[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: lists.d.o Spam (was: Marking BTS spam)



On Thu, Mar 02, 2006 at 07:33:04AM -0600, Cord Beermann wrote:
> it's not ok to bounce us spam from years ago, this address is another
> feedback-loop to enhace our filters, it's useless to get old[1] spam
> there. 

Well, I thought it would be useful to 

a) have spam mails from the mailing list archive
b) enhance the filters

The first idea was not supported in your post (but kind of implied, since
this discussion was based off the web button thingy), the second idea
came from this:

On Fri, Feb 24, 2006 at 07:03:31AM -0600, Cord Beermann wrote:
> The reported mails will be used to enhance our Spamassassin and
> procmail-filters. The mails will be stored non-public.

I bounced a month's worth of spam with that idea in mind. I don't see what
advantage is there in reporting only 2 day's worth of spam. It's not like we
are going to report this to the ISPs (like Spamcop is), are we?

In any case, I will refrain from bouncing "old" spam from now on.

> (and for those that expect immediate healing from this: the input
> won't be piped into some filters for learning, as it would be to
> easy to poison it then.)

I understand that this mechanism could be abused by casual bystanders, but
you could maybe set "trusted moderators" for lists in the sense that *their*
reports (conveniently GPG signed) could be used to remove spam in a more
automatic fashion?

After all, if we trust DDs to develop software that installs in our
systems as root I guess we could trust some of them to report spam fairly.

IMHO, that's one of the issues with the web reporting button: it's anonymous
and might lead to abuse. If you can link report to reporter (through a
GPG-signed e-mail) then you can increase trust (if you trust the reporter)
and consequently, increase your confidence in doing something more automated
with *his* report than with John Doe's. Would that be a possibility? Maybe
long term?

Just my 2c.

Javier

Attachment: signature.asc
Description: Digital signature


Reply to: