Theo Van Dinter <felic...@apache.org> writes: > On Thu, Jan 08, 2009 at 04:37:37PM +0100, Karsten Bräckelmann wrote: >> > It appears to me that the HABEAS rules are hitting only a very tiny >> > fraction of mail, many of the nightly mass-checks don't have a hit >> > at all (or is it that those checks don't contain any network >> > checks?). The aggregated view shows no hits at all for these rules. >> >> Network tests are done once a week, not daily. > > Just to share some data, my last weekly run shows: > > 0.084 0.0000 1.2638 0.000 0.58 0.00 HABEAS_ACCREDITED_SOI > 0.010 0.0000 0.1484 0.000 0.47 0.00 HABEAS_ACCREDITED_COI > 0.000 0.0000 0.0000 0.500 0.44 0.00 HABEAS_CHECKED > > and generating stats from the last weekly run results from everyone: > > 0.039 0.0001 0.6879 0.000 0.62 0.00 HABEAS_ACCREDITED_SOI > 0.003 0.0000 0.0573 0.000 0.51 0.00 HABEAS_ACCREDITED_COI > 0.000 0.0000 0.0000 0.500 0.49 0.00 HABEAS_CHECKED > > There's a handful of spam hits for a couple of people, so it's not > clear if that's misfiling or an abusive sender. But these results are > pretty good IMO.
I searched all of my mail on two machines for HABEAS (I expire a lot of it, though), and came up with a few messages in a spam complaints folder (such as the one that I started this thread with), discussion on this list, private messages to me from people saying they find that these days habeas accredits spammers, and one other message - a misdirected political rant from a friend-of-a-friend-.... that was about gitmo, not spam. So I wonder if the reality is that most of the places habeas accredits are legitimate newsletter senders some of them are spammers habeas does not respond to complaints in any reasonable/useful way people like me who don't get the kind of junky newsletters that need accreditation to be deliverable only get HABEAS-marked mail that is spam and thus I have a different local reality than the overall statistics -- a handful of messages received that are accredited, all of them spam, and no response to complaints. This seems to be the experience of a number of others. So, my point was basically that even if in some statistical sense the rule is valid, is it reasonable to let a for-profit third party sell -4 spamassassin points unless we are convinced that they are very diligent and respond so quickly and appropropriately to complaints that problems are very rare? (Obviously I think habeas does not meet the above test.) If this were -0.2 I wouldn't be so cranky.
pgpHa5UQwFjjy.pgp
Description: PGP signature