On Oct 7, 2013, at 11:56 PM, Martin Rex <mrex@xxxxxxx> wrote:
> Dearlove, Christopher (UK) wrote:
>> dcrocker@xxxxxxxx
>>>
>>> From what you've written, your basic point seems to be that 51% isn't
>>> enough; it's worth making that explicit.
>>
>> To add to the confusion, and to emphasise the point about making clear,
>> British and American English differ here. If three proposals (not the
>> most common case, I agree, but it can happen) have 45%, 35% and 20%
>> of the votes, the first of these has a majority, sometimes emphasised
>> as simple majority, in British English. (We do not - to our loss - use
>> the word plurality. Just 51% is given the strong term absolute majority.)
>> I haven't checked the context here, but saying not just a simple
>> majority might suggest to a British English user that 51% is enough.
>
>
> Voting as is done in elections for political parties is often going
> to produce a political result, a personal preference of a few rather
> than an engineering solution that adequately addresses the concerns
> of the community at large.
>
>
> What we could do in the IETF, is not just trying to pick the lesser evil,
> but rather use the _engineering_ skills to modify and/or merge proposals
> to increase the number of folks that support the result and reduce
> the amount of folks that object the result.
>
>
> With your example of three competing proposals: A, B and C, and by couting
> "votes" the WG chair determine support of 45%, 35% and 20% respectively,
> does this mean that A has signficant support? Not at all. It could be
> that the folks who voted for B and C did so because they are both strongly
> opposed to A.
>
> A WG chair who wants to be neutral on the decision should probably not
> just call:
> which of the three proposals do you prefer: A, B or C
>
> and perform political inferences on the result, but rather
> _ask_ the engineers direclty:
>
> if we were to select A, would you "support it, are neutral, are opposed"
> if you're either neutral or opposed to A, what change(s) to A would
> make you supportive for A?
>
> if we were to select B, would you "support it, are neutral, are opposed"
> if you're either neutral or opposed to A, what change(s) to A would
> make you supportive for B?
>
> if we were to select C, would you "support it, are neutral, are opposed"
> if you're either neutral or opposed to A, what change(s) to A would
> make you supportive for C?
I think in this case the WG chair should do something different. A better question would be:
Which is worse? To decide on *any* of these 3 proposals, or to keep debating this for another few months?
At that point you might get something like
- Just choose any of them - 90%
- No, we need to choose the right one - 10%
But percentages don't really matter. You get the people who are in the 10% (1-3 in most working groups) and get them to the mic or mailing list and ask them what is so terrible about proposal A (or B or C, whichever one or two they really object to) that it's better to delay the process by some extra months. That gets the objections that are really worth discussing.
If you can get those objections ironed out so that there's consensus (rough or not) for the assertion that "Any of them is better than not proceeding", the selection can proceed by any method that seems fair: majority, coin toss, 3-way Rechambeau ([1]), shortest document.
Yoav
[1] http://tools.ietf.org/html/draft-harkins-rochambeau-02