Being generally happy with Akismet spam protection plugin for WordPress 2.0, I’m still kept from total happyness by a fly in the ointment.
Before enabling Akismet I used to use a built-in setting in WordPress called “Comment Moderation” (“Hold a comment in the queue if it contains more than N links.”) Of course, “a common characteristic of comment spam is a large number of hyperlinks” is a pretty dubious spam filter and I did get hundreds of spams in short periods of time with it enabled. But surprisingly or not, now I constantly see about 90% of all spam caught by Akismet being that kind of spam with multiple links (I used “2 links” as a trigger for holding comments). That is, Akismet takes bulk of the spam caught by WordPress and misappropriates it to its own portfolio!
So I’m wondering if Akismet really benefits from breaking this standard feature. Surely they could reduce load on their servers by quite a few-fold having they left the feature alone. Moreover, WordPress webmasters are usually up on with this sort of things and are but irritated with the crazy spam counters in Akismet tab.
I realize Akismet is trying to enjoy powerful promotion in this manner, but doesn’t it go too far by depriving users of the right to judge its effectiveness adequately? Really, the spam count is absolutely irrelevant this way; it’s like telling the size of computer network by the number of copper nuclei constituting all the wires of the network.
Bayesian filtering is nothing new and Akismet may easily loose the competition to a more honest operator. Even more so in the situation where operational data ages quickly and new players catch up promptly with the veterans.