W3C home > Mailing lists > Public > public-audio@w3.org > July to September 2013

Re: Preparing the vote on the data race issue

From: Philip Jägenstedt <philipj@opera.com>
Date: Mon, 29 Jul 2013 09:42:40 +0200
Message-ID: <CAMQvoCm3fN=x5=4kg3u=-B3+a+DCoJfvBZttVrVC1MRQ0OO8Xg@mail.gmail.com>
To: "public-audio@w3.org WG" <public-audio@w3.org>
On Fri, Jul 26, 2013 at 3:50 PM, Olivier Thereaux
<Olivier.Thereaux@bbc.co.uk> wrote:
> Dear all,
>
> It is now most unlikely that the group will reach consensus on the question of "data races".
> Per the W3C process [1], we are therefore preparing a vote on the issue.
> [1] http://www.w3.org/2005/10/Process-20051014/policies#Votes
>
>
> The vote will be held as follows:
>
> * Each organisation and invited expert listed as member of the WG will be able to vote
>  The full list is available at: http://www.w3.org/2000/09/dbwg/details?group=46884&public=1&gs=1&order=org
>
>
> * Each organisation gets a single vote. Invited experts also get one vote each. W3C staff may cast a single vote for their organisation.
>
>
> * There will be a single question put to the vote. Voters will chose one of three options :
>
>  1. No change in the API. The spec will include a clarifying statement about the fact that effects of modifying a buffer while reading its data is not specified and might yield unexpected results.
>
>  2. ROC's proposal, as archived at: http://lists.w3.org/Archives/Public/public-audio/2013AprJun/0644.html
>
>  3. Jer's alternative proposal, as archived at: https://gist.github.com/jernoble/6034137

This doesn't seem like a very good way of voting, see
http://en.wikipedia.org/wiki/Arrow's_impossibility_theorem

I think that most of us care more about whether or not the API is racy
than the specific solution, and splitting votes between options 2 and
3 might result in a situation where option 1 wins even though the
majority preferred either option 2 or 3 over 1.

I would suggest instead that every voter rates the three options
(allowing the same rating for multiple choices) and that some
https://en.wikipedia.org/wiki/Condorcet_method is used to pick a
winner. (I've used https://github.com/bradbeattie/python-vote-core and
Schulze before.)

(A strictly ranked system would also be far better than a single vote,
even if it's not a Condorcet method.)

Philip
Received on Monday, 29 July 2013 07:43:08 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:50:10 UTC