[EM] "Be careful what you wish for"

[Russ Paielli]

Araucaria Araucana araucaria.araucana-at-gmail.com |EMlist| wrote:
> One reason I went ahead and formulated the Marginal Ranked Approval
> Voting page was to illustrate a method that is biased toward higher
> approval.
> 
>         http://wiki.electorama.com/wiki/Marginal_Ranked_Approval_Voting

Hey, I don't know about the method, but I love those colorful matrices! 
Nice job.

> The problem with this method, when compared to DMC, is that it gives
> higher-approved candidates too much of an advantage.  The long term
> game theory effect of MRAV's bias toward higher-approved candidates is
> that voters will give less approval to non-favorite candidates.
> 
> Forest made a comment a few weeks ago about an irritating former
> contributor to the list, Donald Davidson:
> 
>    http://lists.electorama.com/pipermail/election-methods-electorama.com/2002-February/007310.html
> 
> One of his statements (no longer on the web, apparently his web site
> is gone) was he didn't care what method was used, it would eventually
> evolve into IRV.

This Davison guy is very good at pointing out problems with election 
methods other than IRV, but when it comes to IRV he seems to be wearing 
rose-colored glasses.

The one thing IRV is good at is eliminating the effect of spoilers. But 
the same effect could be achieved by simply persuading all voters to 
vote strategically.

Here's what Gore supporters should have told Nader supporters in 2000 
who thought that IRV could have helped them: write you preference for 
Nader over Gore on the bathroom wall, then vote for Gore and imagine 
that your vote transfered from Nader to Gore.

> Well, that has a kernel of truth to it -- candidates are going to try
> to game the system, whatever it is.  So whatever method you set up, it
> needs to have a certain unpredictable aspect to it, even (or
> especially) if it is deterministic, so voters will give up and simply
> state their true preferences.

I don't think I can go along with that. If a little bit of randomness 
helps discourage strategy, then a lot of randomness will help even more. 
Why not just toss dice? The only effective strategy is to somehow load 
the dice.

--Russ