Bot/spam account analysis is just trying to do PR spin. Musk has no legal standing to use any of that to back out of the merger.
> The five-day trial is now scheduled to begin on October 17
So we will see about 2 more months of PR campaigns from Musk and Twitter about the case.
The claim is that Twitters SEC filings had bots at ~5%, and Musk made the investment based on that. From the leaked recordings of staff, everyone knew that the percentage was much higher.
No one is saying that 5% of Twitter's users is bots, except for when Musk is trying to put those words into Twitter's beak.
A company with 95% of its revenue coming from bot-clicks and a company with 5% of its revenue coming from bot-clicks might make the exact same amount of money, but one is substantially more screwed in the future.
"In its disclosures, Twitter claims to have nearly 238 million monetizable daily active users (“mDAU”) who participate on the platform, and tells its investors that this userbase metric is a bellwether for its ability to generate revenue and the “best way to measure [Twitter’s] success . . . .”
I think they are talking about this number being significant less than 238 million.
"They show that in early July fully one-third of visible accounts may have been false or spam accounts—resulting in a conservative floor of at least twice as many false or spam accounts as the 5% that Twitter discloses for the entire mDAU population."
They are talking about the same after bot removal 5%.
> The merger agreement contained no references to false or spam accounts, and Musk didn't ask Twitter for any information to verify the number of spam accounts before signing the merger deal.
> Additionally, our calculation of mDAU is not based on any standardized industry methodology and is not necessarily calculated in the same manner or comparable to similarly titled measures presented by other companies. Similarly, our measures of mDAU growth and engagement may differ from estimates published by third parties or from similarly titled metrics of our competitors due to differences in methodology.
Musk can't barge in and say "but MY calculations have a HIGHER number!" because Twitter already acknowledged this could be the case. Musk decided to buy anyways.
[1] https://www.sec.gov/Archives/edgar/data/1418091/000141809120...
Not to mention that some of Musk's publicly stated reasons for the purchase are to reduce the bot problem to improve Twitter's value.
1. Build a classification model to determine whether an account is a bot or not.
2. Measure the precision/recall curve of that model, and pick a threshold where the precision would be >95% (i.e. >95% of what the model labels as "humans" are really humans).
(Higher recall leads to lower precision and vice versa. Higher recall means a higher mDAU, so it's a balancing act between a high number versus a high confidence in that number.)
3. Let the model loose on all accounts.
4. The number the model tags as humans is the "mDAU", though <5% of these might be false positives.
While this model takes a lot of training data to build, verifying its continued precision/recall characteristics can be completed with very little ongoing sampling - easily handled by the "100 a day" figure Musk has claimed.
They randomly sample 100 accounts per day and have human raters decide if they're bots or not. This is the right way to validate that your fancy ML algorithms are working well.