James Carlson wrote:
> Garrett D'Amore wrote:
>
>> 1) Can a member +1 his own case? ("own" here can also mean two things
>> -- a case he's the sponsor, but not submitter of, or a case which he
>> also is the submitter of.) The document says nothing here, but in a
>> meeting with Tim in the past when I asked for clarification the
>> consensus I understood was that, no, a +1 from a *different* member is
>> required.
>>
>
> I thought that part was quite clear: no, you can't do +1 for something
> you authored or sponsored.
>
> The whole point of this +1 thing (which I didn't like adding, but ended
> up going with the majority opinion) is that we had fast-tracks that, due
> to a lack of specific expertise on the ARC, were being approved without
> substantial review. Something needed to be done about that. If we were
> to allow the sponsor to add the +1, then we're really back where we
> started, and the whole thing becomes an unnecessary exercise.
>
>
>> 2) The threshold for when a +1 is required. The guide says this is
>> required only when the mail file is empty. However, I think we've
>> mostly been applying a higher standard, either requesting
>> an explicit +1, or assuming one if its obvious that review from a member
>> has occurred and consensus has been reached. This is highly subjective,
>> because it requires the chair (or someone else) to judge whether or not
>> a member really reviewed the case without an actual statement (in the
>> form of a +1) from the member that he did. I'd like to either clarify,
>> or change (depending on what the actual current policy is), the policy
>> to explicitly require a +1 in all fast tracks. This will eliminate any
>> ambiguity, and ensure that a consistent standard is applied across fast
>> tracks.
>>
>
> I'm happy with saying that as long as one member has read it enough to
> comment on it, then the process has served its purpose, and no extra +1
> motion is required.
>
> The political issue behind the "+1" process is that the ARC needs to
> have enough breadth of experience to review everything that comes by it.
> If it isn't getting the membership it needs -- if management is either
> assigning only junior engineers to the task, or if there are groups that
> are contributing nothing whatsoever to the process -- then the quality
> of review will suffer. Rather than let that happen, it'd be better to
> affirmatively reject proposals for which the ARC lacks enough resources
> to cover.
>
> The theory is that by doing so, two useful things occur: first, if they
> didn't know before, then managers do know afterwards that there's a
> resource problem here and, second, because groups tend to be focused on
> particular technical areas, it tends to punish those who are putting
> load on the ARC without shouldering the burden. Though there is
> collateral damage involved, which is one of the reasons I wasn't in
> favor of needing +1.
>
My problem with this is that just because an ARC member makes some
comment does not (to me at least) imply that the member feels qualified
in the subject matter at hand to offer a meaningful review. For
example, if an ARC member points out that a case uses interface bindings
which are out fashion (e.g. "Evolving"), as a nit, it should not IMO be
automatically assumed that the ARC member has read the rest of the case,
much less felt qualified to offer substantive review.
Of course, at the other end, if Gary has a detailed conversation with
Darren about a case involving password policies, then we can probably
assume that the case is well reviewed even if neither one explicitly
says "+1".
Those two cases represent the extreme edges. In the middle are cases
where members may be commenting on aspects of the case, but which may
fall outside of areas of (self-acknowledged) expertise. Having a policy
that doesn't require the explicit +1 leaves to many things to be
subjective for no real merit. (If a member really believes a case is
ready, how hard is it to say "+1" in e-mail? If this is too hard, then
IMO the member should reevaluate his ability to commit to ARC in the
first place.)
Further, if anytime I comment on a case its automatically going to be
assumed to mean +1, then I'm going to be a lot more hesitant to comment
on cases unless I'm prepared to either assert that the review is
complete, or derail the case. That means that what might sometimes
still be useful exchanges will not occur because I don't want to be
misjudged as having fully reviewed the case.
I think an explicit +1 policy for *all* fast tracks is the best policy.
-- Garrett
