[python-committers] Timeline to vote for a governance PEP

[Tim Peters]

[Victor Stinner <vstinner at redhat.com>]
> I'm unhappy with the "[] Further discussion" choice. We have a
> governance crisis. Many people would like to see it resolved as soon
> as possible, I don't see the ability to vote for "[] Further
> discussion" as a way to resolve this crisis.

Nobody else does either.  This was added for political reasons:  it's
_expected_ that virtually everyone will rank "Further discussion"
last, so we can "prove" to the world that we really do want it to end
ASAP.  "See?  We voted, and 'further discussion' came in dead last!"


> There are 6 proposed governance PEPs (maybe 7? ;-)). I don't expect
> that everybody will agree on everything in a PEP, but everybody should
> be at least able to order them to vote, no? If no, well, maybe don't
> vote?

I'm missing context for this.  Has someone complained about that?  I
_would_ ;-) , but why bother.  Having to strictly rank forces people
to fabricate distinctions they may not actually believe.  A voting
protocol that forces dishonesty isn't attractive.  Note that Condorcet
methods don't generally require this; for example, Schulze is happy if
you rate any number of choices "all the same to me".  If you in fact
have no preference among (say) 3 of the PEPs, why must you be forced
to say that you strictly most like just one of them - and strictly
most dislike another?

"The reason" appears to be that the more of that people do, the more
likely there won't be a Condorcet winner.

> ...
> I'm not surprised that someone doesn't like one part of the PEP 8001.
> But well, we need to move on and take a decision...

Sure!

>> "Pure Condorcet" is close to trivial to tally:  there is a Condorcet
>> winner, or there isn't.  I wouldn't even bother to write code to
>> figure it out.  For example, write a simple script to convert each
>> ballot to a single line for the following web page, paste the ballots
>> into the text box, and click the "Calculate all winners" button:
>>
>>     https://www.cse.wustl.edu/~legrand/rbvote/calc.html

> Yes, I'm asking for such script. I didn't say that it would be overcomplicated.

Well, that professor already wrote one.  Give me the task of tallying
the ballots. and I'd do just what I said above :-)

> The PEP 8001 is not trivial, it expects a specific format:
>
> **DO NOT LEAVE ANY BRACKETS BLANK!**
> **DO NOT REPEAT A RANKING/NUMBER!**
>
> Maybe it would help to have a script to validate my own vote? (Also
> ensure that all choices are present?)

Someone will have to do that, but it's a different issue than I was
answering above:  your original "Which tool can be used to compute the
vote result?".

> ...
> Hum, it seems like you are unhappy with the chosen voting method.

Not at all!  If we had used a ranked ballot in the poll, "Pure
Condorcet" would have been my #1 choice.  For this election.

> Again, we have to move on and take a decision.

Yup.  In the poll, I voted "approve" for everything, but retracted my
vote for IRV after it was pointed out that _other_ PEPs pointed back
at 8001 to define the voting method _they_ used too.  I was willing to
endure IRV for a PEP 8001 one-shot vote, but not for eternity.

I could live with any of the other methods forever, but if we had a
PEP on voting methods in general, _then_ I'd push for hybrid
score-runoff methods, like STAR and 3-2-1.  They're much easier to
understand, and do better in large-scale voting simulations, than the
others.

> We cannot discuss voting methods forever, and there is no perfect voting
> methods. Only tradeoffs.

Which I know ;-)  But some nevertheless do better than others under
very widely varying assumptions.  That's why, e.g., absolutely nobody
is in favor of plurality.  "But nothing is perfect" isn't actually a
reason for accepting something that's awful in ordinary conditions ;-)

The best of the Condorcet methods are in fact very good.

> I looked at the length of the discussion, and I understood
> that everybody had the opportunity to express their opinion, and the
> discussion gone deeply in voting methods, as Carol, I was impressed by
> the level of the discussion :-)