One thing about captcha's is that bots, as Chris points out, can be adapted to them.
Here is where Jakes method works well. He's using heuristics to limit the amount of spam he gets, then uses human monitoring to filter out the rest. It's better than beysean filtering, but still requires manual work for the admin.
I look at the 'work on the user' issue as a minimal thing. We already enter our name, email and address and type long winded comment... a few more digits is hardly an inconvenience, though I can't disagree with the principal of not putting a barrier between the user and posting a comment.
Back to the adaptable bot scenario though, it seems anything you do - hidden fields, negative captchas, regular captchas, has to be made sufficiently random, or easily modifiable, to make it as trivial for the admin to adjust as it is for spammy to tweak or adapt his bot. Eventually, one party will give up... and bloggers are a stubborn lot!
One thing about captcha's is that bots, as Chris points out, can be adapted to them.
Here is where Jakes method works well. He's using heuristics to limit the amount of spam he gets, then uses human monitoring to filter out the rest. It's better than beysean filtering, but still requires manual work for the admin.
I look at the 'work on the user' issue as a minimal thing. We already enter our name, email and address and type long winded comment... a few more digits is hardly an inconvenience, though I can't disagree with the principal of not putting a barrier between the user and posting a comment.
Back to the adaptable bot scenario though, it seems anything you do - hidden fields, negative captchas, regular captchas, has to be made sufficiently random, or easily modifiable, to make it as trivial for the admin to adjust as it is for spammy to tweak or adapt his bot. Eventually, one party will give up... and bloggers are a stubborn lot!