Message-Id: <199712230552.AAA11991@spot.cs.utk.edu>
From: Keith Moore <moore@cs.utk.edu>
To: "Eric S. Raymond" <esr@snark.thyrsus.com>
cc: Keith Moore <moore@cs.utk.edu>,
        IETF working group on revision of mail standards <drums@cs.utk.edu>
Subject: Re: How to find a solution 
In-reply-to: Your message of "Tue, 23 Dec 1997 00:13:24 EST."
             <19971223001324.44034@snark.thyrsus.com> 
Date: Tue, 23 Dec 1997 00:52:43 -0500
Sender: moore@cs.utk.edu

> Keith Moore <moore@cs.utk.edu>:
> > I can't tell that you've bothered to analyze the problem at all 
> > beyond simply assuming that we cannot invalidate the behavior of
> > the installed base.
> 
> That's a pretty good assumption to start from.  

I beg to differ.  The entire premise of the DRUMS effort is that
the installed base suffers from ambiguities in the existing 
specifications. It's explicitly the job of the DRUMS group to 
fix those problems, which inevitably means invalidating some 
existing behavior.

This doesn't mean, of course, that DRUMS should ignore transition pain
in choosing its solutions.  

> It's not like Internet
> email is an infant technology; there's a huge user and code base, and
> the cost of deliberately breaking compatibility with the 99% of it
> that assumes Personal-Reply-To semantics would be just too high.

Why?  What pain is caused by doing so, as compared with the alternative?  
As far as I can tell, the pain of changing the meaning of Reply-To
is less than that of adding new fields.  If I'm wrong, I'd like to know
why, but I don't blindly assume that we can't adopt the former strategy.

> The way you might like Reply-To to behave in an ideal world is irrelevant. 
> The way *I* might like Reply-To to behave in an ideal world is irrelevant.

I mostly care about allowing the author to specify where he/she wants 
replies to go.  Whether this happens by fixing Reply-To or defining new 
fields doesn't matter much to me.  But I want the functionality fixed 
in the way that gets it done quickly with the least amount of pain. 
And as far as I can tell so far, that's best accomplished by changing 
the meaning of Reply-To (and by encouraging UAs to have better interfaces).

> The best thing we can do for the users at this point is to rule only
> 1% of the implementations out there non-conforming, rather than 99%.

My analysis reaches a different conculsion.

> That's half the reason I support the combination of Personal-Reply-To
> and Mail-Followup-To.  The other half is that Eudora, which is
> definitely one of the 700-pound gorillas among MUAs, is going to
> implement Mail-Followup-To soon, combining it with their existing
> Personal-Reply-To semantics.  

Hey, I like and respect the Eudora folks, but I don't view Eudora 
as dictating the standard any more than Netscape or Microsoft or 
anybody else.  Nor do I see them as in a better position to make
such a decision than the DRUMS group.  

> Having no need to do things by committee nor
> debate endless starry-eyed proposals that ignore transition costs,
> they've followed the logic of the existing situation to its conclusion
> more rapidly than the DRUMS group.  And I'm sure they know it.

It's not clear at all that the solution you recommend is a logical 
conclusion of the existing situation, or one that imposes a 
minimal transition cost, or even one that is well-defined.  

The dogma is getting tiresome.  Why don't you do the math?

Keith