One month of Negative Turing Test
One of Negative Turing Test’s most recent features is a counter of how many spams it eats. I added this feature last month, and made a note to reveal today what it got up to.
The number of spams blocked by NTT from 2007-02-12 to 2007-03-12 is:
6,220
I should probably get around to making it delete those…
March 12th, 2007 at 19:59:02
My question is, how many false positives have occurred, and how many false negatives? I mean, it even took me a little while to work out what I was actually meant to do. (I realise my question involves a double negative, but that’s half the fun)
March 12th, 2007 at 22:13:40
There have only been a couple false positives that I know of, all caused by people forgetting to answer the spam-prevention question. As for false negatives, just one—there’s no automated defense against a spammer who is either doing it himself or hiring a bunch of cheap labor to do it, but fortunately spambots are still cheap and effective enough that human-delivered spam isn’t yet making a major comeback.