[Election-Methods] Corrected "strategy in Condorcet" section

[Abd ul-Rahman Lomax]

At 02:04 AM 8/2/2007, Juho wrote:
>The votres' "stated preferences" are easy to collect but in a
>competitive environment voters tend to exaggerate. I guess the basic
>problem is the feeling you get when Bush wins Gore and you have voter
>G=100, B=80 and your neighbour has voted B=100, G=0.

But your neighbor cared more than you. Look, the kind of example you 
cited is very common. In this election, you actually have very little 
motive to vote (of those are the frontrunners).

You cared very little, your neighbor cared enough to vote as the 
neighbor did. If there were two people making the decision, which one 
would be right?

"I prefer Gore, but Bush is fine with me."
"I prefer Bush, and Gore is awful."

Which one is the choice that makes that neighborhood a nicer place to live?

Do you trust democracy? If you cared about the outcome, why in the 
world did you vote such a weak vote?

It only makes sense to me if you *didn't* care, and you'd be happy 
with Bush winning.

Look, I *wish* I could get outcomes like that! They have been quite 
rare for me.

The assumption is that voters don't vote their *true* preferences, 
and then are upset about the result. If you'd be upset, your true 
preference is not 100/80. Why in the world did you vote that way if 
Bush was not acceptable to you?

Actually, we expect even sincere voters to vote as described very 
little. Most voters will max one and min one. The scenario is 
unrealistic and a narrow view of the election, rooted in ranked 
thinking. The only reason you don't like the outcome is that you 
mistrust the neighbor, you don't think the neighbor was "sincere." 
But people vote to get what they want, and only if it is easy for 
them will they vote weak votes. They vote weak votes if they actually 
don't care.

Would you be upset if you voted that way? If so, you woudn't do it 
again, and, I suggest, you would be improving outcomes.

Bad thinking. Get more precise. Consider the whole context.

>Intermediate votes may not do harm in some cases but as a main rule
>competitive voters are likely to use mins and maxes as efficiently as
>they can.

Approval votes do harm. To those who vote that way. The harm doesn't 
show in the individual strategies, it shows in the overall social 
utility *to the set of voters who vote that way.* Unless the way they 
are voting is sincere. In which case, what is the problem?


> > But in two stages, you can have your Condorcet cake and optimize
> > satisfaction as well.
>
>Some guidance for reading my comments. I tend to split the cases in
>two categories. Sincere elections => Range ok. COmpetitive elections
>=> better to go for Condorcet or Approval or some other strategy
>resistant but more coarse methods.

Said over and over and over with no evidence but rather "this is bad" 
without explaining *how* it is bad and *how much* it is bad.

>  Trying to put these (opinion
>strength measurements and competitiveness) in one package causes me
>confusion. If you can prove that some such system works I'll include
>that method in my favourite methods list.

Confusion for what? Easy. Vote sincerely, pretty much as you would in 
a Range without runoff, only it gets a little easier to simply vote a 
utility transfer, full scale, no stretching. The system will 
determine your preferences from your votes. The preference marker 
allows you to vote full scale for more than one and still indicate a 
favorite. It answers the number one objection to Range. You can 
equally empower and indicate your favorite too.

Utterly easy for the voter to vote what may actually be an optimal 
strategy, or at least close to that. Let me say this, it does not 
make it more difficult. Have a problem figuring out which of two 
equally rated top candidates is your favorite? Fine. Leave it blank. 
You don't have to make a choice. Only mark Favorite if you feel that way!

Why wouldn't it work? It's a range election! But with a preference 
analysis trigger to confirm that the Range winner is actually the 
best, where there is someone else who would win, in the initial 
election, a pairwise comparison.

> > You get people to vote sincerely by trusting what they say and
> > giving it to them!
>
>There is some truth in this. I'm however afraid than in many cases
>the competitive instincts and fear of others using strategies will
>take over.

In a competitive environment, people have stronger preferences, so 
they will vote that way. What is the problem?


> > You have an election between Al Gore, G W Bush, and Osama Bin
> > Laden, I'll pick the current Bete Noir, he  deserves the
> > approbation in my opinion. If I thought that there was a real risk
> > that Bin Laden would win the election, it wasn't just a joke, I
> > would seriously consider, in Range, voting max for both Bush and
> > Gore. And I think Bush is pretty bad!
>
>Note that Condorcet / ranked ballots would be good here. Ballot
>Gore>Bush>Osama says Gore>Bush with maximum strength, Gore>Osama with
>maximum strength, and Bush>Osama with maximum strength.

Sure. Good in one case and bad in another. Most election methods, 
actually, pick the best winner most of the time.


> > The basic rule (in the absence of strategies) in Range is to
> > express one's sincere opinions.
>
>Correct. The only problem is that the strategies can be applied so
>often. In Condorcet the strategic opportunities can at least be
>claimed to be rare.

Strategy in Range is quite different than Strategy in Condorcet. The 
latter requires reversed preferences, in a way, lying to the system. 
The former merely requires deciding how strong an expression to make 
of your preferences. What people do is to posit an internal weak 
utility and an external strong expression.

It's crazy. You are postulating schizophrenic voters. If you want 
your favorite to win so badly that you won't give anyone else any 
votes, well, that's how you feel!

If you want to slavishly follow someone else's recommendations, 
again, that's how you feel and the preferences being expressed are 
theirs, not yours, and the situation is the same, really.

I've been over and over this many times and I've never seen a cogent response.