It's been interesting to watch this story unfold. Twitter is a broadcast system and therefore has many (useful [and not useful]) bots. These aren't users, but not all of them are spam. It's a simple counter-example to show that simple user (and engagement) counts will almost certainly be misleading. Twitter is in a big-time catch 22 with reporting here. It'll take skill to report a qualitative aspect like engagement with a quantitative metric (like they're expected to).
I've advertised on Twitter before and their engagement metrics are curious and a bit dishonest.<p>They measure any click on a tweet's body (not a hyperlink) as an engagement; where as an advertiser I define engagement as a favorite, retweet or reply.