>Is this done to posion bayes before 'the real spam' arrives? >So far spamassassin did *not* react on those Mails, having >no clue that it might be spam (it has some trouble with german >anyway). > >Should I stop 'autolearnig' to avoid repercussions of this? > >Stucki (postmaster at math/inf/mi.fu-berlin.de) > > > It sounds to me like they are just trying to verify email addresses.
As far as bayes poisoning, it actually helps bayes to get emails like that.
Quite frankly, sounds more like an attempt at AWL poisoning than bayes poisoning.
Sending a bunch of poison messages before you send a real spam isn't going to be very effective at poisoning bayes, particularly since you don't know what they will do with the messages, but it could be used poison your AWL.
