Simon Byrnand writes: >Whitelisting this list in spamassassin doesn't help prevent it from being >autolearnt - because the whitelist scores don't contribute to deciding if a >message triggers the autolearn thresholds or not.... > >Having said that, I havn't personally seen a problem with autolearn trying >to learn inappropriate messages from this list.... > >(I don't whitelist this list or have any special exclusions for it...)
No, me neither.
Occasionally (maybe one or two messages per hundred on this list) someone will post a copy of an actual spam which hits enough body tests to trigger and put it in my spam folder, but I probably didn't want to see that message anyway as I get enough spam of my own ;-)
But I've never seen conversation on this list come remotely near the threshold, even conversation discussing things that trigger a couple of rules... partly due to BAYES which tends to give a decent negative score to plain text conversation like this message.
The way I look at it, if the autolearn system is so vulnerable to poisoning by processing a list like this, then its too vulnerable to be used in any sitewide capacity. Fortunately thats not the case... ;-)
BTW the AWL helps.
I've mainly stayed away from AWL because a few people have suggested that sitewide enabling of AWL is a bad idea. Is that the case ?
By default will it try to use one common AWL database, or will each user have their own AWL database ? If both options are possible which is recommended ?
Regards, Simon
------------------------------------------------------- This SF.Net email sponsored by: Free pre-built ASP.NET sites including Data Reports, E-commerce, Portals, and Forums are available now. Download today and enter to win an XBOX or Visual Studio .NET. http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01 _______________________________________________ Spamassassin-talk mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/spamassassin-talk