My biggest problem with Akismet is its inability to deal with false positives. Granted, you can always mark a incorrectly labeled comment as not being spam and use the form on the Akismet website but both seem to be black holes to us.

Regarding the open source-ness of TPAS, it feels a bit like we’re getting into the territory where defintions are less useful and people make their evaluations on a gut level or on an emotional basis. I say that not as a criticism, but just as an observation. Not speaking for Six Apart, but just for myself, it feels like the logic or rules in an antispam service are the data, and we don’t share all of our data. That doesn’t mean the engine that runs them isn’t open source, just that we keep some data private. This is analogous to how, for example, SpamAssassin treats its rulesets, which many users customize, tweak, or invent without sharing.

We’re definitely open to suggestions of how to make these ideas clearer, or how we can improve what we’re doing, but if the first result of TypePad AntiSpam is that it encourages other services to become more free (as in both beer and speech), that seems like it can only be a good thing. Oddly, I haven’t heard more people arguing for that, or at least I’ve been too busy to find those conversations.