RE: An idea for a bot to help combat the comment spam
There's several potential methods to prevent false positives from falling into the blacklist.
Yes, pattern searching is a good. But the more exception filters you add the less likely it is a real human falls into that category (as in someone with lack of originality and compulsive obsession to "leave a comment).
I would add a filter based on the post length→ minimum reading time factor, flagging those bots that read 1500 word articles in 2 minutes (I HATE those).
Also the potential to find markov chains is imperative, it wont be long until comment bots start implementing that (or a similar system), being a step ahead can never hurt.
A third option would be to "legalize" comment bots:
Release a bot easily usable by anyone that offer certain features that make it a better choice over others, and in it, include a command able to "please, don't comment on my posts".
And I forgot to add:
Selfvotes (such as the one above), all comment bots do that!
Great points. Should be quite easy to do a bot that finds the people that comment too quick. It might be my best bet for my first bot before I try to do the more complicated one.