Rough consensus reached. Let's move on.
2004-04-15 10:53:40
I sense a rough consensus that all of the following, applied together,
will be more useful than all but any one of them:
1)HELO checks
2)MAIL FROM checks (The C-ID folks admit this is not a bad idea.)
3)From: (and other 2822) checks (The longstanding LMAP folks admit this
would be nice, but hard, and insist 1) and 2) are essential.)
4)Maybes: MTAMARK, spamhaus's cool .mail proposal! and CBV
5)DNSRBL, DNSRWL, Bayesian Filtering, etc that is already in widespread use.
6)RHSRBL, RHSRWL.
7)Accreditation marks for WL.
8)SRS &| VERP
There hasn't been any consensus movement toward or strong argument for
elimination of 1, 2, or 3. Agreed?
And I think the problem is hard enough that we should assume use of ALL
of them:
I'm hopeful, but not yet confident that all 8 of the above, even if
widespread and applied together, will cause a drastic *long term*
reduction of abuse!
So, MARID should provide the info to support at least 1,2,3,6
(implicitly), and 7.
Does anyone want to assume this consensus and move on? It's not May '04
yet, but I think it's time to move to step two. Let's start thinking
about how one might code this, which will likely inform more exactly
what MARID records need to provide? In other words, I'd like to
We will be doing this NOT to make it part of the spec's *requirements*:
we don't want an RFC that tells people they MUST filter or tag in a
particular way, but I think we need not shy away from saying how people
MAY, and in some clear-cut cases SHOULD tag/filter* if they want to use
the info MARID records in a sensible/BCP way.
It would be a great thing to have pseudocode that supports all of the
above, and be able to look at it and say 'This looks like it'll handle
the whole abuse problem so well we don't need the following stuff in 4
and 5: ______, which will make adoption faster. I noticed that the
media (Network World front page!; dumb article on our "Anti-Spam"
"Crusade") think we're aiming to tackle the mail abuse problem using MTA
authorization, and I think that's our goal, even if it isn't stated in
the charter.)
It would even be a great thing to have pseudocode that supports all of
the above, and be able to look at it and say 'This looks like it won't
handle the spam problem very well in the long term; anyone have a better
idea?
*Probably best to call it "classify", and say that action based on that
classification might be tagging, filtering, prioritizing, bouncing,
dropping, etc, under a policy known to, and preferably defined by the
recipient.
Who wants to collaborate on an LMAP receiving MTA pseudocode (or code)
draft?
<Prev in Thread] |
Current Thread |
[Next in Thread>
|
- Re: User experience, (continued)
- Re: User experience, John Gardiner Myers
- RE: User experience, Harry Katz
- RE: User experience, Harry Katz
- Re: User experience, wayne
- Re: User experience, Matthew Elvey
- Re: User experience, Doug Royer
- Re: User experience; RHSBLs; Strong From: check seems possible, Matthew Elvey
- Rough consensus reached. Let's move on.,
Matthew Elvey <=
- Re: Rough consensus reached. Let's move on., Marshall Rose
- Re: Rough consensus reached. Let's move on., wayne
- Re: Rough consensus reached. Let's move on., Jon Kyme
- Re: Rough consensus reached. Let's move on., Marshall Rose
- Re: Rough consensus reached. Let's move on., Jon Kyme
- Re: Rough consensus reached. Let's move on., Andrew Newton
- Re: Rough consensus reached. Let's move on., Jon Kyme
- Re: Rough consensus reached. Let's move on., Matthew Elvey
- Re: Rough consensus reached. Let's move on., Marshall Rose
- Re: Rough consensus reached. Let's move on., Alan DeKok
|
|
|