Twitter and fake accounts, an old story unearthed by Musk

Dizzy, Elon Musk paused the takeover of Twitter on Friday. In a new twist, the billionaire said he wanted to wait to find out the exact number of fake accounts on the platform before buying it. “The deal is temporarily on hold pending details supporting the calculation that spam and fake accounts actually represent less than 5% of users,” the billionaire explained.

Does the boss of Tesla pretend to discover this element? According to the agency Reuters, the American company would have provided, Monday, a financial document which estimates at less than 5% the number of fraudulent users among the most “active” users of the social network. Subscribers who cannot, therefore, be monetized by the platform and targeted by advertisers without risk for them.

Panic over Twitter action

The figure, quoted by Musk in his tweet, is not a detail. It will be decisive for the financial future of Twitter. The platform, which has 229 million subscribers, is targeting $7.5 billion in revenue for next year. If it fails to attract enough advertisers to its “discussion thread” and develop its advertising offer, Elon Musk’s takeover proposal could overvalue the company and lead it straight to its loss.

Markets reacted strongly to Elon Musk’s announcement. At the opening of the markets on Friday, the title of Twitter had lost 17.7% of its premarket value, falling to 37.10 dollars of the action. A rate much lower than that proposed by Elon Musk to buy the company at 54.20 dollars per share in April. His “last and best offer”, in his words.

Read alsoWhat awaits Elon Musk with the takeover of Twitter

Almost 8% fraudsters in 2018

The presence of spam and automatic accounts on the social network is an open secret. The mass of these users who connect to Twitter from several accounts to manipulate the algorithms has already been the subject of several estimates. In May 2018, the teams of Yoel Roth, the site’s integrity manager, had established at 9.9 million the number of potentially fraudulent users in a blog post. According to these figures, fraud accounted for up to 8% of active accounts at the time out of 112 million subscribers.

The campaigns run by these accounts in a coordinated way have a name. Astroturfing, or the art of flooding the platform with false information. A common drift on Twitter. And well known, in Russia, where the practice of “bot factories” has even become a lucrative business. The possibility of raising a sentence, or a “hashtag”, reinforced by algorithms, is regularly pointed out in cases of harassment and influence campaigns. It was recently studied by sociologist David Chavalarias in his book, Toxic Data, how social networks manipulate our opinions (ed. Flammarion).

Elon Musk, a “troll” past white knight

Would Elon Musk pretend to find out? Not quite. The proposal to delete fake accounts is a hobbyhorse of the billionaire. The boss of Tesla has taken it into his head to eradicate spam. He assures him, he will remove the fake accounts from the platform, “even if it means going through it”, if he fails, he tweeted on April 21. Elon Musk knows something about it. Himself considered a “troll” (a malicious individual, in internet slang), his own page had already been preemptively suspended in 2018 after the company suspected an attempt to hack its account.

At the time, the anecdote seemed to have amused him greatly. Elon Musk himself took great advantage of these automatic accounts to manipulate Tesla shares, recalls the Washington Post. For the billionaire, the episode had highlighted the fact that the moderation rules exercised by the platform could prove to be arbitrary. A point that he has kept repeating in recent weeks to justify his takeover of Twitter and his plan to give back his freedom to the social network.

But tracking down fake accounts won’t be easy. Over the past two years, Facebook has devoted significant resources to this, including the creation of a high-performance algorithm, presented under the name of “Deep Entity Classification” (DEC), capable of learning to spot spam without human intervention and block their login attempts. For its part, Twitter has limited to 400 the number of accounts that a user will be able to follow during the same day, to prevent the platform’s traffic from being manipulated.

The social network, which employs more than 7,000 employees, has relied heavily on its reporting tools to identify fraudulent accounts and false information. The tool “BirdWatch“, put online in the United States, should thus allow its community to report misleading information itself, by filling in notes, on the model of the online encyclopedia, Wikipedia.

A collaborative model, to fight against fake news? Enough to seduce Elon Musk, who hates nothing so much as vertical control and censorship from above. The Billionaire clarified this Friday on Twitter be “still committed” to buying the social network. It is still necessary that he be sincere and not seize the question of false accounts as a pretext to have the operation canceled. And become, in turn, the biggest “troll” in the history of Twitter.