A former Twitter engineer has a novel idea for Elon Musk to test how many fake users are on the site: secretly create his own army of bots.
Musk, who’s currently locked into a $44 billion agreement to buy Twitter, has suggested that as many as 20% of Twitter accounts are fake or spam — four times as many bots as Twitter claims. The Tesla CEO has said his takeover is “on hold” as he awaits proof from Twitter that “spam/fake accounts do indeed represent less than 5% of users.”
Twitter has insisted in regulatory filings it has strong spam detection systems, that the 5% figure is accurate and it plans to go through with the deal.
Now, an ex-Twitter engineer says the best way for Musk to theoretically “de-legitimize Twitter’s bot count” would be for him to create his own spam army.
“I would try to spin-up a bot farm,” the unnamed engineer, who worked for Twitter’s mobile advertising business, said in a recent interview circulated in a note from investment bank Jefferies. “Once spun-up, I would then create millions of fake accounts, and then disclose this to the Twitter board and the world.”
“If such a strategy was successful, this would/could legitimize Musk’s statements regarding bot count,” the ex-employee added in the note. “It would be a lot cheaper than paying the $1bn termination fee if I was Musk!”
In order to prevent Twitter from taking action, Musk would have to create the bot farm in secret, the ex-employee said.
“He just needs to quietly fund a team to do so, and I would via crypto to cover my tracks,” said the former engineer, who worked for Twitter’s mobile adverting business, which was called MoPub and was sold off in 2021.
The note was sent out to Jefferies clients on Friday and was exclusively obtained by The Post. Twitter did not immediately respond to requests for comment, and Jefferies declined comment.
Elsewhere in the Jefferies note, the ex-Twitter employee threw cold water on the idea that a huge proportion of Twitter accounts are bots.
“I would be very surprised to learn that the 5% number was meaningfully different and the company KNEW it was meaningfully different,” the ex-employee said. “There are a bunch of very smart people working on this and my inclination is that they mostly got this right.”
“Are there things that they may not have caught, and could someone point and say the number is really 8-10%?” the ex-employee added. “Sure. But that seems within a reasonable margin of error in my view. Could it be 25-50%? I really doubt it.”
The former employee also said Twitter uses various systems to detect bots, including monitoring whether users are scrolling or liking too many tweets at an unreasonably fast speed.
Twitter’s advertisers also use third-party analysis to verify Twitter’s claims and would’ve likely caught any widespread proliferation of bots, the ex-employee said.
“Advertisers hire technically savvy employees that are aware/can evaluate the impact of bots on their ad campaigns,” the engineer said. “It would be very difficult to ‘hide’ a meaningfully inaccurate 5% bot count at Twitter.”
