From: Vipul V. P. <ma...@vi...> - 2002-01-15 21:55:41
|
On Mon, Jan 14, 2002 at 02:51:46PM -0600, Shane Williams wrote: > -----BEGIN PGP SIGNED MESSAGE----- > > On Sun, 13 Jan 2002, Vipul Ved Prakash wrote: > > > The bigger drawback of client side thresholds is > > that people who set their threshold to a high number, say 5, will only > > benefit from the reports of other people who set their threshold to 5 or > > 5+ (because people with lower thresholds will filter out the spam and not > > report it.) > > I don't think this logic is necessarily correct. I, for one, report > spam to razor regardless of its current razor status, based on quick > inspection. I am not sure if many people do that; I, for example, don't. Troll boxes are the only sources that can (and should) submit irrespective of the Razor status. > Further, I don't think the count has to be a simple threshold. I'm using > spamassasin in conjunction with razor, so spamassassin (not razor) > determines the threshold above which an email is spam. In such a case, I > could have SA score multiple razor hits higher than single razor hits. Submission count isn't a correct measure of spamness. But it's probably a good measure of "probability of spamness". I am going to give this more thought. best, vipul. -- Vipul Ved Prakash | "I almost died, but I made it, so I'm not so Software Design Artist | serious about formal-wear anymore." http://vipul.net/ | -- Gene Boggs |