Elon Musk vs Twitter: the weird number at the heart of the drama

Listening to the lawyers representing Elon Musk and Twitter argue over which documents need to be provided as part of legal discovery today, I began to wonder if it is possible to measure anything at all.

The arguments centered on a wonky count of user numbers: mDAUs, or “monetizable daily active users.” It’s a nonstandard measure — most social media companies use daily or monthly active users — that focuses on which users can see ads. But Twitter is a nonstandard social media company!

Twitter is suing Musk for abandoning a deal to buy the company for $44 billion. Part of Musk’s argument is that he was misled about the amount of spam and bots on the platform and that, therefore, the deal can’t be done. In today’s hearing, Musk’s lawyers claimed Twitter was cherry-picking documents the lawyers had requested on mDAUs. Alex Spiro, Musk’s counsel, told the court that Musk wants to check all Twitter accounts.

“We don’t just have to take their word for it,” Spiro said to the court. “We are allowed to question their process.” Chancellor Kathaleen McCormick, who is overseeing the case in Delaware’s Chancery Court, said she would take the matter under advisement before adjourning the hearing.

I don’t know if mDAU is actually a good metric. But this hearing suggests that revelations from a Twitter whistleblower will shape Musk’s case, at least in terms of guiding Musk’s discovery.

Twitter’s former security head Peiter “Mudge” Zatko — among other allegations — said Twitter’s method for measuring bots, fake accounts, or spam is wrong. Zatko’s complaint suggested executives were offered bonuses of up to $10 million to boost user counts. And so, that’s what executives did instead of removing bots.

It’s also true that Twitter admitted in April that it had overcounted for three years. This was not even the first time it had happened! The SEC is now asking questions!

How did mDAUs come to exist in the first place? Well, lots of us look at tweets in places that aren’t Twitter. One instance is on this here website, where Twitter can’t display ads because they’re our ads, baby.

It gets worse! I don’t use Twitter’s website on my computer as my primary mode of viewing Twitter. I use Tweetbot, which is not made by Twitter and which does not display ads. On mobile, Twitter’s app does show me ads, but I mostly don’t use Twitter on my phone, and honestly, I’m considering switching to Tweetbot there, too. So I am certainly a daily active user, but I’m seeing a lot fewer ads than someone who’s using Twitter’s website. How do you account for me?

That question is at the center of the arguments. The way mDAU gets captured is by taking 9,000 Twitter accounts at random and reviewing them manually. They are then flagged as spam or bots or used to estimate how users see ads. Both private and public data are used to make these calls. But because some data isn’t retained, it’s hard to check Twitter’s work. Twitter’s lawyer said that a “brute force” manual review might be possible, and it would take about a week. Musk wants more data than that.

Musk’s lawyer argued that mDAU was made up in order to juice Twitter’s growth numbers — not a novel critique of mDAU, by the way. He also suggested that tying compensation to mDAU growth was a bad incentive. Twitter says that mDAU is the preferred measurement; Musk says that using it is essentially fraud.

There was one chink in Musk’s lawyers’ arguments, and it was an Elon Musk tweet. “If Twitter simply provides their method of sampling 100 accounts and how they’re confirmed to be real, the deal should proceed on original terms,” he wrote. Twitter’s lawyers suggested that Twitter has, indeed, provided that method, its lawyer said in court.