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

Re: Spam on lists? [was: Re: serious problems with Mr. Troup]

On Sun, 22 Feb 2004, Matt Hope wrote:

> [Santiago Vila <sanvila@unex.es> - 18:42 Saturday Feb 21]:
> > Funny. Do you really care about list abuse? Then please fix Bug #228228,
> > which is the cause for nearly HALF of the spam we receive.
> Regarding your bug, #228228, I've never seen the listmasters claim to
> run razor over the email lists - I don't believe it to be a bug to not
> be doing something you never claimed to be.

Razor2 was working last year. I know very well because it was me who asked
the listmasters to use it (see the archived bugs against lists.debian.org).
It was, and it still is, very effective against spam, and not using it
on an open list which receives a lot of spam would be a bug even if it
had not been used before.

> I'd also like to point out that the listmasters are responsible for
> blocking an insane amount of spam from hitting our lists, and I think
> they should be congratulated. Sure, maybe one or two emails might slip
> through, but compared to the number they block, it is nearly
> statistically insignificant.
> The graphs that Pasc has provided make this pretty clear. For
> debian-devel[1], on some days, the amount of blocked spam can be as
> much as nearly four times the number of messages that went through to
> the list.

There is a common misunderstanding about spam statistics.

What the readers of the different lists care about is the spam they
see over the total of messages they receive, not the spam they see
over the spam the *don't* see.

If normal email remains more or less constant, spam grows exponentially,
and we always block 95% of spam, then the proportion of spam messages
among the total of received messages will also increase exponentially.

It is my opinion that if spam continues to grow exponentially, then
sooner or later we will have to make the lists moderated or closed,
and better spam filtering will only delay the time by which this happens.

Reply to: