On Tue, 05 Dec 2006 09:51:06 -0800, Marc Perkel <[EMAIL PROTECTED]> wrote:
> > >Yet Another Ninja wrote: >> >> Just found a few ... sent directly from DULs. >> (there went my theory...) >> >> :-( >> >> > >I have a theory that spammers are either doing some sort of probe or >sending out nonspam so that ther headers are learned by bayes as good. I >think it's either probes or bayes poison. I can't see 5 digits being of any use in bayes poisoning, with that little data there's not much point feeding it to bayes anyway. As for getting their headers known, surely that is counter-productive? If we (and spamhaus, spamcop etc etc etc) know the headers, the spam are less likely to get through after the probe, The time to send the spam would have been initially? Maybe, like me last night, they had a couple too many beers and fired off their mails without much thought? Nigel
