Date: 27 Dec 1997 14:32:00 +0200
From: kaih@khms.westfalen.de (Kai Henningsen)
To: drums@cs.utk.edu
Message-ID: <6kga3eiXw-B@khms.westfalen.de>
In-Reply-To: <199712270605.BAA03736@spot.cs.utk.edu>
Subject: Re: How to find a solution
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Organization: Organisation? Me?! Are you kidding?
References: <199712270605.BAA03736@spot.cs.utk.edu>
Comment: Unsolicited commercial mail will incur an US$100 handling fee per received mail.

moore@cs.utk.edu (Keith Moore)  wrote on 27.12.97 in <199712270605.BAA03736@spot.cs.utk.edu>:

> > Indeed, this is the same
> > model a couple of million USENET users are already comfortable with.
> > It's been field-tested.
>
> There are several problems with this leap:

Not as much as you seem to think.

> 1. Just because users are comfortable with a behavior doesn't mean
> that it works well.  Many users are "comfortable" with their
> particular interpretation of Reply-To.  (when their use conflicts with
> someone else's use - it's of course the other guy's fault.)

True. Then again, in the case of Usenet, being comfortable means being  
comfortable with the results of hundreds of thousands of people using that  
behaviour. And if you look at behaviour-related flames, they are nearly  
exclusively pointed at software (and people) _not_ following that  
behaviour. That's pretty indicative of "works well", IMO.

> 2. USENET isn't email.  USENET is used primarily for access-to-anybody
> group conferencing, and the vast body of experience with USENET is in
> this area.  Email is also used for interpersonal communication,
> communication within ad hoc groups, communication within closed
> mailing lists, etc.

Well, the vast number of problems with Reply-To: certainly seem to stem  
from its use in the very same area that Usenet is used in.

Besides, there _are_ ad hoc groups and closed groups done with Usenet  
technology, and I couldn't point to any serious problems with those.

> Even assuming that USENET's reply model works well for USENET's target
> audience (which is not clear), this is no reason to conclude that it
> will work well for other uses of email.

It seems easy to just not put in a Mail-Followup-To: in areas where it  
doesn't make sense. I sure dont see a possible conflict here.

> For example, consider an author-supplied Followup-To.  As long as the
> author specifies followups to newsgroup(s) that are universally
> available, there is little need for a responder to edit or override
> the Newsgroups field when composing a followup.  After all, anyone who

That's not quite true, these days. Followups get frowned upon in some sort  
of exponential relation to the number of groups involved (and rightly so,  
IMHO). Of course, if the original Followup-To: is selected with some care,  
there's no need to change it. No reason why that would be any different  
with Mail-Followup-To:.

> wants to continue following the discussion can simply pick up the
> followups in the new groups.  But translate this to an email world
> where a great many lists are closed and you cannot distinguish lists
> from personal addresses, and now the responder has an increased
> responsibility to consider where his responses should go.  Which in
> turn implies the same kinds of changes to the user interface that are
> needed for email with the current Reply-To field.

That's pretty much a distinction in search of a difference. There's little  
practical difference between

Newsgroups: news.groups
Followup-To: local.jokes

and

To: linux-kernel@vger.rutgers.edu
Mail-Followup-To: internal-jokes-list@bigcorp.com

Actually, with Usenet, it's _harder_ to follow up - you probably can't  
post to local.jokes directly, you need to crosspost to some other, widely  
distributed group to get your article to where local.jokes exists (if it  
exists). That's not a problem with mail.

As to "simply pick up the followups in the new groups", more often than  
not, that's not as simple as it sounds. Part is because of using the wrong  
UI model in some news clients, but part is because for _any_ offline news  
reader, you'll have to subscribe to that goup first and will surely miss  
the first posts there. Sound familiar?

As to needed UI changes, yes,of course we'll need some. No way around  
that.

> 3. The USENET model doesn't consider mailing lists, which *will* mung
> headers.  If you define new header fields for lists, you might get
> them to mung those fields rather than the ones supplied by the author.
> If you don't define new header fields for lists, they will mung the
> ones intended for use by the author, just like today.  Either way, the
> recipient needs more choices than just reply/followup.

That's not that much different from moderated groups, except that the  
submission address is generally handled fully automatically. I've seen  
more than one discussion about what sort of header (and body) munging a  
moderator should/may do.

> The real solution to the reply problem is to fix the user interfaces.

That's only part of the solution. We also need to make the underlying  
model more expressive, and it's painfully clear that that's impossible to  
do with Reply-To: alone.

> > I don't understand why there is even an argument here.
>
> Neither do I.

Nor do I.

MfG Kai