...... Original Message .......
On Thu, 19 May 2005 07:01:49 +0200 Frank Ellermann
<nobody(_at_)xyzzy(_dot_)claranet(_dot_)de> wrote:
Scott Kitterman wrote:
The nightmare reputation system exists.
http://spamassassin.apache.org/tests_3_0_x.html
My concept of a reputation system was something like SIQ, input
domain and/or IP, output "value". SA scores could still fit,
but it's done locally without third parties like senderbase or
IADB.
Only a confusion of terminology, I'd say "scoring system" for
SA, not "reputation system".
OK. That's probably better.
The patch to add NEUTRAL rules has already been accepted:
http://bugzilla.spamassassin.org/show_bug.cgi?id=3616
The default score for a PASS is AFAIK already the minimal value
supported by SA, +/-0.001 or whatever it was, sign irrelevant.
My prediction is that once SA 3.1 is released,
INTENTIONALNEUTRAL will be a tough place to be.
Maybe you could propose something better ? I have no idea how
they handle such issues, only a vague impression based on some
X-Posts on the SURBL list, apparently they are sensible folks
(just like the SPF community, so don't hold your breath... ;-)
Well they are pretty strong on the idea that whatever the statistics from
their corpus of ham/spam says the score should be is what the score should
be. This was discussed here and in the comments to the bug. IIRC, the
patch author is/was on this list.
Problem for me is that as the little guy my ?include: ... gets totally
swamped by stuff that falls off the end of Aol and such's record.
One point should be clear: PASS never meant "no spam", a PASS
without a white or black list is pretty useless. If they now
plan to handle NEUTRAL like PASS it has the same restrictions:
It's pretty useless without corresponding black or white lists.
Bye, Frank
It's a low fidelity reputation system where each SPF result gets a
collective reputation. IMO, the only things SA should score on are Fail
and Softfail.
Scott K