> This really is highly fragile answer. The VERP RFC, for one example,
>
> ...
>
> I would just warn everyone to be aware of the likelihood
> of false positives, especially from well-behaved, even
> distinctly anti-spam, sources.
Hi Sandy
Yes you're right: There can be legitime messages containing more then 3
digits in the from adress.
But I'm sure you know that no spam-test is a 100% indicator of spam.
We've implemented every single test based on a detailed statistical
research on our server. We've seen that this two test are very good
indicators of spam. Less then 2% of positive SenderwithCode-tests are
false positives (4% for SenderwithCodeMaybe).
Can you say this about SPAMCOP, NOABUSE, REVDNS an Co?
I consider the second test checking for unreadable sequences a lot more
"dangerous" For this we gave them only a small weight: 2 and 1 on a
20-point-weighting system. Probably 2 points are already to high.
For all SPAMCHK-Users:
We consider to add the following consonant-sequences to the
FreqConsSequences-parameter:
sts,stm,adm,str,wsl
This should cover from-adresses containing PoSTMaster, ADMinistrator,
NeWSLetter, ...
Markus
---
[This E-mail was scanned for viruses by Declude Virus (http://www.declude.com)]
---
This E-mail came from the Declude.JunkMail mailing list. To
unsubscribe, just send an E-mail to [EMAIL PROTECTED], and
type "unsubscribe Declude.JunkMail". The archives can be found
at http://www.mail-archive.com.
