If Twatter admitted that it only samples 100 accounts in determining the percentage of bots, that's basically an admission of fraud itself. The only reason to use such a tiny sample is to bias the results. It's almost certainly an automated process, so there's no real reason not to sample more accounts anyway. That said I expect if it hits the courts it will be with an ultra-far left shitlib kike judge who will pretend that deliberately kneecapping your own bot detection is standard procedure and that if the real number is much greater than 5% that's fine because shareholders and advertisers shouldn't expect to get the service they paid for.