Re: ...

[Jim Popovitch <jimpop@gmail.com>]

On Sun, Jul 4, 2010 at 19:31, Stephen Gran <sgran@debian.org> wrote:
>
> No, Russ implied that reality occasionally intrudes on fantasies of
> spam-free inboxes.

Russ stated:
    It's unlikely to get substantially better than it is (I believe
we're already
    rejecting something like 95% of the incoming mail), so if it's still not
    good enough for you, you should probably consider unsubscribing.

I beleive that 99% is achieveable, and I believe his final
"unsubscribe" sentence is akin to walking away from the problem.

> If, as you imply, you're a professional mail admin familiar with environments
> of vastly divergent requirements in a single ruleset, catering to users who
> speak every language on Earth, submitting legitimate mail from every corner
> of the globe, you will understand some of the challenges.

I am exactly that person, and I have 10+ years experience doing exactly that.

> If you couple that with prizing getting good bug reports and user feedback
> over eliminating every single spam, you might begin to get an idea of the
> difficulty of the task.

Just like there are different roles in the *@l.d.o lists, there can
(and should) be
different inbound policies.   In the past I've personally unsubscribed from
several l.d.o lists due spam on lists not even associated with
bug/user feedback.

> It's not as if we can just ditch mail based on presence in a DNSBL or
> non-ascii character sets, or even a wildly misconfigured mail server
> (that may be what the bug report is about, after all).

If corporations, governments, policy bodies, heck even the U.N can do it, so
can Debian.   Again, different policies for different list addresses (just like
abuse@ and postmaster@ are generally handled differently than support@)

> Don't get me wrong, I would welcome new blood to contribute fresh ideas
> and energy.  It's just that very frequently the spam issue seems to be
> one of those where people are very interested in telling you about what
> works for them in a very different environment to what there is in Debian,

While there is no such thing as one-solution-fits-all, Debian is not so unique.

> and it is energy draining to keep having the same discussions over and
> over about something that seems to be both a hot button issue and largely
> irrelevant.

Perhaps it's time to look into addressing the hot button issue and not just
ignoring it.  I'll remind you that this thread started due to an
easily identifyable
spam slipping through.

> I think that, at present, Debian accepts something like
> less than 1% of all mail offered to it across our various mail servers.
> Yes, it is possible to do better, but probably not significantly better,
> mathematically speaking.  I think that that may be a pretty good place
> to strike a bargain between openness and spam fighting.

-Jim P.