95TheAuthor(s)2020F.
Stjernfelt,A.
M.
Lauritzen,YourPosthasbeenRemoved,https://doi.
org/10.
1007/978-3-030-25968-6_10InaDanishcontext,thefirsttimethepublicwasexposedtotheseflawswasin2012.
ItwastheseeminglybanalcaseofdocumentarywriterPetervigKnudsen,whopublishedthetwo-volumebookHippieabouttheDanishhippiemovementandtheirso-calledThyCampin1970.
Thephotodocumenta-tionofthesebooksincludedphotosfromthecamp,wherehippieswalkedaroundnaked,swamundressedandthelike.
Itwas,ofcourse,acentralpartofthehippiemovement.
Theseseeminglyinnocentdocumentaryphotos,however,turnedouttoviolateseveralstandardsofseveraltechgiants.
OnNovember1,2012,itbecameapparentthatApplerefusedtomarketvig'sHippie2—referringtothefactthatthebookcontainedphotosofnakedpeople,makingitanactofindecentexposure.
vigstatedaboutApple'siBookstore:"IfApplebecomesthemostdominantbookstoreinDenmark,wehaveasituationwherecensorslocatedintheUS,acrosstheAtlantic,decidewhatbookscanbeboughtinthebiggestbookstoreinDenmark.
"Hethenadded:"Thismakesforaveryunsettlingfutureoutlook.
WehavenoinfluenceoverApple,andourattemptsatdialogwiththemrevealsthattheyhavenointerestwhatsoeverindiscussingthiswithus.
"1Subsequently,theonlyreasonthebookcouldbepurchasedthroughApple'sbookstorewasthatthepublisherGyldendal1Ritzau"Petervig:Apple-censurergrotesk"EkstraBladet.
11-01-12.
Chapter10NipplesandtheDigitalCommunity96preparedacensoredversion,coveringthemoresensitivepartsoftheimages.
AsanironicreferencetoApple,tinyredappleswereplacedacrossgenitalsandbreastsonthephotos.
DanishMinisterofCulture,UffeElbk,declinedtodoany-thingaboutthematter,referencingAppleasaprivateplayerwhoisfreetosetitsownguidelines.
Inthespringof2013,thecasespreadtoFacebook,whereseveralofvig'spostsweredeletedandhewassuspendedfromtheplatform,againcitingindecentexposurepresentonsomeofthephotospostedonhisFacebookpage.
WhenhesubsequentlylinkedfromtheFacebookpagetoawebsitefaroutsideofFacebook,describinganddiscussingwhatheper-ceivedasacensorship,hisFacebookpagewasblockedagain.
Facebookwouldnotevenallowcriticaldiscussionofthecaseoutsidethereachoftheirownservice.
TelevisionnetworkTV2askedFacebook'sDanishrepresentative,ThomasMyrupKristensen,whatwaswrongwithnudity:"Inandofitself,thereisnothingwrongwithnudity.
HereintheNordiccoun-tries,wearequiterelaxedaboutnudity.
Butotherculturesseethingsdifferently.
AtFacebookwehavetoprotectasetofguidelinesthatcoversourwholecommunityof1.
1billionpeople,"saidThomasMyrupKristensen,adding:"Therearerulesastohowwedefinethem.
Icannotgetintoexactlyhowwedoit.
Itisamatterofnudenipplesnotbeingallowed.
Theyarenot—thatiswhatthisisabout.
"2AuthorPetervig'sproblemswouldreturnin2016whenhepublishedabookontheDanishsquatterBZ-Movementoftheearly1980s.
Onceagain,hisFacebookwassuspended—thistimebecauseitcontainedalinktovig'sownwebsitewhichfea-turedaphotofromanudeprotestorganizedbytheBZ-movementinfromoftheCopenhagenTownHallin1983.
Thisconstitutesadetailedregulationthatnotonlyaffectscontentpostedonsocialmediabutalsocontentforsaleinbookstores,aswellascontentonprivatewebpagesentirely2"Facebooktilvig:Derforsletterviditngenfoto"TV2Nyheder.
06-13-13.
Chapter10.
NipplesandtheDigitalCommunity97outsidetherealmofFacebook,aslongasthesepagesarelinkedtofromtheperson'sFacebookpage.
Orinvig'sownwords:"Notbeingabletodecidewhattoputonyourownwebsiteissimplybizarre.
"3ItmayseemaperipheralandharmlessmatterwhetherornottechgiantspreventoldphotosfromtheDanish1970sleftwingfromspreadingontheweb.
Anditisalmostcomicaltofollowhow,forexample,Facebook'sstruggleagainstnudityhasmigratedintotheworldofartandculturalhistoryremov-ing,amongotherthings,classicalworksofartsuchasDelacroix'semblematic"LaLibertéguidantlePeuple"(1830),wheregoddessoffreedomMarianneleadsrevolution-aryFranceinbattle,equippedwiththeFrenchtricolore,abayonettedmusketandbarebreasts.
AnotherremovalwasCourbet'sfamouspaintingfrom1866,"L'OrigineduMonde",featuringaclose-upofastraddlingwoman,aswellasiconicpressphotosfrommoderntimessuchas"TerrorofWar",NickUt'sPulitzerAward-winningphotoofnakedVietnamesechildrenfleeingfromanAmericannapalmattackin1972.
EvenHansHolbein'sdrawingofErasmusbyRotterdam'snakedhandfromaround1532didnotmakeitthrough.
4Overthepasttenyears,anextensiveandstillongoingconfronta-tionhasemergedbetweenFacebook'sremovalpolicyandactivistgroupsofbreastfeedingmotherswho,notentirelywithoutreason,considerFacebook'sremovaloftheirhappyselfiephotosofbabiessuckingtheirbreastsasjustanotherelementinamajorsuppressionofbreastfeedingmothersinthepublicsphere.
ThefactthatFacebookcategorizessuchphotosofmother-childidyllwithrelativelylimitednudityas"obscene"or"adult"hasprobablyoffendedthesemothersjustasmuch—ifnotmore—asthoseimagesoffendedthesensibleuserswhofelttheyneededtocomplainaboutthem3Kjr,B.
"Facebookstrammercensur-skruenoverforPetervig"Politiken.
10-03-16.
4Jones,J.
"FacebookbannedHolbein'shand–butitisn'tevenart'ssauci-est"TheGuardian.
08-31-16.
98tobeginwith.
5Intermsofnudityinadvertisements,Facebook'spolicyisevenmorerestrictive,whichhasmadeBelgianmuseumsdesperate—theycannolongeradvertiseclassicmasterpiecessuchas"AdamandEve"byRubens.
Therefore,theysubmittedanofficialcomplainttoMarkZuckerberginJuly2018.
6However,thesomewhatcomicalfearofnudityillustratesamuchbiggerandmorefundamentalproblem:thetechgiants'increasingremovalofusercontent.
Toalargeextent,thetechgiantslongmanagedtokeeptheirmachineryofremovalhiddenfromthepublic,butascommunicationsresearcherTarletonGillespiepointsout,contentmoderationisoneoftheessentialservicesprovidedbytheplatforms—itmightevenbepartofwhatdefinesthem.
7Althoughthecom-paniesdonotproducecontent,theirreasonforbeing—andwhatseparatesthemfromtheremainingunfilteredwebaroundthem—liesintheirservicesofmoderation,prioritiza-tionandcontentcuration.
Intheeyesofthetechgiants,thefactthatmostusersdonotdiscoverthismachineryandper-ceivetheplatformasopenanduncuratedisasignoftheverysuccessoftheremoval:whenmanyusersperceivetheplat-formsasopentoallideas,itisbecausetheyhaveneverrunintoproblemsthemselves.
Thisisnotthecaseforthemanyuserswhohaveactuallyhadtheircontentremovedandtheiraccountsblocked.
Withanumberofusersinthebillions,thenumberofremovalscanofcoursenotbeamarginaltask;onthecontrary,itissoextensivethatthemachineryofremovalmusttakeplaceonanindustriallevel.
Duetoalackoftrans-parency,itisdifficulttogetqualitynumbersontheseremov-als,butina2014TEDTalk,TwitterVicePresidentDelHarveysaidthat"GiventhescalethatTwitterisat,aone-in-a-millionchancehappensfivehundredtimesaday.
(.
.
.
)Say5InChapter6,Gillespie(2018)mapsthecourseofeventsinthebreast-feedingcontroversy.
6"It'sRubensvs.
FacebookinFightoverArtisticNudity"(AP)NewYorkTimes.
07-28-18.
7Gillespie'sCustodiansoftheInternet(2018)isthefirstcomprehensiveandthoroughstudyof"contentmoderation"onsocialmedia.
Chapter10.
NipplesandtheDigitalCommunity9999.
999percentoftweetsposenorisktoanyonewhatsoever.
Therearenothreatsinvolved.
.
.
Afteryoutakeoutthat99.
999percent,thetinypercentageoftweetsremaining,worksoutroughly150,000permonth.
"8Obviously,thisnum-berhasincreasedsince.
InMarch2017,MarkZuckerbergmentionedthatFacebookgets"millions"ofcomplaintseachweek(formoreexact2018numbers,seech.
12).
9Althoughthepercentageofcontroversialpostsmaybesmall,inabso-lutenumberstheyareenormousandmanyusersareexposedtoremovalsandsanctionswhichittakesalotofresourcestodecideonanddeploy.
AsnotedbyGillespie,mostremovalactivitybythetechgiantstakesplaceonanumberoflevels,aspartofanintricatecollaborationscheme.
10Atthehighestlevelisthecompanytopmanagement.
Rightbelowisthedepartmentresponsibleforongoingupdatesoftheremovalpolicy,whichconsistsofonlyafewhighlyplacedpersons.
Belowthemisthemanage-mentandrecruitmentofthecrowdworkerswhodotheactualimplementationoftherulesbyremovingandmakingsanc-tionsagainstuserswhohavepostednon-standardcontent.
Theyareusuallylow-paidandlooselyemployedpeople,whocanbehiredandfiredaccordingtothegivenworkload,farawayfromthelushmainofficesinPaloAlto,inplaceslikeDublininIreland,HyderabadinIndiaorManilainthePhilippines.
Inaddition,thereisanincreasingamountofautomatedartificialintelligencethat—unlikethehumancen-sorsworkingwithalreadypostedmaterial—flagsuploadedmaterialbeforeitispostedonlinesothatitcanberemovedimmediately,insomecasesevenwithouthumaninvolvement.
Inthiscontext,theefficiencyofAIisexaggerated,however.
In2016,itwasreportedthatFacebook'sAInowfindsmoreoffensivephotosthanthehiredstaffdoes.
11However,the8QuotedfromGillespie(2018)p.
74.
9Ibid.
p.
75.
10Ibid.
p.
116ff.
11Constine,J.
"FacebookSparesHumansbyFightingOffensivePhotoswithAI"TechCrunch.
05-31-16.
100computersmerelyflagthephotoswhichthenstillneedsub-sequenthumancheckbyaremovalworker.
AlthoughthedevelopmentofsuchAIishighlyprioritizedinthecompanies'R&Ddepartments,Gillespiepointstoitsinherentlyconservativecharacter.
AIstillsuffersfromtheproblemthatthemachine-learningprocessthatthesoftwareistrainedbycanonlyteachtheartificialintelligencetorecog-nizeimagesandtextthatarecloselyanalogoustotheappliedsetoftrainingexamples.
Whichessentiallyareselectedbypeople.
Moreover,theAIsoftwareisnotsemanticallysavvy—itrecognizesproxiesonlyfortheproblematiccontent,notthecontentitself.
Forexample,largeareasofskinsurfacetonesareaproxyfornudityandsex.
Theexactwordingofcertainsexualslang,pejoratives,swearwordsorthreatsareproxiesforthepresenceofporn,"hatespeech"orthreateningbehav-ior.
ThismeansthattheAIdetectorshavelargeamountsof"falsepositives"(e.
g.
sunsetswithcolornuancesofpaleskin)aswellas"falsenegatives:"(e.
g.
pornimagesmostlyfeaturingdressedmodels).
Italsoimpliesthattheyareeasyforcreativeuserstofoolbywriting"fokking"insteadof"fucking","bo@bs"insteadof"boobs",etc.
,orbycoloringthepornographicmod-elsincolorsfarfromthatofpaleskin.
Somethingsimilarappliestopoliticalandreligiousterms,whereitispossibletojustwrite"mussels"insteadof"Muslims","migr@nts"insteadof"migrants",etc.
Ofcourse,theprogramscanberevisedsoastocrackdownonsuchcreativespellingaswell,butthisrequiresongoinghumaninputandrevisionwhichcannotinitselfbeautomatized.
Anotherwayistoidentifycontrover-sialcontentindividually,piecebypiece—imagescan,forinstance,beidentifiedbyapixelsequencesothattheycanalsoberecognizedaspartsofeditedormanipulatedversions.
Thistechnologywasdevelopedtocurbchildpornography,butitrequiresthatanextensive"library"ofsuchbannedimagesiskept,againstwhichelementsfoundonlinecanthenbeautomaticallychecked.
Suchtechnologywasusedin2016tofightterrorism,aspartofacollaborationbetweenFacebook,Microsoft,Twitter,andYouTube.
TheysetupaChapter10.
NipplesandtheDigitalCommunity101shareddatabaseofalreadyidentifiedterroristcontent,sothatonlinereuseofthiscontentcouldberemovedautomatically.
12However,saidtechnologyisofcourseonlyabletorecognizealreadyidentifiedcontentandittakescontinuoushumanefforttoupdatethedatabasewithnewcontentfoundonline.
Itisanothervariantoftheoverallfactthatthesoftwarecannotrecognizee.
g.
pornographic,terrorist,threateningcon-tent,otherwiseinstantlyrecognizabletohumans,unlessthematerialhasbeenanticipatedinthesoftware'strainingsyl-labus.
Despitethetechno-optimisminherenttotheindustry,itdoesnotseemasifAIwillbeable—anytimesoonatleast—torespondtonewtypesofcontroversialcontentwith-outhumanmonitoring.
Forexample,AIwillnotperceivenewterroristthreatswithadifferentpoliticalagendathanalreadyknownones.
Inadditiontothesehumanandmechanicalremovalpro-cedures,mosttechgiantsalsorelyonthecontributionfromuserswhohavetheoptiontocomplainaboutcontenttheyhavewitnessedontheplatformbyflaggingit.
13Theuseofthisoptionmakesup,inasense,userco-creationofthesiteasawaytopatrolitsborders,fittingnicelyintomanyofthetechgiants'idyllicself-descriptionasself-organized"communi-ties".
Thisisnotfreefromproblems,however.
Theusersarenoneutral,impartialpoliceforce.
Itisbettertocompareittoaself-organizedmilitiaor"posse"ofaparticulargrouporurbanneighborhood,whomightbecapableofcarryingoutcertainprotectivetasks,butwhomayalsohaveitsownagen-dasandwhoisnotobligedbytheruleoflaw,letaloneknow-ingaboutthelaw.
Ofcourse,thetrendisforthemostannoyedorquarrelinguserstobetheonesfilingthemostcom-12Perez,S.
"Facebook,Microsoft,TwitterandYouTubecollaboratetoremove'terroristcontent'fromtheirservices"TechCrunch.
12-06-16.
13Inadditionto"flaggers",Gillespiepointstoyetanothergroupofmod-erators—communitymanagers,whoareleadersof,forexample,Facebookgroupsandtheso-called"redditors"onReddit.
Theymayhelpenforcethecompanies'policiesandaddtheirown,aslongastheydonotcontravenethoseofthecompany.
102plaints—ratherthanjustleavingthecontentinquestion.
Nooneisforcedtostickaroundandwatch.
Thereisnoguaranteewhatsoeverthatcomplainantsprimarilycomplaintodefendtheplatformpolicy.
Infact,thereisnoguaranteethatcom-plainantsevenknowaboutthepolicy.
Othermotivationsforcomplaintsrangefromdiffuselysensingsomethingtobeinappropriate,toperceivedoffense(whetherornottheoffenseisaviolationofanygivenrule),tooffenseonbehalfofothers(oneisnotpersonallyoffendedbutbelievesthatthecontentmayoffendsomeoneelse,whocomplainantsfeelmoreorlessintheirrighttodefend),tomoralistpushiness,totakingactionagainstviewsthatoneisindisagreementwith,toprivaterevengeagainstpeoplebelonginganopposinggroup.
Sometimesgroupsofusersevenjoinforcestocarryoutcoordinatedflaggingofspecificpersons,groups,orviews,simplytohavethemremovedfromtheplatform.
Whatismore,thereisnotevenanyguaranteethatthecomplaintinfactrelatestothecontentinquestion.
Thesecircumstancesaresupportedbythefactthatthecomplainantremainsanonymousbothtotheremovalstaffandtheaccusedparty,thelatterofwhichisnotinformedaboutwhoreportedandoverwhatexactcontentaspect—usersarecompletelyfreetocomplaininthesensethattheycannotbecriticizedoraccusedoffakecomplaints.
Inmostcases,communicationtotheproscribeduserseemsrudimentary.
Oftenitisnotmadeclearexactlywhatcontentisdeemedproblematicorwhatrulewasallegedlyviolated,appealsmechanismsarenotmen-tioned,andwithmosttechgiantsitisnotoriouslydifficulttogetincontactwithstaffmemberstoobjecttoanunfairdecision.
Someadditionaldetailsseemtoaffectthisflaggingproce-dureandthesubsequentbriefchecksbythetechgiantstaffmembers(oftenoutsourcedtoothercompaniessuchasTaskUs,whichhavespecializedincontentremoval):theuseof"superusers"whosecomplaintpatternenjoysparticularconfidenceandwhosecomplaintsarealmostautomaticallyfavored.
Forexample,YouTubehasa"TrustedFlagger"cat-egory,in2016expandedandrenamedasnothinglessthanChapter10.
NipplesandtheDigitalCommunity103"YouTubeHeroes".
Theprogramturnsflaggingintoakindofcomputergamewithpointsandprizes.
Conversely,techgiantsappeartobeworkingwithlistsofparticularlysuspectuserswhocouldsomehowbeexpectedtopostquestionablecontent—possiblybasedonpostspreviouslycomplainedaboutandremoved,orsimplybasedontheirsuspiciousbehavioronthesiteoranywhereelseontheInternet.
Subsequently,suchproblematicusersmaybesubjecttospe-cialmonitoringaimedatquickandconsistentcrackdowns.
Finally,thisentirecomplexinternalstructureiscontinu-ouslyaffectedfromoutsidebyleakedcasesdiscussedandcriticizedonthesite,elsewhereontheweborintraditionalmedia—which,insomecases,mayputpressureonorjeopar-dizethereputationofthetechgiantinquestion,whomaythenmodifythesetofrulesorpractices,inpublicorcon-cealedfashion.
Giventhelargeamountofcomplaintsandthemanyoddlevelsofthecomplaintmechanism,nobodycanexpectconsistencyinthedecisions,andtherearemanyreportsofpeoplewhohavehadtheiraccountsdeletedforpoststhatareneverthelessstilluponotheruserprofiles.
Userflagginghasseveraladvantagestotechgiants.
Inacertainsense,itoutsourcesanenormousandunforeseeabletasktotheuserswhoworkforfree.
Itmayalsoseemtosup-portthebeliefoftechleadersthattheremovalofcontentismadebyaself-regulating"community"—asifthecomplain-antswereinmutualagreementandasiftheywerenotcom-plainingaboutwidelydifferingthingsandwithwidelydifferingmotives.
Sometimes,thetechelitecomesoffasifbydefinitiontherearenootherproblemsontheirplatformthantheonesflaggedbyusers—suchaswhenZuckerbergspokeatthecongressionalhearings.
Theseromanticideasareofcoursecontradictedbythesimplefactthatitisthecompanieswhomakeandenforcetherulesandnottheuserswhoactasreviewersonly,somewouldsaysnitches.
Differentdeviationsandvarietiesofthisgeneralscenarioarefound.
Filteringcanbeasupplementoranalternativetoremovalssothatcertaincontentcategoriesarereservedforcertainusers.
Forexample,someplatformsusefilteringfor104certaincategories—suchasTumblrwithporn—inasetup14whereusersratetheirpoststhemselvesbasedonhowporno-graphicornottheyare.
Thatwayonlyuserswhohaveprevi-ouslyacceptedithaveaccesstothatcontentcategory.
Certainserviceshave"safesearch",anoptioninwhichusersbefore-handdeselectdifferentcategoriesofcontentfromtheirsearch.
However,asnotedbyGillespie,notenabling"safesearch"doesnotmeanthattheusergetsaccesstoallcontentwhendoingneutralsearches(pornmoviesdonotappearwhensearching"movies",only"pornmovies"),soevenwith"safesearch"disabled,moderationtakesplace.
Filteringisofcourseamilderkindofcontrolthanthefarmorewidespreadpolicyofremoval,butifitisnotdoneaccordingtoveryexplicitprinciples,makingtheuserawareofwhatisgoingon,itnaturallyincreasestheriskoffilterbubbles.
EvenTumblr'srelativelyliberalstancetowardspornhascomeunderattack.
InDecember2018,theplatformannouncedfullprohibitionofallpornographicmaterial.
Thenewruleprohibits"AdultContent.
Don'tuploadimages,vid-eos,orGIFsthatshowreal-lifehumangenitalsorfemale-presentingnipples—thisincludescontentthatissophotorealisticthatitcouldbemistakenforfeaturingreal-lifehumans(nicetry,though).
Certaintypesofartistic,educa-tional,newsworthy,orpoliticalcontentfeaturingnudityarefine.
Don'tuploadanycontent,includingimages,videos,GIFs,orillustrations,thatdepictssexacts.
"15TheargumentbehindseemstobethatinordertoaccommodatetheAppStorepolicyagainstchildpornography,itwasdeemedsafesttoremovethemuchlargercategoryofpornographyingen-eral,inordertoavoidtheissueofborderlinecasesbetweenchildpornandnormalporn.
ThisantiliberalstepgaverisetooutcryfromtheLGBTcommunity,whichhadseenTumblrasaforumofself-expressionandexperiments.
Simultaneously,14Thecategorieshavechangedovertime.
Atonepointthereweretwomarkingsbesidetheneutralstandard:NSFW("NotSafeForWork")for"occasional"nudityandAdultfor"substantial"nudity.
15Tumblr"CommunityGuidelines"modified12-17-18.
Lastvisited12-20-18:https://www.
tumblr.
com/policy/en/communityChapter10.
NipplesandtheDigitalCommunity105thisstepseemstocorroboratethesuspicionthatAppleisincreasinglysettingthelimitsforonlineexpression–andapplyingittoothertechcompanies–becauseofthecentralroleoftheAppStoreasagatewaytoAppledevicesfortheothertechgiants.
Allinall,contentremovalisacomplexprocessinvolvingmanylevels.
Coordinationandcommunicationbetweenlay-ersisaprobleminandofitself,aggravatedbytheexistenceofdifferentversionsofthesamesetofrules.
Onemayaskwhycontentremoval—notonlyonFacebookbutwithmosttechgiants—mustbekeptsosecret.
MediaresearcherSarahRobertshighlightsseveralreasonsbehindthis:thefactthatremovalisconcealedsousersdonotknowthedetailedremovalcriteriamaypreventusersfromtryingtobypassorgametherules,anditalsohidestheelementaryfactthattheentireplatformisaresultofongoingactiveselectionbasedonmonetarymotives.
16Onemightaddthatthewholeideol-ogyofafreeself-organizingcommunityisprobablybestmaintainediftheextentoftheremovalindustrydoesnotseethelightofday.
Atthesametime,secrecygivesusersintoler-ableconditionswhentherulesapproachthestatusoflaws—sincetheRenaissance,acoreaspectoftheideaofruleoflawhasbeenthatlawsmustbepublic.
Foralongtimenow,websiteshavebeenarounddocu-mentingthiscensorship,asitisoftendirectlydubbed—e.
g.
https://onlinecensorship.
org.
Whenlookingathowhyper-detailedandsensitivetheservicesarewhenitcomestotar-getingcontentandadsatindividualusers,itiscurioustoseehowinsensitivetheircommunitystandardshavegenerallybeen.
Bydefault,theyapplytothewholeworld,to"ourentirecommunity",astheDanishFacebookspokespersonsaidinsugar-coatedterms.
Thestandardsarenotinanywayadapted16Roberts,S.
T.
"CommercialContentModeration:DigitalLaborers'DirtyWork"Scholarship@WesternMediaStudiesPublications2016.
Robertsquotesananonymouscontentmoderatorworkingforoneofthetechgiants:"Wehavevery,veryspecificitemizedinternalpolicies.
.
.
theinternalpoliciesarenotmadepublicbecausethenitbecomesveryeasytoskirtthemtoessentiallythepointofbreakingthem"p.
5.
106tocultural,geographicalletalonejurisdictionaldifferences.
TheylargelyreflecttheworldviewofasmalleliteofwealthymenandsoftwareengineersatthetopofSiliconValley.
InmostWesterncountries,thesecommunitystandardsaremorestringentthanlocallawsonexpressions,whereasinothercases,especiallyintheMiddleEastandAsia,theymaybemorelenient.
Buttheprinciplethatthestandardsshouldbethesamethroughouttheworldgivesthemanaturaltendencytobeorganizedbythelowestcommondenominator.
Ifsome-thingistabooedorfeelsoffensiveinonepartintheworld,thenviathecommunitystandardsthatsomethingisexpandedtoallusersacrosstheworld.
vig'sharmlessimagesofnakedhippiebreastsdidnotraisemanyeyebrowsinDenmarkwherehisbookwasinitiallypublished.
Buttheybecameaproblembecauseinother,morestraitlacedplacesoftheworld,therearepeoplewhoclaimtobeshockedatthesightofpeople'snaturalbodies.
Somethingsimilargoesforthecritiqueofreligion,whichiswidelyusedinsomeplacesandevenregardedasanintellectualendeavorofacertainstan-dard,whileelsewhereintheworlditisregardedasblasphemyandmayeveninvokethedeathpenalty.
Inthatarena,thetendencyisalsoforthelowestcommondenominatortopre-vail.
Whenitcomesto"hatespeech",thereareverydifferentstandards—asmentioned,itisnotacrimeintheUS,butinmanyEuropeanandothercountriesitis,anditusuallyalsoappearsamongexpressionsregulatedbytechgiantstandards.
Althoughbydefaultthesamestandardsapplytothewholeworld,inrecentyearspressurefromdictatorstatesandauthoritarianregimeshasmadeseveraltechgiantstightenthecriteriatoalignthemwithlocallaws.
Forinstance,FacebookgaveintoTurkishpressurebyforbid-dingspecificcritiqueofIslamontheirTurkishFacebook—includingabanoncaricatureimagesofprophetMuhammad.
ThathappenedinJanuary2015,onlytwoweeksafterZuckerbergproudly,promptedbytheCharlieHebdomassacreinParis,hadstatedthefollowing:"WeneverletonecountryorgroupofpeopledictatewhatChapter10.
NipplesandtheDigitalCommunity107peoplecanshareacrosstheworld.
"17In2010,AmericanillustratorMollyNorrisannounced"EverybodyDrawMohammedDay".
ItwasbasedontheideathatiftheInternetdrownedindrawingsofMuhammad,censorshipofthemwouldbepracticallyimpossible.
PakistanfiledproteststoFacebookandthreatenedtoclosedowntheservice.
Morris'initiativewasnotcoveredbyFacebook'srulebook,soFacebook'simprovisedreactionwastoremoveNorris'sbigFacebookgroupwithhundredsofthousandsofusersinPakistan,IndiaandBangladesh.
EvenGoogleagreestosuchlocalconcessions:ontheRussianversionofGoogleMaps,CrimeaisnowapartoftheRussianFederation.
Thusthetendencyisthatincaseswheretheprincipleofoneandthesamestandardforthewholeworldisdepartedfrom,thedepartureistowardsfurtherlocalnarrowingofwhatcanbesaid,notliberaliza-tion.
18Thereisnogeneralopennessortransparencyabouttherangeandnatureoftheselocalcompromisesmadewithnon-democraticregimes.
Gillespiepointsoutthatsuchmodifications—justlikepersonalfilterbubbles—areinvisiblebecausenotonlyiscontentremoved,butthereisnomentioningofsaidremoval:theproblemis".
.
.
theobscuredobscuringofcontentiousmaterialfromthepub-licrealmforsome,notforothers.
"Itcreatesfilterbubblesonthenationalorreligiouslevel,sotospeak.
The"globalcommunity"maynowlookcompletelydifferentindiffer-17Dewey,C.
"MarkZuckerberg'sFacebookcensorsimagesoftheProphetMohamedinTurkey–twoweeksafterhedeclared'JeSuisCharlie'"Independent.
01-28-15.
18FacebookisallegedlyenteringintorestrictiveagreementswithPakistanoncompletelyremovingblasphemy(whichissubjecttothedeathpenaltyinthecountry),withVietnamonremovingcommentscriticalofthegovernment,andwithThailandonremovingcriticismoftheroyalfamily,cf.
Gillespie(2018)p.
38.
Inthisrespect,FacebookseemsmoreadaptablethanGoogle,whopulledoutofChinaduetothatcountry'sdemandforcensorship.
108entcountriesandmaynotbeasglobalasonceproclaimed.
Lookingatthestandardsofthedifferenttechgiants,wefindsomethingstriking.
FromthePinterest"TermsofService":"YougrantPinterestandourusersanon-exclusive,royalty-free,transferable,sublicensable,worldwidelicensetouse,store,display,reproduce,save,modify,createderivativeworks,perform,anddistributeyourUserContentonPinterestsolelyforthepurposesofoperating,developing,providing,andusingPinterest.
NothingintheseTermsrestrictsotherlegalrightsPinterestmayhavetoUserContent,forexampleunderotherlicenses.
WereservetherighttoremoveormodifyUserContent,orchangethewayit'susedinPinterest,foranyreason".
19Aswithothertechgiants,usersgivethecompanytheright,toasurprisingextent,tousetheirper-sonaluploadedcontentinanumberofways,includingcom-mercialones.
Butthecrucialpartherearethelastthreewords:"foranyreason".
Therighttoremoveormodifyusercontentisreservedforanyreason.
Similarunspecifiedformu-lationsarefoundinmanytechgiantpolicies,whichmaytalkaboutremovingcontent"atanytime"orlaunchinglistsofcriticalcontentwiththephrase".
.
.
including(butnotlimitedto).
"20Thisisanalogoustocensorshiplawswhichfailtospecificyexactlywhichtypesofstatementsareinfactcrimi-nalized.
Intheabsenceofsuchspecification,thelawsmaybeappliedtoanunlimitedamountofundefinedinfringements;potentiallytoallkindsofstatements.
Thelegalprotectionoftheuserisnon-existentiftheextenttowhatcanberemovedremainsunknown.
InaDanishcontext,wearesomehowback19Pinterest"Termsofservice"Mostrecentvisit06-25-18:https://policy.
pinterest.
com/en/terms-of-service.
20Forinstance,theconditionsofFacebook-ownedInstagram:"Also,InstagramreservestherighttoremoveanyContentfromtheServiceforanyreason,withoutpriornotice".
Lastvisited12-19-18:https://help.
insta-gram.
com/478745558852511,or"SpotifyreservestherighttoremoveordisableaccesstoanyUserContentforanyornoreason".
Lastvisited12-19-18:https://www.
spotify.
com/legal/end-user-agreement/Chapter10.
NipplesandtheDigitalCommunity109attheDanishpolice-issuedcensorshipof1773,inwhichapolicecommissionerhadthesovereignrighttoconfiscatewhichbookshedeemedillegal,withoutanypriorcourtdeci-sion.
Itisalsocharacteristicthatcontentremovedbythetechgiantsforanyreasonisremovedoffhand,withoutthecriteriabeingclear,withoutknowingwhetheritwasdonemanuallyorautomatically,withouttheuserbeinginformedwhy,with-outanythingthatlookslikeacourtdecisioninwhichargu-mentsproandcontracanbeputforwardbasedonasoundlegalbasisandfinallywithoutanyformalorclearappealpath.
WhenitcomestoPinterest's"TermsofService",thepolicyonprohibitedcontentfitsononesingleline:"DonotpostpornographyorspamoractlikeajerkinfrontofothersonPinterest.
"Thismaysoundfunky,idiomaticandstraight-forward,butitexposesalackofmoreprecisedescriptionofwhattypesofactionsmakesoneconsidered"tobeajerk".
Thisgivesthatparticularcompanyalegallicensetodoasitpleases.
ThestandardprocedureforcontentremovalonFacebookstartswithusersviewcontentthattheyfindproblematicoroffensive.
Theythendecidetoflag,tosubmitacomplaint,whichthenconstitutesthebasisforremoval.
ThenitisuptoFacebook'ssafetyandsecuritystaff—ateamasperearly2018consistingof7,500people,butwhichafterFacebook'scrisisinthespringof2018wassetouttoaquickexpansiontoaround30.
000,manyofwhichareallegedlybasedinthePhilippines—tocheckthegivencontentanddeterminewhichcomplaintsshouldbefavored.
Thisessentiallymeansthattheneedsof"offended"usersaremet—butitalsomeansthatonlyproblemsvisibletotheuserscanbeaddressed.
21AtthecongressionalhearingsinApril2018,whencommenting21However,farfromallpotentiallyproblematiccasesarevisiblefromthepointofviewoftheindividualuser.
Forexample,ifanadisonlytargetedtowhitebuyers,non-whitebuyerswillneverknowanythingabout,simplythisbecausetheydonotseethead.
Manyissuesregardingtargetedandpersonalizedcontentmayescapenoticebecauseusershavenotransparencywhenitcomestopersonalization.
110onvariousproblematicissuesraisedbyUSpoliticians,ZuckerbergrepeatedlysaidthatFacebookwouldobviouslytakecareoftheseissuesif,itshouldbenoted,anyusersfiledrelevantcomplaints.
Thisistosaythatifnouserswerecom-plaining,thentherecouldbenorealproblemandthepoliti-ciansneednotworry.
Accordingtothislogic,itisthecomplainants(andtheadvertisers)whosettheagendaandthattheyinfactcomprisethesolespecificationorcorrectionofthecompanyguidelines.
Manyhavediscoveredthisandareabletoabuseittoorganizesocialmediashitstormsagainstviewsandpeopletheydislike—whomaythenberemovedfromtheplatformbyitsspeedyrepresentatives,seebelow.
Accordingtothesamelogic,structural,general,andstatisticalaspectsofthealgorithms–invisiblefromtheview-pointofsingleusers–couldneverappearasrealproblemsthecompanyoughttoaddress.
Asmentionedearlier,manualremovalofcontent,asaresponsetocomplaints,isincreasinglysupportedbytheautomatedremovalofcontentbasedonalgorithmscon-stantlydevelopedandmodified.
Therehasbeennopublicaccessto,neitherthealgorithmsthemselvesnortheexactcriteriaonwhichremovalsarecarriedout.
Thecriteriaarerevisedandmodifiedonaregularbasis,andrarelycommu-nicatedtothepublic.
AtthehearingsinApril2018,Zuckerbergplacedthemainresponsibilityfortherecentremovalof"terrorist"contentfromFacebookonthealgo-rithms,claimingthat99%ofISIS-relatedmaterialwasnowremovedautomatically.
Herepeatedlycitedfurtherdevel-opmentandsophisticationofAI-basedalgorithmsasthesolutiontomanyoftheissueshighlightedbymembersofcongress—includingtheremovalofmanydifferenttypesofcontent,boththosealreadycoveredbythestandardsandcontentfoundproblematicbydifferentpoliticians.
Butthepublichasnoinsightintoorguaranteethatalltheafore-mentioneddeletedISISpostsactuallycontainseriousactssuchasthreats,incitementtoviolence,organizationofter-roristcells,etc.
Needlesstomention,nothingcouldbefar-therfromthewritersofthisbooktohavetheremotestChapter10.
NipplesandtheDigitalCommunity111sympathywithISIS,butthatisexactlythereasonthesecasesaresuitabletoillustratetheproblem.
InterpretationsofforinstancetheISIStheology,itsunderstandingofgov-ernment,politics,strategy,propagandaetc.
shouldnotberemoved,inouropinion,andcertainlynotautomaticallyandwithoutcontrol.
Theclassiclineofargumenttodefendthepositionthatsuchcontentshouldalsobetoleratedgoesasfollows:ifnotallviews—alsoabominableones—areallowed,theresultisapublicspherecharacterizedbydis-honesty,pretenseandhypocrisy,becausepeopleareforcedtohidewhattheyreallymean.
Ifcertainviewsarebanned,theywillnotjustdisappear,butinsteadgetorganizedintheunderground—forinstanceonTheDarkWeb—wheretheymaygainasheenofmartyrdom,becomeradicalizedandevenmoredifficulttocontrol.
Theresultisfragmentationofthepublicsphere.
Itmakesitharderbothforthepublic,forresearchersandforsecretservicestounderstand,forexam-ple,whatkindofphenomenonISISis,ifnoonehasaccesstotheirdistortedviews.
Andifradicalviewsarerejected,supportersofsuchviewsaremorelikelytoconcludethatdemocracydoesnotincludethem;therefore,theymusttakeanti-democraticaction.
Onecouldalsoaddthatthepres-enceofgrotesqueandextremeviewsinthepublicspherehasadiversityoffunctions.
Thisdoesnotonlymakefortherecruitmentofmoresupporters.
Italsogivesthepublictheopportunitytobeawareof,disgustedatandrenouncesuchpositions—likeavaccinethatmakessocietyreadytobattleagainstsuchpositions,areadinesswhichmayloseforceiftheseideologiesdevelopinmoreclandestinefashions.
Thecensorshipprocedurenotonlyincludestheremovalofcontent,beitalgorithmicallybycensorshipbeforethefactormanuallybycensorshipafterthefact.
Italsohappensbycategorizingcertaintypesofusersassuspiciousandthensubjectingthemtospecialmonitoringoftheironlinebehav-ior.
Categorizationofsomeoneassuspiciouscanofcoursebethesimplefactthatonrepeatedoccasionsthepersonhashadcontentremoved,manuallyoralgorithmically.
Butitcanalsojustbeasignificantchangeinone'smetadata—data112mappingone'snetworkconnectivityandonlinebehavior.
Ifthatbehaviorsuddenlyinvolvesnewgroups,nottomentiongroupsincontradictiontooneanother,orifitcomesoffasstrikinginanyotherway,thenthe"standing"withthetechgiantcanbechangedfromlightgreentored,whichmeansthattheuser'sbehaviorissubjecttospecialmonitoringinordertoimmediatelycrackdown,iftheuserpostssome-thingthatviolatesthecommunitystandards.
Monitoringusershappens,ofcourse,primarilyforadvertisingandcom-mercialpurposes—butitalsohasapoliticaldimension,sotospeak,inthewaythatitcategorizesusersbeforethefact,iftheyarethoughttopossiblyviolatethecommunitystan-dards.
22InAugust2018,thepublicwasgivenasmalltastesampleofthiswhenFacebookannounceditsuseofaTrustIndextorateusersonascalebetween0and1.
Thepurposeistoidentifymalicioususerstocrackdownupon.
Theindexmostlikelyincludesrepeatedviolatorsofthecommunityrules.
AccordingtoFacebook,thiscredibilityassessmentalsoseemstoincludewhetherusershavereportedsome-thingasfalsewhichinfactwasonlyamatterofdisagree-ment—apparentlythefirstattemptatpossiblesanctioningagainstabuseoftheflaggingfeature.
Similarly,theindexalsoincludeswhichpublishersareconsideredcrediblebyusers.
Asalways,thereisnotransparencyabouthowthesecomprehensivecredibilityassessmentsarearrivedat,whetherallusersareassignedacredibilityscore—andthereisalsonoindicationofwhetheritisevenpossibleforuserstogaininsightintotheirownindexandpotentiallyfileanappealincaseoferror.
23Theoverallpictureisthatbyusingtheirdefactomonop-oly,thetechgiantsareactuallyintheprocessofturningtheircommunitystandardsintothenewlimitsofthepublicsphere.
22Wehavethisdescriptionofthecategorizationofusersfromanexpertsourcewhohasvisitedandinterviewedthetechgiantsonseveralocca-sionsandwhohashadprofilesclosed;weareawareoftheidentityofthepersonwho,however,haspreferredtoremainanonymous.
23E.
Dwoskin.
"FacebookistheTrustworthinessofitsUsersonaScalefromZeroto1"WashingtonPost.
08-21-18.
Chapter10.
NipplesandtheDigitalCommunity113Fromtheiroriginaseditorialprinciplestheseguidelinesslowlytransformandbecomedefactocensorshiplegislation.
Andthesenewlimitsarefarnarrowerthanthebroadlimitsforfreedomofspeechgenerallysetoutinmodernliberaldemocracies.
OpenAccessThischapterislicensedunderthetermsoftheCreativeCommonsAttribution4.
0InternationalLicense(http://creativecom-mons.
org/licenses/by/4.
0/),whichpermitsuse,sharing,adaptation,distri-butionandreproductioninanymediumorformat,aslongasyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreativeCommonslicenseandindicateifchangesweremade.
Theimagesorotherthirdpartymaterialinthischapterareincludedinthechapter'sCreativeCommonslicense,unlessindicatedotherwiseinacreditlinetothematerial.
Ifmaterialisnotincludedinthechapter'sCreativeCommonslicenseandyourintendeduseisnotpermittedbystatutoryregulationorexceedsthepermitteduse,youwillneedtoobtainpermissiondirectlyfromthecopyrightholder.
今天遇到一个网友,他在一个服务器中搭建有十几个网站,但是他之前都是采集站点数据很大,但是现在他删除数据之后希望设置可能有索引的文章给予404跳转页面。虽然他程序有默认的404页面,但是达不到他引流的目的,他希望设置统一的404页面。实际上设置还是很简单的,我们找到他是Nginx还是Apache,直接在引擎配置文件中设置即可。这里有看到他采用的是宝塔面板,直接在他的Nginx中设置。这里我们找到当前...
老薛主机怎么样?老薛主机这个商家有存在有一些年头。如果没有记错的话,早年老薛主机是做虚拟主机业务的,还算不错在异常激烈的市场中生存到现在,应该算是在众多商家中早期积累到一定的用户群的,主打小众个人网站业务所以能持续到现在。这不,站长看到商家有在进行夏季促销,比如我们很多网友可能有需要的香港vps主机季度及以上可以半价优惠,如果有在选择不同主机商的香港机房的可以看看老薛主机商家的香港vps。点击进入...
搬瓦工最近上线了一个新的荷兰机房,荷兰 EUNL_9 机房,这个 9 的编号感觉也挺随性的,之前的荷兰机房编号是 EUNL_3。这次荷兰新机房 EUNL_9 采用联通 AS9929 高端路线,三网都接入了 AS9929,对于联通用户来说是个好消息,又多了一个选择。对于其他用户可能还是 CN2 GIA 机房更合适一些。其实对于联通用户,这个荷兰机房也是比较远的,相比之下日本软银 JPOS_1 机房可...
pornmovies为你推荐
商标注册流程及费用商标注册流程及费用?商标注册流程及费用注册商标的程序及费用?陈嘉垣反黑阿欣是谁演的 扮演者介绍psbc.com邮政银行卡6215995915000241921是哪个地区的同ip站点查询如何查看几个站是不是同IP网站检测请问,对网站进行监控检测的工具有哪些?mole.61.com摩尔庄园的米米号和密码我都忘了 只记得注册的邮箱 怎么办-_-m.2828dy.combabady为啥打不开了,大家帮我提供几个看电影的网址广告法请问违反了广告法,罚款的标准是什么www.jsjtxx.com苏州考驾照,理论考试结束后,要在网上学习满12小时,网站是什么
asp网站空间 vps优惠码 美国加州vps 国外php主机 免备案cdn 免费cdn加速 xfce 天猫双十一抢红包 国外在线代理 最好的免费空间 泉州移动 免费活动 网通服务器托管 银盘服务是什么 无限流量 域名与空间 免费外链相册 视频服务器是什么 中国电信网络测速 智能dns解析 更多