Re: GNOME Board of Directors Foundation Elections Spring 2009 - Preliminary results

[Filippo Argiolas <fargiolas gnome org>]

On Fri, Jun 26, 2009 at 3:38 PM, Dave Neary<dneary gnome org> wrote:
> Hi,
>
> Dave Neary wrote:
>>
>> A small correction to explain exactly how random transfers work:
>
> Well, actually, I just found out from the OpenSTV guys, that how Filippo
> said is how they work.

Oh, nice, I had a feeling it was slightly more complicated. Glad to
hear it's not :)

> In count 1, Vincent has 60 votes, they're shoved into a stack. The top 33
> votes from the stack get redistributed in count 2. No randomness at all, no
> shuffling, and we don't look at the distribution of the 2nd preferences to
> calculate who gets what.

In my opinion the issue is not the randomness itself. After all if you
count the ballots always in the same order (the one they arrived) the
process is somewhat deterministic and it gives always the same result.
The problem is with assuming a relatively small sample to be
representative of all the electors. With random transfer a significant
number of ballots only gets counted for their first preference.
As Shaun said, there is a good chance that the sample you take, not
being shuffled, could have a non trivial correlation making the
results even more unfair.
I believe the only reason random transfer exists is to be able to
count ballots by hand but if you're using a software to compute
results fractional transfer is clearly to be preferred.

Cheers,

Filippo