Brian E Carpenter wrote:
On 2007-01-08 11:08, Brian E Carpenter wrote:
a) we believe that it is indeed the document shepherd's
job to summarise issues and take them back to the WG, as
stated in section 3.3 of draft-ietf-proto-wgchair-doc-shepherding.
This certainly seems reasonable. Unfortunately, it has two fundamental flaws.
One, of course, is that it introduces the opportunity for the person making the
summary to get things wrong. The more generic the AD's concern, the more likely
this error, I believe. The wg can then pursue a resolution that turns out not
to match what the AD meant.
The second is that it permits the AD to offer relatively vague concerns, and
even to change them over time. Since the AD has not been required to document
their concerns fully, there is no history to indicate that something they offer
later was not in their original Discuss.
If an AD is going to impose the considerable costs of the extra effort by the
wg, to resolve a Discuss, the AD should be required to fully document their
concerns, to whatever level of detail is appropriate for that Discuss. Key,
here, is to avoid generic statements of concern about reliability, security,
efficiency, interoperability, or the like, and instead to make specific
statements about the relevant flaws they see. That way, the working group can
evaluate the concerns concretely and can have a good idea that they are pursuing
a resolution that will be acceptable.
There is also the obvious benefit of transparency, making the AD's concerns
fully public, encouraging public review, comment, and assistance. In terms of
ensuring that transparency, the comments need to be circulated to the working
group, actively, rather than passively posting them on a web page and assuming
that wg participants know when and where to look. (This can be accomplished
easily, I believe, by sending a notice, with a link, to the wg when the AD
comments are created or revised.)
In the current model, any follow-on discussion really is between the Design Team
and Chairs, with the AD. This introduces the possibility of significant
late-stage changes that are agreed to by a smaller set than the whole working
group. Even for items that are then returned to the wg, the wg cannot see the
basis for the proposed changes; worse, having this all be late-stage, along with
the added delay during the effort to resolve the Discuss, stands a good chance
of resulting in less careful inspection by wg members interested in expediting
closure.
Brian E Carpenter wrote:
>
>> Why not simply:
>> - copy all Comments and Discusses to the WG mailing list
>> - hold all discussions on the WG mailing list until resolution
>
> Why would we do this for technical typos and other things that
> are essentially trivial?
At the moment, it's not being done at all. Besides that, the idea that a
document would be blocked because of a typo, or anything else trivial, is worth
re-considering.
In any event, why should an AD discuss be subject to less transparency than what
is supposed to take place during normal working group activities, where folks
post all sorts of comments, trivial and substantial.?
> This is what should, IMHO, be the PROTO shepherd's job to decide
> about, as well as consolidating issues when more than one AD
> (or other reviewer) finds the same thing.
WG formation is a whole other matter, although the more that is done on an open
mailing list, the more likely the wg charter will be in synch with community
desires and needs.
Brian E Carpenter wrote:
> On 2007-01-15 17:11, Michael Thomas wrote:
>>> ...I'd expect an AD to enter WG discussion
>>> when raising fundamental issues, but not for straightforward
>>> points.
>>
>> ...This sets up sort of a
>> representative democracy kind of situation vs. a direct
>> democracy that would be a conversation directly on the wg list.
>> I can understand the IESG's desire for filtering, but that does
>> place a lot of power in the hands of the wg's representatives.
>> And power always begats abuse at some point... is this really
>> what was intended?
>
> Abuse wasn't intended, obviously, but delegation was.
Michael's phrasing is quite apt.
Delegation invites errors, whether in the form of abuse or just plain getting
issues wrong. The premise of the IETF is that broad-based participation ensures
adequate review. Having AD Discusses get resolved primarily by a subset of the
wg ensures less review and less accountability.
d/
ps. Meta-point: There is a consistent tendency to cite an exemplar from one
side of an issue as justifying a choice by IETF management, when the
counter-side is just a valid -- and often more of a concern. I thought IETF
decision-making was about seeking balance?
--
Dave Crocker
Brandenburg InternetWorking
bbiw.net
_______________________________________________
Ietf@xxxxxxxx
https://www1.ietf.org/mailman/listinfo/ietf