On Tue, Jun 13, 2006 at 10:08:52AM +0200, Martin Wuertele wrote: > * Steve Langasek <vorlon@debian.org> [2006-06-13 00:31]: > I strongly disagree with censoring Mails coming from hosts that _also_ > host tor nodes. The host several DDs including me use > (asteria.debian.or.at) is a tor (tough not exit) node hosting also a > mixmaster anonymous remailer. > I feel offended and hurt that you want to censor me because I use a host > where we provide a imo useful service for others (there are many valid > reasons to use tor and mixmaster). I don't want to censor anyone; I want to protect the mailing list from abuse. I feel offended that people who *aren't* trolls are referring to this as "censorship" at all when the goal is clearly to prevent disruptive behavior, *not* to suppress information or viewpoints. In any case, the fact that the machine in question is not a tor exit node should mean there is no reason to filter on that machine's IP, correct? And the filter could be further refined to stop only mails that both include that IP and a sender address belonging to a webmail service, for instance? > Trolls like the ones posting to the lists should be treated like spam: > ignored and best filtered by bogofilter and/or spamassassin et al. Yeah, great theory, except trolls *aren't* treated like spam: nobody bats an eye when the 2000th spam of the day makes it through their filters, and nobody takes their suggestions to make pills fast with penny incest stocks seriously, but when a troll shows up, people mistake it for a message of value, engage it, and lose valuable time to it -- and this pattern repeats itself as long as there are trolls and new people to fall for those trolls. Heck, we have long-standing *developers*, with no excuse for not knowing better, who think it's a *good idea* to try to respond to trolls on mailing lists! So how do you minimize the damage done by trolls on a mailing list which is expected to by design be a gateway into the Debian community, where there will always be an influx of newbies getting drawn in -- or driven away -- by trolls of this nature? It has been suggested that some central bayesian troll-filtering would be beneficial here, and I agree it's worth trying; but AFAIK pure bayesian filtering still is not as effective as bayesian+blacklists, so I think a rational discussion of appropriate IP blacklisting is still in order. Cheers, -- Steve Langasek Give me a lever long enough and a Free OS Debian Developer to set it on, and I can move the world. vorlon@debian.org http://www.debian.org/
Attachment:
signature.asc
Description: Digital signature