Devangshu Datta on how fake accounts drive the world of Twitter HOW TWITTER BOTS WORK
Remembering Girija Devi, the voice of Benares Celebrating Kabir by the Ganga Inside Petra, the lost city of stone Your guide to becoming a superbiker
OnOctober15, at1.56pm, theofficialTwitteraccount ofRahulGandhireactedtoatweetbyUSPresident DonaldTrumppraisingAmerican-Pakistani relationswiththistweet:“Modijiquick, lookslike PresidentTrumpneedsanotherhug.” Thetweet quicklyzoomedto30,000-plusretweets. OnOctober21, newsagencyANIcarriedareportclaimingthat Gandhiwasusingfaketwitteraccounts— so-calledtwitterbots fromKazakhstan, Russiaandothernations— toretweethistweets andtherebyamplifytheapparentpopularityofhisaccount.
ThestoryshowedscreenshotsofseveralautomatedTwitter accountsthatretweetedthatTrumptweet. Thisstorywasseized upongleefullybytheBharatiyaJanataParty. Itssocialmediacell spreadthestoryfarandwideusinghashtagslike#RahulWaveinKazakh. BJPministerSmritiIranimadesnideremarkstotheeffect thatGandhiintendedtowinelectionsinRussiaandKazakhstan.
TheCongresssocialmediateamrespondedquicklybypointingoutseveralodditiesabouttheTwitterbotsthatwerecitedinthe story. TheseaccountshadbeencreatedseveraldaysafterGandhi hadmadethatTrump-relatedtweet. Theseaccountsspecifically statedlocationdata(whichcanbeturnedonandoffatwill) allowingthemtobeidentifiedasKazakhandRussianaccounts.
TheBJPITcellhadalsostartedpublicisingtheANIstorya coupleofhoursbeforeitwasactuallyreleased— asignthatitwas clearlyintheknowofthewholeaffair. DivyaSpandana, who managessocialmediaanddigitalcommunicationforthe Congress, saidthestorywas“factuallywrong” andmanyexperts saidthatthislookedlikeaclumsilyfakedstingoperation wherethebotshadbeencreatedbyright-wing supporterstotryandembarrassGandhi.
Thisisthelatestinstalmentintheongoing battletowinvisibilityonsocialmedia. TheBJP hasthelargestandmostorganisedsocialmedia apparatusandPrimeMinisterNarendraModi himselfhasover38millionfollowerscomparedto RahulGandhi’srelativelymeasly3.9million. Butthe Congressispushinghardtoincreaseitssocialmedia visibilityandtheAamAadmiPartyisalsopretty goodatmanagingtheTwittersphere.
TheCongress’socialmediawingcategoricallysays thatitwouldnotusebotstotryandgenerateTwitter volume. ButdespitetheCongress’assurances, fakeTwitter accountswillalmostcertainlybedeployedbyallsidesinthisbattle forsocialmediavisibility. Infact, fakeaccountscontrolledand deployedbyneutralthirdpartiescouldendupoperatingfreelance onbothsides!
Any21st-centurypoliticalcampaignwillinvolvealotof jockeyingforsocialmediaterritory. Thehighertheprofileofthe campaign, themorelikelyitistodraw“freelancers”. Boththe TwitteraccountsofModiandGandhihaveahugenumberoffake followersandmanyofthosearefreelance.
Everyhigh-profileaccountattractsbots. Botcontrollersseek outhigh-profile, high-volumeaccountsthattheycanfollow andretweetinthehopeofgainingmorefollowersthemselves. For example, aEuropeanescortserviceusedtofollowModiandfaithfullyretweetedhimforawhile, in-betweenshillingforclients!
Itmayseemabsurdthatanoverseasescortservicewould attempttogaintractionfromtheSwachhBharatMissionbutthat istheabsurdwayinwhichsocialmediasometimesworks. Followingandretweetingahigh-profileaccountisalow-costwaytogain somevisibility. Perhapsonlyoneoutofeverymillionrealfollowers willnoticethefakeaccount, butit’safreeride. Ifahackercontrols 10,000bots(andthat’snotimpossible), hecanmultiplyhis chancesofbeingnoticed. It’sevencheaperthantheoldgameof sendingoutspamemails. Thatinvolvedbuyinge-maildatabases. Twitterisafreeservicewithmostaccountspubliclyvisible.
Trump’spresidentialcampaignisknowntohavegained tractionfromonesuchfreelanceoperation. Macedoniaisoneof severaltech-savvyEasternEuropeannationswhereexploiting socialmediatogarneradrevenueisapopularformof entrepreneurship. InthefinalweeksoftheUSpresidential election, thesmalltownofVelesinMacedoniaattainedshort-lived famebecauseatleast100pro-Trumpwebsiteswerebeingrunout ofthere. ManywerefilledwithblatantfakenewssuchasTrump slappedavoteronthecampaigntrail; HillaryClintonfaced criminalcharges; thePopeendorsedTrump, andsoon.
In early 2016, a tech-savvy teenager from Veles was running multiple blogs, cut-pasting content on a dozen different topics (sports, politics, health care) in the hope of generating ad revenue. He discovered that news about Trump generated lots of views, which of course translated into better advertising revenue. He also set up similar blogs extolling Bernie Sanders and Clinton but Trump supporters seemed more gullible and generated more ad clicks.
SohefocussedonTrump, settingupwebsiteslike PoliticsHall.comandUSAPolitics.cotoridetheTrumpcampaign. Hesourcedcontentfromdozensofobscureright-wingpro-Trump Americansources, directlycut-pasting. HecreatedFacebook groupswithnameslike“MyAmerica, MyHome”;“The Deplor-ables” and“FriendsWhoSupportPresidentDonaldJ Trump” topublicisethecontent. Heandhisequallytech-savvy friendscreatedfakeFacebookprofilesandfakeTwitteraccounts topushthoselinksout. Eachofthosestorieswasimmediately discreditedbymainstreammediabutthatdidn’tmatter. Social mediaisabubbleandmanyTrumpsupporterscontinuedto believethelies. Itallfellapartaftertheelection, butbythenthe Macedonianshadmadetheirpile.
Itwasn’tjusttheMacedonians, ofcourse. TheTrump campaignisreckonedtohavehadfourtimesasmanybots workinginitsfavourcomparedtotheClintoncampaign. Bythe finalstages, theTrumpbotsweregeneratingseventimesas muchTwittervolumeontherelevanthashtags. TheTrump campaignspentmillionscraftingitsFacebook-Twitter campaignandmicro-targetingthemessagetosuitspecific swingstates. And, ofcourse, theRussiansareallegedtohave helpedTrumpfortheirownreasons.
TheBrexitcampaignhadasimilarhigh-botpresence. Lies andfakestatisticsweregeneratedbythepro-Brexitcampand thenpropagatedbytheuseoftheseautomated mechanisms. Itworked. Itwasonlyafterthe referendumthatalotofBrexitvotersrealised thattheyhadbeensoldadummy.
TheIndianTwitterspherehasseen similaruseofautomationssinceat least2013. TheAnnaHazareagitation, right-wingfriendsoftheBJPandthe BJPitselfhaveallusedsocialmedia withconsummateskill. Hashtagshave beentrendedbyusingacombinationof verifiedhigh-profiletweeterstomakethe originaltweetsfollowedwithmanyautomated accountsspewingoutcoordinatedretweets.
Quiteoftenthecontenthasbeenspurious. Photos sourcedfromBangladeshhavebeenpassedoffas communalincidentsinKerala; theAhmedabadriverfront hasbeenPhotoshopped, quotesfrommediapersonshavebeen selectivelysplicedtogether tocreateamisleadingimpression.
Inadditiontopropagatingthissortoffakenews, Twitterhas alsobeenusedcleverlytobuildBrandModiandinversely, to brandRahulGandhiasincompetent, usingthetoolsof satireandirony. TheCongressisonlybeginningtocatch up. Ithaslatelystartedtomakeanimpactbyusingcatchy songslikethe“VikasGandoThayoChey( Developmenthas gonemad)” lampooninGujarati.
Botshavealsobeenusedtorelentlesslytrollpolitical opponentsandtotrollmediapersonswhoareperceivedtobe unsympathetic. Thisisoftendoneinthevilestoftermswithabuse, rapethreatsanddoxingofinformationthrownin. Inatleastone case, Delhi-basedjournalistSwatiChaturvediwastrolledbya tsunamiofright-wingbotsemanatingoutofSouthThailand. Of course, trollingisnotthesolepreserveofpolitical tweeters. Moviecriticsaretrolledbyfansofmoviestars. Sportsfanstrollsportsjournalists.
Thetechnologyisagnostic. Botsarealsousedto propagateusefulinformationandtorunFAQsfor prominentcorporatebrands. Entertainmentchannels releasetheirprogrammesusingbots. Generalknowledge questionsareansweredbybots. Botsflagchangesin Wikipediacrowd-sourcededitsofsensitivesubjects.
But nobody has yet figured out a way of nailing down fake news and driving it off social media. Truth will certainly come out and every fake story is usually rapidly debunked by credible sources. Indeed, there is now a cottage industry of tweeters who make it their business to track down and debunk fake news on social media. Pratik Sinha, for example, runs Alt News that is dedicated to dissecting and debunking fake news and identifying fake accounts. But no matter how often a story is debunked, the social media bubble ensures that the fake narrative also stays alive and some people continue to believe it.
Indeed, forgetaboutfakenews, nobodyhasfiguredoutafoolproofwaytoidentifyfakeaccountsforcertain. Thisisconsidered soimportantthattheUSDefenseAdvancedResearchProjects Agency(thesameagencythatfundedtheresearchthatledtothe internet) fundedthe“TwitterBotDetectionChallenge” inviting researcherstodevelopmethodstoidentify“influencebots”.
Thebattleofthebotswillcontinue. The2019electionandthe run-uptoitwillcertainlyseebotsbeingdeployedinlargenumbers onallsidesandthefreelancerswillpileonaswell. Sothenexttime youseeatweetbeingretweetedby20,000people, beawarethatit mightjustbeacoupleofgeekyteenagersdoingtheirthing.
ATwitter bot is an automated account. It can tweet, retweet, like, follow, unfollow, or direct message. Bots are used for broadcasting helpful information, auto generating content, et cetera. They are also used for spamming and spreading fake news and propaganda and trying to trend hashtags by increasing the volume of tweets. Sometimes they are hired by users who just wish to show a high follower count. Multiple bots – a panel – can be controlled by one person using software that’s easily available on the dark web. About 25 per cent of tweets are estimated to be created on bots, and Twitter submitted an estimate in 2016 that about 8.5 per cent of all accounts are bots. Most bots have few followers but clever manipulation can increase the follower count for bots by picking the right accounts to follow and tweeting on hot button subjects. While many bots perform legitimate tasks such as flagging Wikipedia edits or promoting brands, there is increasing concern about the way these are used to promote fake news and propaganda. Bots were extensively used during the Russia-Ukraine dispute and also during the 2016 Brexit and US presidential election campaigns. Most high-visibility Indian twitter accounts, including prominent politicians, have large numbers of bot followers. The 2014 Lok Sabha election saw a strong social media campaign crafted by the Bharatiya Janata Party that included the use of many bots. There are often determined efforts to trend hashtags on Indian twitter by using bots. There are accusations and counter-accusations by most political parties about the use of bots and there is apparent evidence in many cases. There have been many attempts to find fool-proof methods of flagging bots. Some of these are purely technical. Bots may tweet in inhumanly high volumes — sometimes they can churn out thousands of tweets in a day. Typically, bots panels will tweet and retweet the same contents in unison as well. But clever “bot managers” can also disguise themselves as legitimate accounts. The latest tool for bot detection is itself a bot called @Probabot that identifies and analyses political tweets using the Botometer, a tool developed by the Indiana University Network Science Institute and the Center for Complex Networks and Systems Research. This looks at the content of tweets, who is tagged, and how often, and who else is in their network.