headers
Andrew Pimlott | 5 Dec 2002 17:08

Re: current A.6 draft

On Wed, Dec 04, 2002 at 10:15:39PM -0500, Raul Miller wrote:
> On Wed, Dec 04, 2002 at 06:32:26PM -0500, Andrew Pimlott wrote:
> > YAExample: sincere preferences are
> > 
> >     3   ABD
> >     2   BAD
> > 
> > but voters vote strategically
> > 
> >     3   ADB
> >     2   BDA
> > 
> > We are deadlocked.
> 
> D defeats B 3:2
> A defeats B 3:2
> A defeats D 3:2
> 
> A wins.

Sorry--I had written initially (before editing) that A, B, and D are
as in your example:  A requires a 2:1 supermajority, B a simple
majority, and D is the default option.

Andrew
Permalink | Reply |
headers
Raul Miller | 5 Dec 2002 18:53
Picon
Favicon

Re: current A.6 draft

On Thu, Dec 05, 2002 at 11:08:43AM -0500, Andrew Pimlott wrote:
> > >     3   ADB
> > >     2   BDA
> > > 
> > > We are deadlocked.
...
> Sorry--I had written initially (before editing) that A, B, and D are
> as in your example:  A requires a 2:1 supermajority, B a simple
> majority, and D is the default option.

Ok:

2:1 supermajority means that these developers are trying to overrule a
technical committee decision.  This implies that the technical committee
does not agree with either A or B, and that the issue is technical
in nature.

So, in this case, the fact that no option has enough approval from the
developers (none defeats the default option by by 2:1) means that it's
probably a good idea to talk through the issue a bit more.

Again: this is by design.

--

-- 
Raul
Permalink | Reply |
headers
Andrew Pimlott | 5 Dec 2002 23:42

Re: current A.6 draft

On Thu, Dec 05, 2002 at 12:53:46PM -0500, Raul Miller wrote:
> 2:1 supermajority means that these developers are trying to overrule a
> technical committee decision.  This implies that the technical committee
> does not agree with either A or B, and that the issue is technical
> in nature.
> 
> So, in this case, the fact that no option has enough approval from the
> developers (none defeats the default option by by 2:1) means that it's
> probably a good idea to talk through the issue a bit more.

Sorry again that my example has been fragmented and is thus unclear,
but:  The original premise is that everyone really thinks that both
A and B are better than the default (obviously, the technical
committee is locked in a closet and can't vote).  A and B might in
fact be minor variations on the same idea.  But supporters of A rank
D above B to sabotage B, and vice versa.  Under your system this
works, reliably.  And there is no way out (within the voting
system).

Maybe Debian voters have enough integrity that this is not really a
problem.  But I find it worrying that the incentive exists.

Regarding your other message:  This is one _specific_ disadvantage
to your system that I don't think has been looked in the face.

Andrew
Permalink | Reply |
headers
Raul Miller | 5 Dec 2002 23:50
Picon
Favicon

Re: current A.6 draft

> > So, in this case, the fact that no option has enough approval from the
> > developers (none defeats the default option by by 2:1) means that it's
> > probably a good idea to talk through the issue a bit more.

On Thu, Dec 05, 2002 at 05:42:02PM -0500, Andrew Pimlott wrote:
> Sorry again that my example has been fragmented and is thus unclear,
> but:  The original premise is that everyone really thinks that both
> A and B are better than the default (obviously, the technical
> committee is locked in a closet and can't vote).  A and B might in
> fact be minor variations on the same idea.  But supporters of A rank
> D above B to sabotage B, and vice versa.  Under your system this
> works, reliably.  And there is no way out (within the voting
> system).

[1] Why is it obvious that the technical committee is locked in a closet?

[2] Why is "further discussion" not the right outcome if people are
expressing preferences that they don't believe in?

> Maybe Debian voters have enough integrity that this is not really a
> problem.  But I find it worrying that the incentive exists.

If we can't trust them to vote what they think, why should we trust the
outcome of such a vote?

> Regarding your other message:  This is one _specific_ disadvantage
> to your system that I don't think has been looked in the face.

I'm not sure why you don't think this has been "looked in the face".
In my opinion "further discussion" is the right outcome for this
(Continue reading)

Permalink | Reply |
headers
Andrew Pimlott | 6 Dec 2002 00:48

Re: current A.6 draft

On Thu, Dec 05, 2002 at 05:50:28PM -0500, Raul Miller wrote:
> If we can't trust them to vote what they think, why should we trust the
> outcome of such a vote?

Empirically, many people look at their vote as a means for achieving
the best outcome, not a statement of belief.  (Eg, those who
preferred Nader but voted for Gore.)  Do you assume that there are
no such people in Debian, or would you propose to exclude them?

I think your position is roughly:

    Given

        1. Debian developers' general sense of integrity
        2. social pressure to vote sincerely
        3. a sufficiently "expressive" ballot

    developers will vote sincerely even though

        1. some developers will not see their ballot as a moral
           statement, per se
        2. voting insincerely will be reliably effective in some
           elections

If so, I'll stop arguing.

Andrew
Permalink | Reply |
headers
Matthias Urlichs | 6 Dec 2002 08:52
Picon

Re: current A.6 draft

Hi,

Andrew Pimlott:
> Empirically, many people look at their vote as a means for achieving
> the best outcome, not a statement of belief.  (Eg, those who
> preferred Nader but voted for Gore.)  Do you assume that there are
> no such people in Debian, or would you propose to exclude them?
> 
Given that Debian has a sensible voting system (or at least we're trying
to build one) where "statement of belief" is basically the same as "voting
for the outcome you think best", this is a null question.

>         2. voting insincerely will be reliably effective in some
>            elections
> 
This cannot be fixed by any voting system. Condorcet (and Approval), at
least, has a few rather nice constraints as to what you can affect by an
insincere vote.

--

-- 
Matthias Urlichs     |     noris network AG     |     http://smurf.noris.de/
Permalink | Reply |
headers
Raul Miller | 6 Dec 2002 01:54
Picon
Favicon

Re: current A.6 draft

> On Thu, Dec 05, 2002 at 05:50:28PM -0500, Raul Miller wrote:
> > If we can't trust them to vote what they think, why should we trust the
> > outcome of such a vote?

On Thu, Dec 05, 2002 at 06:48:00PM -0500, Andrew Pimlott wrote:
> Empirically, many people look at their vote as a means for achieving
> the best outcome, not a statement of belief.  (Eg, those who
> preferred Nader but voted for Gore.)  Do you assume that there are
> no such people in Debian, or would you propose to exclude them?

I propose to respect their votes.

In the case you presented, there were insufficient voters who voted
against the 2:1 supermajority requirement required by the two candidates
on the ballot and the default option won.

If that's not what the voters wanted, they shouldn't have voted that way.

--

-- 
Raul
Permalink | Reply |

Gmane