But why would the analysis care about bots that are viewing? They aren't creating disinformation by viewing (maybe by "liking" though). Bots-as-followers are benign. The "bot problem" is bots being used to post disinfo at scale to cram the entire space with propaganda, that is: to quote Steve Banon "flood the zone with shit". Which is what bots and troll farms have done successfully over the past 7 years.
Engagement signals are what drive ‘the algorithm’.
Bots that create fake engagement signals (fake clicks, fake media views, fake DM engagement, as well as fake replies and likes) can manipulate what content gets promoted onto real users’ feeds.
From a business point of view, the problem is that if people can manipulate and promote messages on Twitter by using fake engagement, they are not paying Twitter to promote tweets. As well, I guess, as undermining regular users’ confidence that the things that show up in ‘trending topics’ really are ‘trending’.
But bots posting propaganda - if they’re any good at it - might, weirdly be a good thing for Twitter’s numbers, because they create controversial posts that drive engagement into which Twitter can sell advertising.
Looking at the firehose tells you very little about any of that though. There could be a million bots who post lorem ipsum tweets at a rate of hundreds a second and they would all show up in the firehose, but if twitter’s algorithm means none of those tweets are ever shown to a human, does it matter?
Musk’s whole schtick with the bot numbers has just been a thorough demonstration that he has barely even thought about Twitter’s business or what a social network even is.
Didn't he get access to the firehose, which is, AFAIK, -ALL- the data? (not the 50GB he tweeted about but supposedly a 2TB/day stream?)