[EM] ignoring "strength of opinion"

[James Gilmour]

> rob brown Sent: Thursday, December 01, 2005 1:38 AM
> On 11/30/05, James Gilmour <jgilmour at globalnet.co.uk> wrote:
> > rob brown Sent: Wednesday, November 30, 2005 11:17 PM
> > >From a purely utilitarian point of view (i.e. "greatest
> > happiness"), it makes a lot of sense to give more weight to
> > the opinions of those who feel more strongly.  But common 
> > sense tells us why this is a bad idea.
> 
> By "common sense" I presume you mean, in the context of 
> elections, that it is fundamentally undemocratic.  
> 
> Not sure that's really what I meant, because it all depends 
> on the definition of "democratic" and don't think I want to 
> go there.  :) I suppose its unfair, but even that is 
> debatable. 

What I had in mind was if I vote 1, 2, 3, 4 (1 = most preferred, the one I want to see win) for candidates A, B, C, D,
and you vote 100, 99, 2, 1 (1 = most preferred) for the same four candidates, it would be fundamentally undemocratic if
your vote counted for more in determining the result just because you expressed your preferences more strongly that I
did.

I can understand that there might well be some difference in the social utility assessments of our preferences against
the outcome, but the issue here is democratic representation and "one person, one vote" surely comes above all else.
Once you depart from that (no matter what your opinion of the other electors!), you are on a very slippery slope indeed.


>> I think this circle can be squared, if you normalise the 
>> responses so that each respondent contributes equally to the 
>> determination of the result.  That is democratic ("one 
>> person, one vote") but still allows those who wish,  to show the
>> relative strengths of the preferences they express.  Brian 
>> Meek, inventor of Meek STV, described such a system for 
>> normalising weighted preferences in multi-winner elections.
> 
> Is that possible if you only have one question on the poll?

Yes, provided you don't restrict the permitted responses to "yes" and "no".  In opinion polling it is quite common to
ask respondents to indicate on a scale how strongly they are for or against a proposition.  You can set all kinds of
scales and you have numerous ways of summing the results (and so assess the imputed "social utility", "regret",
"happiness", etc).  But if you want to use those same responses to arrive at a democratic yes/no answer, then each
person's response must contribute equally to the determination of the result.  That can be achieved by normalisation.
In a single-question poll, in practice that would simply mean counting "for" and "against", but that has normalised the
scores.

I made my comment following this paragraph of yours:
>  Intentionally
> ignoring this information (or, more likely, not collecting it 
> in the first place) is the only thing that makes 
> sense....otherwise people who had any opinion whatsoever would 
> have an incentive to vote insincerely, saying they felt very 
> strongly so as to have the most impact on the outcome. 

I had not assumed your comment in this paragraph was about a single question poll, but about single-winner elections in
general, where there might usually be several candidates.  The only example of normalisation of weighted preferences I
have seen was for multi-seat elections, but I see no reason why the principle is not equally applicable to single-winner
elections where the votes are recorded as weighted preferences.

James Gilmour