If you join, you might relax a bit on rejecting spam, but saving it
for masschecks.Thats what I do... I do reject something, but not
everything I could.
That's probably not a good idea if it leads to unrepresentative spam.
In particular it may lead to botnet related tests being seriously
overscored, causing extra FPs for little benefit to the TP rate. This
seems to be already happening.
There's could be a similar problem with spamtrap spam too. For RBLs and
hashing it's OK to look at everything that goes to the address. SA
QA should only use the spam that would have made it through to SA.
That would tend to *under*score those rules for sites that have SA but
few or no MTA-time DNSBL checks, wouldn't it?
Yes, I know, "proper admin"; but such sites probably do exist - should
we punish them by underscoring those rules?
Okay. Now we need a consensus on this subtopic, right? I do not want to
do harm to the project or users of it.
The spam scores should be tuned for a well-configured server. Mail that
can be trivially rejected by greylisting, rbl, spf and similar tools isn't
all that interesting to use as a basis for the scores.
--
Kim Roar Foldøy Hauge
Event:Presse - The Gathering 2016
webmas...@samfunnet.no
Root@HC,HX,JH,LZ,OT,P,VH
