The problem I have in assigning credibility to Twitter's position on bots is that they seem to have held multiple seemingly inconsistent positions (all paraphrased):
1. "Finger in the air estimate based on sampling", aka. "don't read too much into it"
They are completely consistent. They have always said 'no more than 5% according to their sampling' plus a long row of disclaimers and that they sampled based on things like activity and only from monetizeable users, neither of which can be tracked without Twitters internal data on the user.
1. "Finger in the air estimate based on sampling", aka. "don't read too much into it"
2. "Not more than 5%"
3. "Methodology can't be understood externally"