with the re­searchers. Stillwell was soon joined by an­other re­searcher, Michal Kosin­ski, and their pro­ject took off. Peo­ple were happy to share in­ti­mate de­tails, their likes and dis­likes (both on­line and off ), their age, mar­i­tal sta­tus and place of res­i­dence. Be­fore long, the two doc­toral can­di­dates owned the largest ever dataset com­bin­ing Face­book pro­files and psy­cho­me­t­ric scores. In 2012, wrote Hannes Grasseg­ger and Mikael Krogerus in an ar­ti­cle for Das Ma­gasin and Mother­board, Kosin­ski proved that,

on the ba­sis of an av­er­age of 68 Face­book “likes” by a user, it was pos­si­ble to pre­dict their skin colour

(with 95 per­cent ac­cu­racy), their sex­ual ori­en­ta­tion

(88 per­cent ac­cu­racy), and their af­fil­i­a­tion to the Demo­cratic or Repub­li­can party (85 per­cent) … Seventy “likes” were enough to outdo what a per­son’s friends knew,

150 what their par­ents knew, and 300 “likes” what their part­ner knew. More “likes” could even sur­pass what a per­son thought they knew about them­selves.

On the day Kosin­ski pub­lished these find­ings, “he re­ceived two phone calls”, re­ported Grasseg­ger and Krogerus. “The threat of a lawsuit and a job of­fer. Both from Face­book.” Shortly af­ter­wards, Face­book made “likes” pri­vate by de­fault. The per­sonal in­for­ma­tion users put on Face­book had al­ways been owned by the com­pany, to an­a­lyse or sell, but what it was worth was only just be­com­ing clear. (It has a long his­tory of chang­ing pri­vacy set­tings with­out much no­tice or ex­pla­na­tion.)

Face­book wasn’t the only one to regis­ter the po­ten­tial of this tool. A young as­sis­tant pro­fes­sor from the Cam­bridge psy­chol­ogy depart­ment, Alek­sandr Ko­gan, soon ap­proached Kosin­ski on be­half of an­other com­pany that was in­ter­ested in the myPer­son­al­ity data­base. Ko­gan ini­tially re­fused to di­vulge the name of this com­pany, or why and how it planned to use the in­for­ma­tion, but even­tu­ally re­vealed it was Strate­gic Com­mu­ni­ca­tion Lab­o­ra­to­ries. SCL is a com­mu­ni­ca­tions group whose “elec­tion man­age­ment agency” does mar­ket­ing based on psy­cho­log­i­cal mod­el­ling; its off­shoots, one of which was named Cam­bridge An­a­lyt­ica, had been in­volved in dozens of elec­tion cam­paigns around the world. The com­pany’s own­er­ship struc­ture was opaque, and Kosin­ski, who by this stage had be­come deeply sus­pi­cious of its mo­tives, even­tu­ally broke off con­tact.

Kosin­ski was there­fore dis­mayed, if not al­to­gether sur­prised, to learn of Cam­bridge An­a­lyt­ica’s role in last year’s elec­tion of Don­ald Trump.

“We are thrilled that our rev­o­lu­tion­ary ap­proach to datadriven com­mu­ni­ca­tion has played such an in­te­gral part in Pres­i­dent-elect Trump’s ex­tra­or­di­nary win,” said Cam­bridge An­a­lyt­ica’s 41-year-old CEO, Alexan­der Nix, in a press re­lease.

In Septem­ber 2016, speak­ing at the Con­cor­dia Sum­mit in New York, Nix had ex­plained how Cam­bridge An­a­lyt­ica ac­quires mas­sive amounts of per­sonal in­for­ma­tion (legally) – from shop­ping data to bonus cards, club mem­ber­ships to land reg­istries, along with Face­book in­for­ma­tion and other on­line data – and com­bines it (in­clud­ing phone num­bers and ad­dresses) with party elec­toral rolls into per­son­al­ity pro­files.

“We have pro­filed the per­son­al­ity of ev­ery adult in the United States of Amer­ica – 220 million peo­ple,” Nix boasted. Ac­cord­ing to Mat­tathias Schwartz, writ­ing in the In­ter­cept, Ko­gan and an­other SCL af­fil­i­ate paid 100,000 peo­ple a dol­lar or two to fill out an on­line sur­vey and down­load an app that gave them ac­cess to the pro­files of their un­wit­ting Face­book friends, in­clud­ing their “likes” and con­tact lists. Data was also ob­tained from a fur­ther 185,000 sur­vey par­tic­i­pants via a dif­fer­ent un­named com­pany, yield­ing 30 million us­able pro­files. No one in this larger group of 30 million knew that their Face­book pro­file was be­ing har­vested.

It doesn’t take a great deal of imag­i­na­tion to see how use­ful this could be to the Trump cam­paign. As Grasseg­ger and Krogerus re­ported:

On the day of the third pres­i­den­tial de­bate be­tween Trump and Clin­ton, Trump’s team

tested 175,000 dif­fer­ent ad vari­a­tions for his ar­gu­ments, in or­der to find the right ver­sions above all via Face­book. The mes­sages dif­fered for the most part only in mi­cro­scopic de­tails, in or­der to tar­get the re­cip­i­ents in the op­ti­mal psy­cho­log­i­cal way: dif­fer­ent head­ings, colours, cap­tions, with a photo or video.”

The Trump cam­paign, heav­ily out­spent by the Clin­ton cam­paign in tele­vi­sion, ra­dio and print, re­lied al­most entirely on a dig­i­tal mar­ket­ing strat­egy.

“We can ad­dress vil­lages or apart­ment blocks in a tar­geted way,” Nix claimed. “Even in­di­vid­u­als.” Ad­ver­tis­ing mes­sages could be tai­lored, for in­stance, to poor and an­gry white peo­ple with racist ten­den­cies, liv­ing in rust-belt districts. These would be in­vis­i­ble to any­one but the end user, leav­ing the process open to abuse.

In Fe­bru­ary, the com­mu­ni­ca­tions di­rec­tor of Brexit’s Leave.EU cam­paign team re­vealed the role Cam­bridge An­a­lyt­ica had played. The com­pany, re­ported the Guardian, “had taught [the cam­paign] how to build pro­files, how to tar­get peo­ple and how to scoop up masses of data from peo­ple’s Face­book pro­files”. The of­fi­cial Vote Leave cam­paign, Leave.EU’s ri­val, re­port­edly spent 98% of its £6.8 million bud­get on dig­i­tal me­dia (and most of that on Face­book).

Trump’s chief strate­gist, Stephen Ban­non, was once a board mem­ber of Cam­bridge An­a­lyt­ica. The com­pany is owned in large part by Robert Mercer (up to 90%, ac­cord­ing to the Guardian), whose money en­abled Ban­non to fund the right-wing news site Bre­it­bart, and who funds cli­mat­e­change de­nial think tank the Heart­land In­sti­tute.

The crit­i­cal point is not that wealthy con­ser­va­tives may be ma­nip­u­lat­ing pol­i­tics – this is hardly new – but that pol­i­tics has be­come so vul­ner­a­ble to covert ma­nip­u­la­tion, on a scale never be­fore ex­pe­ri­enced.

There is good rea­son for the strict reg­u­la­tions around the world on the use and abuse of the me­dia in elec­tion cam­paigns, yet gov­ern­ments have al­most com­pletely ab­ro­gated re­spon­si­bil­ity when it comes to so­cial me­dia.

Ac­cord­ing to La­bor se­na­tor Sam Dast­yari, chair of the fu­ture of pub­lic in­ter­est jour­nal­ism in­quiry, Aus­tralia’s se­cu­rity agen­cies “are very clear that [de­lib­er­ately mis­lead­ing news and in­for­ma­tion] is a real and se­ri­ous threat … We would be very naive to be­lieve it’s not go­ing to hap­pen here.”

ABuz­zFeed News anal­y­sis found that in the three months be­fore the US elec­tion the top 20 fake-news sto­ries on Face­book gen­er­ated more en­gage­ment (shares, re­ac­tions and com­ments) than the top 20 real-news sto­ries. The Pope en­dorsed Don­ald Trump. An FBI agent sus­pected of leak­ing Hil­lary Clin­ton’s email was FOUND DEAD IN AP­PAR­ENT MUR­DER-SUI­CIDE. In other news, Wik­iLeaks con­firmed that Clin­ton had sold weapons to ISIS, and Don­ald Trump dis­patched his per­sonal plane to save 200 starv­ing marines.

Face­book’s al­go­rithm, de­signed to en­gage peo­ple, had sim­ply given Amer­i­cans what they wanted to read.

The crit­i­cism was heated and wide­spread, prompt­ing Mark Zucker­berg’s ‘Build­ing Global Com­mu­nity’ es­say.

Sure, there are “ar­eas where tech­nol­ogy and so­cial me­dia can con­trib­ute to di­vi­sive­ness and iso­la­tion”, Zucker­berg wrote, and there are “peo­ple left be­hind by glob­al­iza­tion, and move­ments for with­draw­ing from global con­nec­tion”, but his an­swer to these prob­lems was con­sis­tent and uni­form: peo­ple need to be more con­nected (on Face­book). His prom­ises to build a bet­ter net­work – to counter mis­in­for­ma­tion, for ex­am­ple, in a veiled ref­er­ence to the US elec­tion cam­paign, or fil­ter out abuse – rely to some ex­tent on our good­will and credulity. We’re de­nied ac­cess to Face­book’s in­ter­nal work­ings, and that’s as Zucker­berg in­tends it. Which is within his right as chair­man and CEO of a busi­ness. But a net­work this large, this in­flu­en­tial, this se­cre­tive is more than a busi­ness. In many ways, it’s a test of our be­lief in the mar­ket.

Prom­ises to fix prob­lems ranged from in­tro­duc­ing a dif­fer­ent sys­tem for flag­ging false con­tent to work­ing with out­side fact-check­ing out­fits. Per­haps changes were also made to the News Feed al­go­rithm, but if so they re­mained con­fi­den­tial. How would the new com­mu­nity stan­dards be ap­plied? Would Face­book ever make changes that crimp its busi­ness prospects? Does it ac­cept that so­cial obli­ga­tions come with such editorial de­ci­sions? All this re­mained ob­scure, not­with­stand­ing a mess of cor­po­rate non­sense posted by both Zucker­berg and com­pany PR fig­ures.

The News Feed al­go­rithm works like this: in or­der to en­gage you, it chooses the “best” con­tent out of sev­eral thou­sand po­ten­tial sto­ries that could ap­pear in your feed each day. The sto­ries are ranked in or­der of per­ceived im­por­tance to you (Your best friend’s hav­ing a party! Trump has bombed North Korea again!), and the News Feed pri­ori­tises sto­ries you’ll like, com­ment on, share, click on, and spend time read­ing. It recog­nises who posted things and their prox­im­ity to you, how other peo­ple re­sponded to the post, what type of post it is and when it was posted.

As TechCrunch writer Josh Con­s­tine puts it, “The more en­gag­ing the con­tent … the bet­ter it can ac­com­plish its mis­sion of con­nect­ing peo­ple while also earn­ing rev­enue from ads shown in News Feed.”

Over time, as mil­lions have joined Face­book, the num­ber of po­ten­tial posts that might pop­u­late a feed has mul­ti­plied, so the al­go­rithm has be­come not only in­creas­ingly nec­es­sary to pre­vent users from drown­ing in “con­tent” but also in­creas­ingly sub­ject to hu­man de­sign.

Despite this, Face­book in­sists it’s not a me­dia or­gan­i­sa­tion. It’s a tech­nol­ogy com­pany and a neu­tral plat­form for other peo­ple’s con­tent. It is cer­tainly true that it pig­gy­backs on other com­pa­nies’ con­tent. But it is also con­stantly test­ing, sur­vey­ing and al­ter­ing its al­go­rithms, and the changes have vast ef­fects. Kurt Gessler, deputy editor for dig­i­tal news at the Chicago Tri­bune, started notic­ing sig­nif­i­cant changes in Jan­uary, and three months later wrote a post about them, ti­tled ‘Face­book’s al­go­rithm isn’t sur­fac­ing one-third of our posts. And it’s get­ting worse’, on Medium. The Tri­bune’s num­bers of posts hadn’t changed over time, nor had the type of posts. The news­pa­per had a steadily ris­ing num­ber of Face­book fans but the av­er­age post reach had fallen pre­cip­i­tously. “So,” he asked, “is any­one else ex­pe­ri­enc­ing this sit­u­a­tion, and if so, does any­one know why and how to com­pen­sate? Be­cause if 1 of 3 Face­book posts isn’t go­ing to be sur­faced by the al­go­rithm to a sig­nif­i­cant de­gree, that would change how we play the game.”

Face­book has made it clear that it has been in­creas­ingly giv­ing pri­or­ity to videos in its News Feed. Videos and mo­bile ads are, not co­in­ci­den­tally, the very things driv­ing Face­book’s rev­enue growth. It has also been re­ward­ing pub­lish­ers that post di­rectly to Face­book in­stead of post­ing links back to their own sites. None of which bodes well for the Chicago Tri­bune.

There is one way to guar­an­tee your articles will be sur­faced by Face­book: by pay­ing Face­book. As ev­ery so­cial me­dia editor knows, “boost­ing” a post with dol­lars is the surest way to push it up the News Feed. Greg Hy­wood and Huf­fPost Aus­tralia editor-in-chief Tory Maguire pointed out to the Se­nate’s fu­ture of pub­lic in­ter­est jour­nal­ism in­quiry that even the ABC pays Google and Face­book to pro­mote its con­tent. “Traf­fic is dol­lars,” said Hy­wood, “and if the ABC takes traf­fic from us by us­ing tax­pay­ers’ money to drive that traf­fic, it’s us­ing tax­pay­ers’ money to dis­ad­van­tage com­mer­cial me­dia or­gan­i­sa­tions.”

“This is nor­mal mar­ket­ing be­hav­iour in the dig­i­tal space,” replied the ABC, “and is crit­i­cal to en­sur­ing au­di­ences find rel­e­vant con­tent. It is [also] used by other pub­lic broad­cast­ers like the BBC and CBC.” As well as, need­less to say, thou­sands of other me­dia or­gan­i­sa­tions, in­clud­ing Fair­fax and Schwartz Me­dia.

This is what passes for nor­mal mar­ket­ing be­hav­iour in 2017: news or­gan­i­sa­tions, haem­or­rhag­ing un­der the costs of pro­duc­ing news while los­ing ad­ver­tis­ing, are pay­ing the very out­fits that are killing them. Could there be a more di­rect ex­pres­sion of the twisted re­la­tion­ship be­tween them? Could the power bal­ance be any more skewed?

Mark Thomp­son, CEO of the New York Times Com­pany, re­cently put it like this: “Ad­ver­tis­ing rev­enue goes prin­ci­pally to those who con­trol plat­forms.” Over time, this will mean they also con­trol the fate of most news or­gan­i­sa­tions. So it is some­what trou­bling that one of the few ways to keep a check on the power of Face­book is by main­tain­ing a ro­bust fourth es­tate.

It was on so­cial me­dia that I stum­bled across Rachel Baxendale’s Aus­tralian ar­ti­cle about the anti-Semitic abuse di­rected at Ariella on so­cial me­dia. It was also from so­cial me­dia that I ac­quired Baxendale’s con­tact de­tails, to ask her about the ar­ti­cle.

Baxendale had heard of the story through a con­tact, Dr Dvir Abramovich, chair­man of the B’nai B’rith An­tiDefama­tion Com­mis­sion. Abramovich had ver­i­fied the story him­self, and Baxendale then went back and forth with the school­girl over sev­eral days, check­ing de­tails, look­ing at the screen­shots of the abuse, dis­cussing whether to use a pseu­do­nym and so forth. Baxendale had ex­plained the story to her bureau chief, and it then went to the Aus­tralian’s main editorial team in Syd­ney for ap­proval. It was run past the le­gal team and then sub-ed­i­tors, and Face­book was ap­proached for com­ment. All of this is rou­tine at a news­pa­per. If any­one has a com­plaint, it can be taken to the Aus­tralian Press Coun­cil, who will study it im­par­tially be­fore mak­ing a pub­lic rul­ing. Or read­ers can, of course, get in con­tact with the jour­nal­ist in ques­tion or her ed­i­tors.

By con­trast, Face­book has a sin­gle email ad­dress for all global me­dia en­quiries, and its mod­er­a­tors had mo­ments to deal with the mat­ter of Ariella’s abuse. There were 50 pages of screen­shots.

The Guardian re­ported in May it had seen more than 100 in­ter­nal train­ing man­u­als, spread­sheets and flow­charts used by Face­book in mod­er­at­ing con­tro­ver­sial user posts. The Guardian also re­vealed that for al­most two bil­lion users and more than 100 million pieces of con­tent to re­view per month (ac­cord­ing to Zucker­berg) there were just 4500 mod­er­a­tors. That is one per 440,000+ users. Most work for sub­con­trac­tors around the world; Face­book won’t di­vulge where.

This is what passes for nor­mal mar­ket­ing be­hav­iour in 2017: news or­gan­i­sa­tions are pay­ing the very out­fits that are killing them.

They are trained for two weeks, paid lit­tle, and of­ten have “just 10 sec­onds” to cast judge­ment on is­sues in­volv­ing child abuse, sui­cide, an­i­mal cru­elty, ra­cial dis­crim­i­na­tion, re­venge porn and ter­ror­ist threats, and must bal­ance these against the de­sire to re­spect free­dom of speech. Fake news? Factcheck­ing would be im­pos­si­ble. To help them, the mod­er­a­tors are pro­vided with in­struc­tion man­u­als, which con­tain guide­lines for deal­ing with mat­ters from threats and spe­cific is­sues to live broad­casts and im­age cen­sor­ship. Face­book for­mu­lates coun­try-spe­cific ma­te­ri­als to com­ply with na­tional laws, and Zucker­berg of­ten refers to the com­pany’s at­tempts to fol­low com­mu­nity stan­dards, but he re­ally means Face­book’s Com­mu­nity Stan­dards, which it de­ter­mines. (“You can host neoNazi con­tent on Face­book but you can’t show a nip­ple” is Scott Lud­lam’s short­hand char­ac­ter­i­sa­tion of these stan­dards.) Ex­am­ples of this guid­ance, re­layed by the Guardian, give some sense of the im­pos­si­bil­ity of the mod­er­a­tors’ task:

Re­marks such as “Some­one shoot Trump” should be deleted, be­cause as a head of state he is in a pro­tected cat­e­gory. But it can be per­mis­si­ble to say: “To snap a bitch’s neck, make sure to ap­ply all your pres­sure to the mid­dle of her throat”, or “fuck off and die” be­cause they are not re­garded as cred­i­ble threats.

Videos of vi­o­lent deaths, while marked as dis­turb­ing, do not al­ways have to be deleted be­cause they can help cre­ate aware­ness of is­sues such as men­tal ill­ness.

Some pho­tos of non-sex­ual phys­i­cal abuse and bul­ly­ing of chil­dren do not have to be deleted or “ac­tioned” un­less there is a sadis­tic or cel­e­bra­tory el­e­ment.

Pho­tos of an­i­mal abuse can be shared, with only ex­tremely up­set­ting im­agery to be marked as “dis­turb­ing” …

Videos of abor­tions are al­lowed, as long as there is no nu­dity.

Face­book will al­low peo­ple to livestream at­tempts to self-harm be­cause it “doesn’t want to cen­sor or pun­ish peo­ple in dis­tress”.

In De­cem­ber 2015, Face­book gave a com­mit­ment to the Ger­man govern­ment that it would re­move crim­i­nal hate speech from the plat­form within 24 hours; how­ever, a year­long Ger­man govern­ment study re­ported by the New York Times re­cently found that in some months Face­book man­aged to delete only 39% in the time frame sought by the Ger­man au­thor­i­ties, and that its per­for­mance had been get­ting worse in re­cent months. In March, the Ger­man govern­ment pro­posed leg­is­la­tion in this area, with the threat of fines up to €50 million.

Face­book had al­ready an­nounced plans to hire an ad­di­tional 3000 mod­er­a­tors, and Monika Bick­ert, Face­book’s head of global pol­icy man­age­ment, told the Guardian, “We feel re­spon­si­ble to our com­mu­nity to keep them safe and we feel very ac­count­able. It’s ab­so­lutely our re­spon­si­bil­ity to keep on top of it.” As ever, the com­mu­nity will have to take their word for it – and rely on unau­tho­rised leaks to the me­dia for the de­tails.

The in­quiry into the fu­ture of pub­lic in­ter­est jour­nal­ism, driven by sen­a­tors Sam Dast­yari, Scott Lud­lam, Nick Xenophon and Jac­qui Lam­bie, was set up to ex­am­ine “the im­pact of search en­gines, so­cial me­dia and dis­in­for­ma­tion on jour­nal­ism in Aus­tralia”. At pub­lic hear­ings in May, a pa­rade of speak­ers ex­plained the ad­verse ef­fects.

Union rep­re­sen­ta­tive Paul Murphy ex­plained that 2500 jour­nal­ism jobs had dis­ap­peared in Aus­tralia since 2011, and that pay rates for free­lancers had also de­clined sig­nif­i­cantly. The in­quiry heard ex­am­ples of reg­u­la­tions that ap­plied to the Aus­tralian me­dia but not to the tech com­pa­nies, such as those re­lat­ing to lo­cal con­tent and me­dia own­er­ship, and heard time and again about the lack of tax paid by the tech com­pa­nies. Lo­cal me­dia had obli­ga­tions, so­cial, le­gal and cul­tural; Face­book and Google traded on the lo­cal me­dia’s

con­tent, smashed their busi­ness mod­els in the process, and gave lit­tle back to the com­mu­nity in re­turn. How to rem­edy this is what the in­quiry was set up to ex­plore.

Face­book re­ported that it earned $326.9 million of rev­enue in Aus­tralia in 2016. Google re­ported rev­enue of $882 million last year. But these fig­ures rad­i­cally un­der­rep­re­sent the to­tal amount they col­lect from Aus­tralia, which is widely re­garded as at least three or four times larger. For 2016, Face­book and Google re­ported tax bills of $3.3 million and $16.6 million re­spec­tively. A pit­tance. (Per­haps Aus­tralia should count it­self lucky: in the UK in 2014, Face­book paid £4327 in tax, less than what the av­er­age worker paid.)

“Un­til this year,” wrote jour­nal­ist Michael West, “Google and Face­book en­ter­tained a cor­po­rate struc­ture that booked the bil­lions of dol­lars of rev­enue they made in Aus­tralia di­rectly off­shore.” Face­book sales were booked to an as­so­ci­ated en­tity in Ireland, and re­ferred to as the “pur­chase of ad­ver­tis­ing in­ven­tory”. Now, Face­book has de­clared it­self to be a re­seller of lo­cal ad­ver­tis­ing in­ven­tory, and the fed­eral govern­ment has de­clared that its new Di­verted Prof­its Tax – the “Google tax” – will reap bil­lions of ex­tra dol­lars in rev­enue from multi­na­tion­als over com­ing years, though such pro­jec­tions gen­er­ally rely on multi­na­tion­als not al­ter­ing their tax ar­range­ments in re­sponse, and not fight­ing them out in court for the next decade.

Pro­fes­sor Peter Fray, Pro­fes­sor of Jour­nal­ism Prac­tice at the Univer­sity of Tech­nol­ogy Syd­ney, summed up the prob­lem for jour­nal­ism: “There is no doubt there are is­sues around tax for Google and Face­book and they should pay their fair share, but I can­not see how pub­lish­ers, jour­nal­ists or politi­cians can blame Google and Face­book for the fact that dig­i­tal rev­enue streams did not, do not and will not re­place those of print or that in dig­i­tal en­vi­ron­ments au­di­ences have mul­ti­ple choices for con­tent on de­mand 24/7.”

Put a dif­fer­ent way, how could ex­tract­ing a rea­son­able amount of tax from Google and Face­book save lo­cal jour­nal­ism and a col­laps­ing busi­ness model?

Only by fun­nelling that tax rev­enue into jour­nal­ism. For which no mech­a­nisms yet ex­ist, and to which the ob­jec­tions are ob­vi­ous. His­tor­i­cally, there’s been am­ple rea­son to fear govern­ment in­volve­ment in pri­vate me­dia, and lit­tle rea­son to pro­pose or sup­port it. But how quickly things have changed.

“The eco­nomic model no longer works,” Sam Dast­yari tells me. “So ei­ther govern­ment in­ter­venes and finds a way to sup­port it – with­out go­ing so far as to tip the scales of what is and isn’t jour­nal­ism – or we let it die. There is no third op­tion.”

In­de­pen­dent se­na­tor Nick Xenophon is wary of the term “in­ter­ven­tion” when it comes to a govern­ment re­sponse; nev­er­the­less, he agrees that “do­ing noth­ing is not an op­tion”.

“It’s more a case of govern­ment lev­el­ling a play­ing field which has been tipped into a state of im­bal­ance and dys­func­tion by the ad­vent of dis­rup­tors,” he clar­i­fies.

“This is not like the horse-and-buggy and au­to­mo­bile ar­gu­ment of 120 years ago. This is a case where they are pig­gy­back­ing off tra­di­tional me­dia to make a quid. And that should be re­flected in some way in a com­pen­satory mech­a­nism.

“I don’t want us to end up in a world where we just have so-called cit­i­zen jour­nal­ists and a whole range of blog­gers, where there are no stan­dards, where any­thing goes.”

Scott Lud­lam con­curs. “There is a pub­lic pol­icy role here … be­cause the mar­ket’s wip­ing these [me­dia] en­ti­ties out. The mar­ket couldn’t give a shit whether there’s strong and in­de­pen­dent and di­verse jour­nal­ism go­ing on in a so­ci­ety.”

“I don’t think we’ll have gen­eral daily news­pa­pers on a week­day within the next two or three years,” the Greens se­na­tor adds. The loss of ded­i­cated pro­fes­sional re­porters in health, ed­u­ca­tion, state pol­i­tics, arts, sci­ence, en­vi­ron­ment or so­cial af­fairs will have in­cal­cu­la­ble ef­fects.

Do we still have a democ­racy ca­pa­ble of cre­atively mod­er­at­ing the worst ef­fects of the mar­ket? With just a few ex­cep­tions, gov­ern­ments have shown lit­tle in­cli­na­tion to take on cor­po­rate in­ter­ests to pro­tect civil so­ci­ety, or to in­ter­vene to pre­vent mar­ket oli­gop­ol­ies.

Look­ing at the op­er­a­tions of Face­book around the world, you could eas­ily con­clude that Zucker­berg makes the rules.

Peter Fray, also a for­mer editor and chief pub­lisher of the Syd­ney Morn­ing Her­ald and editor of the Can­berra Times and the Sun­day Age, warned the Se­nate in­quiry that me­dia in­de­pen­dence would be at stake un­der a di­rect sub­sidy model, where pay­ments go straight from govern­ment to the me­dia. This is a con­cern shared by the sen­a­tors, who each stress to me the im­por­tance of main­tain­ing the in­de­pen­dence and diversity of the me­dia.

“It’s not an at­tempt by the state to con­trol the me­dia,” says Lud­lam. “It’s a gen­uine in­quiry into how we can sup­port [it].”

One of the key pro­pos­als be­ing can­vassed is a levy on Face­book and Google, which will be used to pay for pub­lic in­ter­est jour­nal­ism. Dast­yari, Xenophon, Lud­lam and Jac­qui Lam­bie are all open to this idea. (Lib­eral se­na­tor James Pater­son, also a mem­ber of the se­lect com­mit­tee, is not.)

“That would free up mil­lions of dol­lars to fur­ther pub­lic in­ter­est jour­nal­ism,” says Xenophon.

How the pro­ceeds of such a levy might be dis­bursed will be the sub­ject of con­sid­er­able de­bate in com­ing months. One method that’s been raised is a Euro­pean-style grants coun­cil, in which a panel ap­pointed by govern­ment de­cides how money is al­lo­cated. How­ever, there are al­ready reser­va­tions among the com­mit­tee mem­bers. “Who de­ter­mines the grants?” asks Dast­yari. “Why do you get a grant and some­one else doesn’t? How in­de­pen­dent is it?”

The La­bor se­na­tor be­lieves the tax sys­tem is the best

way of sup­port­ing the in­dus­try, whether it’s tax breaks for in­di­vid­u­als buy­ing news sub­scrip­tions or those who make do­na­tions or other in­vest­ments in jour­nal­ism. The fact that Xenophon, Lud­lam and Pater­son are also open to such an idea im­plies broad cross-party ac­cep­tance.

“What we’re talk­ing about,” Dast­yari adds, “is actually quite a rad­i­cal re­think of the role of govern­ment as it comes to jour­nal­ism.”

Xenophon is also keen to ex­plore what he calls the “copy­right ap­proach” to sup­port­ing me­dia, via the Competition and Con­sumer Act. He says a for­mula should be de­vel­oped by which the use and shar­ing of in­tel­lec­tual prop­erty is val­ued and com­pen­sated fairly.

The ob­vi­ous chal­lenge all are weigh­ing up is how to de­fine pub­lic in­ter­est jour­nal­ism. “Where do you draw the line?” asks Dast­yari. “Is food blog­ging jour­nal­ism?” Is opin­ion a form of jour­nal­ism? Tele­vi­sion news and cur­rent af­fairs?

“I think the or­gan­i­sa­tions likely to take ad­van­tage of it are not nec­es­sar­ily the ones that most ad­vo­cates have in mind,” Pater­son also warns. “I sus­pect the first peo­ple to seek tax-de­ductible funds to fund a news ser­vice are those seek­ing to pro­mote a par­tic­u­lar world view. On the left, I’d ex­pect to see a GetUp! News, Green­peace News and maybe Asy­lum Seeker Re­source Cen­tre News. On the right I think we’d see an In­sti­tute of Pub­lic Af­fairs News, Aus­tralian Chris­tian Lobby News and Busi­ness Coun­cil of Aus­tralia News.”

Any def­i­ni­tion, em­pha­sises Xenophon, will need to en­sure that an or­gan­i­sa­tion’s dom­i­nant pur­pose is to pro­vide news and opin­ion and that it’s not the arm of an ad­vo­cacy or­gan­i­sa­tion.

The com­mit­tee has un­til De­cem­ber to pub­lish its fi­nal re­port, but it’s al­ready clear that these are the key ques­tions ex­er­cis­ing the minds of com­mit­tee mem­bers – not whether pub­lic in­ter­est jour­nal­ism needs sav­ing, or even whether govern­ment should play a role.

“I sus­pect we’re go­ing to end up with a whole menu of things that ide­ally could work well to­gether,” says Lud­lam.

Re­gard­less of what hap­pens next, it is a re­mark­able shift in pub­lic de­bate.

The talk about the col­lapse of news­rooms in Aus­tralia has un­til re­cently tended to fo­cus on news­pa­pers and mag­a­zines. It’s be­com­ing ob­vi­ous that tele­vi­sion broad­cast­ers both free-to-air and ca­ble are un­der ma­jor pres­sure too. In mid June, Chan­nel Ten went into ad­min­is­tra­tion soon af­ter re­port­ing a $232 million loss driven by flag­ging ad­ver­tis­ing rev­enue.

The fed­eral govern­ment prom­ises to over­haul me­dia own­er­ship laws and scrap some of the con­straints on the big me­dia com­pa­nies, lead­ing to more con­cen­tra­tion and pre­sum­ably greater ef­fi­cien­cies for them. But it voted against the es­tab­lish­ment of the Se­nate’s jour­nal­ism in­quiry and shows lit­tle en­thu­si­asm for rein­ing in the tech gi­ants. The task of pro­tect­ing the diversity of the Aus­tralian news me­dia re­mains be­yond the scope of its am­bi­tion. For the mo­ment, it’s up to oth­ers.

“His­tory,” writes Zucker­berg por­ten­tously, “is the story of how we’ve learned to come to­gether in ever greater num­bers – from tribes to cities to na­tions. At each step, we built so­cial in­fra­struc­ture like com­mu­ni­ties, me­dia and gov­ern­ments to em­power us to achieve things we couldn’t on our own.”

I re­turn to the Face­book page to read Zucker­berg’s es­say one last time, and a mes­sage pops up on my nor­mally dor­mant ac­count. Cit­ing se­cu­rity con­cerns and un­usual ac­tiv­ity, it re­quests that I prove – with photo iden­ti­fi­ca­tion – that I am who I say I am. I hes­i­tate.

“To­day we are close to tak­ing our next step.”

And as Zucker­berg runs through touted im­prove­ments and in­spi­ra­tional in­no­va­tions, ever so ca­su­ally he drops this: “Re­search sug­gests the best so­lu­tions for im­prov­ing dis­course may come from get­ting to know each other as whole peo­ple in­stead of just opin­ions – some­thing Face­book may be uniquely suited to do.” It’s so neatly in­cor­po­rated you barely no­tice it. We “whole peo­ple” and Face­book are sud­denly in­di­vis­i­ble.

Newspapers in English

Newspapers from Australia

© PressReader. All rights reserved.