4 thoughts on “The Perils of Akismet”

  1. Scoble and Andy are both basically presenting the idea that *any* false positive means the product is crap.

    What they haven’t considered is that Akismet is an organic concept that is constantly learning.

    If valid comments are getting caught up – then it needs to be taught some more.. Turning it off only means one less avenue for Akismet to learn.

    Kind of like biting the hand that feeds you, really. 🙂

  2. Brendan, there is also a hidden assumption in Andy’s criticism – that there aren’t false positives when you do it by hand. I can tell you from experience that before Akismet when I was going in and cleaning out 100 or more of these every morning, it is entirely possible that I deleted valid comments. One way or another, whether a human or system is dealing with it, the load of spam that I get is nearly impossible to handle without any FP. To use my favorite saying “You got to break a few eggs to make an omelet.” You are right, here at least there is a chance to improve on it.

  3. First of all Robert wasn’t really commenting on the accuracy of Akismet, and if you actually read my post, I think you would find that the majority was intended to help make it better.

    The one major criticism I have is that it doesn’t seem to learn who are “good” people rather than spammers.

    If you made a post on Roberts blog and it ended up in the spam bin, do you really want to add to his email queue?

    What effect can someone falsely accusing you have on your ability to post comments on other blogs?

    2 false positives that I know of. I like those odds

    You will probably find there were a few more that people didn’t bother emailing you about.

    I don’t have any problems with false positives.

    Some comments end up as a “maybe” but those are easily handled, and that is only because recently I added subscribe-to-comments and wanted to be extra careful.

Comments are closed.