[Spambayes] SpamBayes feedback

[skip at pobox.com]

    Scott> [-] I found that SpamBayes to be much less accurate "out of the
    Scott> box" than other free anti-spam products I've tested. It's my
    Scott> opinion that SpamBayes took much longer to train than I expected
    Scott> in order to reach a level of usefulness. I have fed SpamBayes
    Scott> well over 1000 spam messages. 

How many hams have you trained on?  What sort of settings do you have for
the thresholds?  Out-of-the-box the spam and ham thresholds are pretty
conservative.

Skip