[EM] Condorcet strategy

[Dr.Ernie Prabhakar]

Hi James,

For the record, I appreciate your efforts to constructively share your  
point of view, even if it s unpopular.   Of course, that doesn't mean I  
agree with you. :-)

On May 28, 2004, at 5:20 AM, James Green-Armytage wrote:
> Here is an example in which it takes fewer
> insincere votes to execute a burial:
>
> Sincere preferences
> 27: A>B
> 25: B>A
> 24: C>A
> 24: C>B

I presume you interpret this as A>B>C, etc.

> Pairwise comparisons
> A:B = 51 : 49
> A:C = 52 : 48
> B:C = 52 : 48
> 	A is a Condorcet winner. But if just a fraction of the B voters  
> reverse
> their preferences,

meaning voting B > C > A instead of B > A > C?

> they can steal the election for B.

But if they've guessed wrong by two votes, they could hand it to C,  
right?

Y'know, this reminds me in an odd way of probabilistic methods for  
finding prime numbers.   They're not perfect, but the odds of them  
being wrong are less than than of deterministic method being wrong  
because of a cosmic ray corrupting the program, so they are at the  
"effective limit of accuracy."

That's sort of how I feel about voting systems. At some point further  
refinements are "in the noise" - they deal with effects that are below  
statistically significant fluctuations.  Things I consider sources of  
statistical fluctuations are:
	- the actual voting population is a subset of the potential  
electorate.  Sure, one can posit non-voters as "A=B=C", but that's  
still just an approximation.   This may even be a systematic -- rather  
than statistical -- error, since certain demographics are less likely  
to vote.
	- any polling technology has a non-zero error rate (hey, *I* have  
mis-marked my ballot at least once).
	- views shift over time - having the election on, say, a Tuesday vs. a  
Wednesday would likely produce different results, just due to the  
fluctuating nature of many human opinions

One could argue that all these results are small (and we should work to  
make that smaller) and I'd agree with you.  But so is the margin  
required for many of these problems to become serious.   I rather like  
the TopFive.com (humor site's) analysis of the 2004 presidential  
election  
<http://66.102.7.104/search?q=cache:GILV_w8vJc4J:www.topfive.com/ 
arcs_top5/t5032701.shtml+site:topfive.com+florida+votes&hl=en>

>  Number of votes by which George W. Bush carried Florida: 862 +/- 9301

The fact is, sometimes the results are simply in the noise, and the  
resolution is pretty much arbitrary.  Its unpleasant, but that's just  
the way it is.

At least for me, its hard to take seriously strategy problems that only  
occur in what is effectively a statistical tie, or require  
foreknowledge of greater precision than possible with polling data.   
Put another way, I think it is important to clearly specify the  
*context* in which these strategy problems become serious.   Life is  
full of (arguably an infinite number) of theoretical problems -- which  
are real --  but not worth worrying about in practice since they're  
swamped by other problems.

I don't know that your problem falls into that category, but I don't  
think I've seen compelling (to me) evidence that it doesn't.  I'm  
willing to be proven wrong, though.

Best regards,
-- Ernie P.
-----------
Ernest N. Prabhakar, Ph.D. <DrErnie at RadicalCentrism.org>
RadicalCentrism.org is an anti-partisan think tank near Sacramento,  
California, dedicated to developing and promoting the ideals of  
Reality, Character, Community and Humility as expressed in our Radical  
Centrist Manifesto: Ground Rules of Civil Society  
<http://RadicalCentrism.org/manifesto.html>