Crypto News: Musk's Twitter bot problem

Hey all, it's Kurt Wagner in San Francisco. Elon Musk's battle against Twitter bots continues. But first…

Today's must-reads:

• Crypto believers have had their confidence rocked by the Voyager collapse
• GameStop declined in late trading after it fired its chief financial officer amid reports of job cuts
• Ramesh "Sunny" Balwani, the former president and ex-boyfriend of founder Elizabeth Holmes, was found guilty of fraud and other charges in the collapse of health startup Theranos

Elon Musk's plan to buy Twitter is reportedly in jeopardy. Again.

The social networking company held a media call Thursday morning to talk about how it measures spam bots on the service, an issue that has become a significant factor in Musk's planned $44 billion acquisition. Twitter says spam bots are a small part of the company's total user base – less than 5% – but Musk believes the number is much higher.

It's a significant disagreement, because Musk has said repeatedly that he might walk away from the purchase if Twitter doesn't do more to prove that bots are indeed the small problem that the company claims. On Thursday afternoon, that threat was issued yet again when the Washington Post reported that the deal was in jeopardy because Musk's team didn't feel Twitter's bot numbers were verifiable.

It was a dramatic twist to a day that started with a media call where Twitter executives explained how the company calculates the number of bot accounts on its service. The company repeated a few things it has said before:

The number of spam bots on Twitter is well below the 5% threshold it mentions in filings.
It calculates these bots using human review of thousands of random accounts each quarter.
It uses some personal account data – IP addresses and phone numbers – to help determine if accounts are run by real people.

Executives also reiterated that they don't believe anybody can accurately determine the number of bots on Twitter without that personal account data, something that Chief Executive Officer Parag Agrawal tweeted weeks ago. That data, Twitter execs said, is not shared with outsiders for privacy reasons.

Musk seems to agree, according to the Post, which reported that "Musk's camp concluded that Twitter's figures on spam accounts are not verifiable." For weeks now, Musk has had access to Twitter's "firehose" of tweets, presumably using that data to come to his own estimate about the number of bots of Twitter. But the Twitter firehose includes data about public tweets, not the kind of account-specific data Twitter says you'd need to determine which accounts are run by real people.

So why did Twitter go through the trouble of sharing a bunch of details about bot counting that it had already shared before?

My guess is that Twitter wants to pre-empt Musk's next move, which was foreshadowed by the Washington Post story. Given that it's widely believed that Musk wants to renegotiate his deal, it seems likely his estimate of Twitter bots – whenever he shares it – will be much higher than the company's. This would give Musk ammunition to argue that he paid Twitter for an audience of users that was much smaller than the company claimed.

But Twitter said again it doesn't believe you can count bots accurately without the kind of private data that only the company has. In other words, Twitter already laid out its case for why any future bot estimates from Musk will be wrong.

So yes, the Twitter-Musk deal may indeed be in jeopardy. But it's been that way for weeks. Thursday was just a reminder.—Kurt Wagner

The US Supreme Court's decision to overturn Roe v. Wade has put big technology companies on the spot. What will they do when law enforcement wants personal data that may show someone seeking an abortion.