Politics on this echo?

OK, in summary, Spamassassin (aka SA) is a filter, written in Perl, that scores messages for spamminess and can be added into almost any pipeline. It reads messages from stdin, examines (both headers and body to calculates a spaminess score and writes the message on stdout, adding headers containing the score and a list of the rules that the message triggered. SA rules contain a regular expression and a score - the score is added to the overall message spamminess score if the regular expression matched anything in the message. Thats all SA does: calculates a score (by convention less than 5 is ham, over 5 is spam), adds some spam-related headers to vte message ande passes it on.

I run it as part of a pipeline that passes messages read via getmail through SA and a spam discriminator and delivers non-spam to a local MTA for delivery on my local LAN. The discriminator quarantines any messages with a spam score of over 5. Using the full SA may be a bit OTT for what you're trying to do, though loking at its documentation may give you some goo ideas, but writing some sort of filter in Perl that uses a configurable collection of regexes that trigger on spam indicators should be doable.

HTH

Generally a good rule of thumb is not to blacklist groups, but to limit crossposts. Something posted to more than maybe two groups has a much greater likelihood to be spam/trolling.

It won't help the ongoing fires when the discussion splinters off into parts only posted to one group, but should block the initial spark.

Theo

On 10 May 2021 at 12:00p, Theo pondered and said... Th> Generally a good rule of thumb is not to blacklist groups, but to limit Th> crossposts. Something posted to more than maybe two groups has a much Th> greater likelihood to be spam/trolling.

Thanks Theo :) Yep I have attempted to do this and targeted a number of political newsgroups. I guess time will tell but at least we can try.

On 10 May 2021 at 10:53a, Martin Gregorie pondered and said... MG> Using the full SA may be a bit OTT for what you're trying to do, though MG> loking at its documentation may give you some goo ideas, but writing some MG> sort of filter in Perl that uses a configurable collection of regexes MG> that trigger on spam indicators should be doable. MG>

MG> HTH

Thanks for taking the time to walk me through SA. It's likely outside my technical ability to implement but good to know the background etc. I've started with pulling some levers I have available in the gating tools in that hope that may mitigate the spam x-post stuff in the newsgroup passing over to the Fido side.

Anywhoo happy Tuesday from over here in this side of the world.