Spamcop
in [UK Linux]
|
From: Andrzej Adam Filip on 23 Nov 2006 06:26 MJ Ray <mjr(a)phonecoop.coop> writes: > Andrzej Adam Filip <anfi(a)onet.eu> wrote: >> Spamcop.net can be used for three purposes: >> a) "binary blocking" (blocking based on spamcop.net listing *ONLY*) >> you and I "discourage+" it :-) >> b) as part of "cumulative" blocking score >> spamassassin uses it for this purpose with the following scores in >> version 3.1.7: >> score RCVD_IN_BL_SPAMCOP_NET 0 1.332 0 1.558 >> c) as a tool to ease sending abuse reports >> [ *every* spam received constitutes solicitation for abuse@* report :-) ] >> >> *"a" is brave/stupid but "b" and "c" do make sense*. > > I disagree. b is just adding a random number to your spamassassin > score The score is "adjusted" by spamassassin's folks at every release. As I understand it is "automatically fine tuned" between releases. Do you challenge their "score assigning" wisdom? > and c is better done with whois.abuse.net directly - spamcop abuse reports > are pretty hopeless IMO and often seem to go to the wrong postmaster. I report 100+ spams a day, every "helper" counts :-) I use *reachable* free email accounts in my usenet posts and I spent 2-3 seconds per spam for personal verification of spamassassin's "classified as spam". Everything else is fully automated. > [...] -- [pl2en: Andrew] Andrzej Adam Filip : anfi(a)priv.onet.pl : anfi(a)xl.wp.pl
From: MJ Ray on 24 Nov 2006 07:11 Andrzej Adam Filip <anfi(a)onet.eu> > The score is "adjusted" by spamassassin's folks at every release. > As I understand it is "automatically fine tuned" between releases. > > Do you challenge their "score assigning" wisdom? If they assign any multiplier above 0 to spamcop, yep. > > and c is better done with whois.abuse.net directly - spamcop abuse reports > > are pretty hopeless IMO and often seem to go to the wrong postmaster. > > I report 100+ spams a day, every "helper" counts :-) If one reports through spamcop, it is almost useless, for the reasons previously given. Use whois.abuse.net directly instead. Regards, -- MJR/slef
From: Nix on 24 Nov 2006 18:13
On 24 Nov 2006, MJ Ray said: > Andrzej Adam Filip <anfi(a)onet.eu> >> The score is "adjusted" by spamassassin's folks at every release. >> As I understand it is "automatically fine tuned" between releases. >> >> Do you challenge their "score assigning" wisdom? > > If they assign any multiplier above 0 to spamcop, yep. Actually, if anything, SpamCop, like Bayes, probably got an excessively low score in 3.1.x and 3.0.x (Bayes was hand-adjusted up, but nobody touched SpamCop). That's because when the training mass-checks are done, the text-based rules haven't been picked up by many spammers, so will hit a lot of spam; the scores for other rules are adjusted down as a consequence. Then SA is released, the spammers adapt to the fixed-text rules, and suddenly the scores for the other rules are too low :( Current statistics for RCVD_IN_BL_SPAMCOP_NET as of the most recent network mass-check are available at <http://ruleqa.spamassassin.org/?daterev=20061118-r476474-n&rule=%2FSPAMCOP&srcpath=&s_detail=on&g=Change> MSECS SPAM% HAM% S/O% RANK SCORE NAME 0.00000 37.5185 1.3859 0.964 0.62 0.00 RCVD_IN_BL_SPAMCOP_NET (The score is 0 because the ruleqa runs don't try to assign scores to network or Bayes rules.) A rank of 0.62 isn't very good: nearly all the truly effective rules are above it. (It seems high, but a good few rules hit nothing, and some effectively non-forgeable rules, especially header rules, are *intended* to hit ham and thus will have a very low rank and receive a negative score.) I suspect that at this point the spamcop rules are most useful in meta rules, but we'd need to do a network scoring run to be sure. >> > and c is better done with whois.abuse.net directly - spamcop abuse reports >> > are pretty hopeless IMO and often seem to go to the wrong postmaster. >> >> I report 100+ spams a day, every "helper" counts :-) > > If one reports through spamcop, it is almost useless, for the reasons > previously given. Use whois.abuse.net directly instead. I concur. Plus it's vastly annoying to report any significant volume of spam through SpamCop, to the point that I don't bother at all. -- `The main high-level difference between Emacs and (say) UNIX, Windows, or BeOS... is that Emacs boots quicker.' --- PdS |