Paul, On Fri, Nov 01, 2002 at 07:10:12PM +0000, Paul Makepeace wrote: > > I've been curious about this for some time too. > > I don't actually report a whole lot of spam since after SA there isn't > that much to report(!) but am considering revoking a lot more often. > > Without giving away corporate secrets, how does the revoke trust level > work in relation to the report? > > It strikes me that if a large number of people are auto-reporting, which > appears to be the case based on razor false positives I see, then there > is presumably a fairly strong correlation between these reporters. Now, > if only a handful of people are revoking, surely the numbers still > hugely favor the reporters? Are we pissing against the wind here? > Are those revokers simply hurting their own trust by going up against > the masses? > > I'd certainly feel a lot more comfortable about revoking if I knew* I > wasn't shooting myself in the foot here.
Trust assignments are based on the collective notion of correctness and trust doesn't necessarily correlate with volume and/or number of people who agree/disagree on a decision. People who report indiscriminately don't have a high trust rating, even if they are prolific (and part of what moderates them down is revokes from other trusted users). In general, you should do what _you_ think is correct. If you are careful about your assertions, other trusted reporters (other people who are careful about their reports) are likely to agree with you. Trust systems are emergent, so there's never a guaranteed result - you should voice your opinion irrespective of the outcome. In the larger scheme of things, individual trust ratings are not important, it doesn't really matter whether you have negative or positive trust, what matters is that confidence values generated by TeS are consistent so you can set the confidence threshold that suits your needs. One of the features I want to add to razor-users is an easy way to inspect the confidence values of spam, so people can set local cf values. > "What is a thousand times nothing much? Fast, then expel worms." > -- http://paulm.com/toys/surrealism/ Cool toy. :-) cheers, vipul. -- Vipul Ved Prakash | "The future is here, it's just not Software Design Artist | widely distributed yet." http://vipul.net/ | -- William Gibson ------------------------------------------------------- This sf.net email is sponsored by: See the NEW Palm Tungsten T handheld. Power & Color in a compact size! http://ads.sourceforge.net/cgi-bin/redirect.pl?palm0001en _______________________________________________ Razor-users mailing list [EMAIL PROTECTED] https://lists.sourceforge.net/lists/listinfo/razor-users
