[EM] James--Strategy, part 1

[James Green-Armytage]

James replying to Mike...

Mike:
>James, say there are two gentlemen who sometimes  punch you in the face.
>One 
>of them might  punch you in the face if someone tells him that you said 
>something about his mother, unless you tell him that it isn't true.  The 
>other will punch you in the face whenever he feels like it, for no 
>particular reason. And he feels like it fairly often.
>The 2nd one is better than the first one, right? Because the 2nd one
>isn't 
>"vulnerable to false mother-insult reports".

	Let's say that there is a man named Irving, who sometimes punches you in
the face with no provocation. However, he doesn't do any more damage than
that, and if he doesn't feel like punching you in the face, he won't do it
even if you try to provoke him.
	Let's say that there is a woman named Marge, who is quite gentle if you
don't provoke her, but if you do provoke her, she can do some terribly
violent things... crack open your head with a baseball bat, cut your
throat, cut off your fingers. And yes, she'll punch you in the face
sometimes, too.
	Let's say that there is a man named Waverly. If you don't provoke him, he
is gentle. What is he capable of, if provoked? Will he ever do anything
worse than punch you in the face? How often are people likely to provoke
him?

	(Provocation = strategic incursion. Violent acts = undesirable results.
Hopefully you can guess who the three people are.)
>
>Duverger's law is about how Plurality results in a two party system. So
>then 
>is a Duvergian scenario a scenario with only two winnable contenders as a 
>result of Plurality being the method in use?
>You seem to be using "Duvergian scenario" to mean any situation with only 
>two winnable candidates, regardless of what method is in use. 

	Correct.

>That seems 
>like a wrong use of "Duvergian scenario".

	Perhaps so. Or at least an unprecedented one.
>
Mike:
>No, given your assumptiont that there are only 2 winnable candidates, 
>offensive order-reversal disappears as a problem. "Widespread truncation"
>is 
>the obvious expected way of voting for supporters of the big-two
>candidates. 
>That would be known to every voter.

	Yes, I agree with you. But when there are more than 2 winnable
candidates, widespread truncation could create limitation/distortion in WV.
	However, the two-viable-candidate case is an important one to look at if
we're interested in comparing WV and IRV. It is possible that WV performs
well in all or nearly all of the scenarios where IRV performs well. If
that is the case, then there is no argument for IRV over WV (assuming a
Smith-efficient WV). This is an interesting area for further inquiry: are
there any general scenarios where IRV performs well but WV performs poorly?
>
James:
>	But anyway, yes, Condorcet methods (including WV) do have a problem that
>IRV doesn't have, because they fail LNH
Mike:
>Maybe I neglected to tell you that I'm not interested in the question of 
>whether, by some way you define "new problem", wv has a new problem. I 
>didn't say that before? 

	Hard to say. You wrote "...it's meaningless to say that wv brings a new
problem.The loss of a CW, or the violation of majority wishes, is the same
undesirable result in wv as in Plurality or IRV, but in wv it takes much
more to make it happen, and it takes less to protect against it."
	I interpreted this to mean that Condorcet does NOT bring a new problem.
Perhaps I misunderstood. Anyway, we can probably agree that the question
of whether Condorcet methods add a "new problem" depends on exactly how
one defines "new problem", and leave it at that.

Mike:
>You say that you don't repeat, but what do you call that? You've told me 
>many times that wv is vulnerabale to burying. Every time you say that, I
>say 
>this: The undesirable results that can happen due to "burying" in wv will 
>often happen in IRV and Plurality without anyone using any offensive 
>strategy at all. Remember the gentleman who isn't vulnerable to false 
>mother-insult reports?
>And then I add this: The difference? IRV and Plurailty will give those 
>undesirable results automatically, on their own, while in wv they're a 
>result of offensive order-reversal. Additionally, in IRV and Plurality, 
>those undesirable results can be prevented only by defensive 
>favorite-burial, while in wv they can be prevented by mere equal ranking. 
>And, in fact, mere truncation is enough to deter the offensive strategy
>that 
>could otherwise cause those undesirable results.
>I've answered that statement for you about twenty times now.

	And although you may not believe me when I say this, I do understand your
point, and have understood it since the first time that you stated it. You
have to understand that I am trying to reply directly to this point, not
ignoring it.
>
James:
>	I call your "majority wishes" definition the "WV-defined immune set".
Mike:
>why do you feel a need to rename what I 
>define?

	That's fine. I will refer to it (when I do) as "Mike's majority wishes
definition" (or "your majority wishes definition"), and to your majority
rule definition as "Mike's majority rule definition." I won't call your
majority rule definition "majority rule", because I think that you define
majority rule too narrowly (as we discussed).

Mike:
>you can't elect a Condorcet loser unless you fail to elect 
>a CW, and you can't violate preference Pareto unless you violate majority 
>wishes. Your worse results tend to be things that can only happen when my 
>undesirable results happen.
>Your goals are just a lot more modest than mine. Where I want to elect
>the 
>CW, you want to elect anyone but the Condorcet loser. Where I want to
>avoid 
>violation of majority wishes, you will settle for only not violating
>Pareto. 
>There's nothing wrong with being willing to settle for less. 

	You've misunderstood my point. Yes, you can't violate Condorcet loser,
Pareto, or do anything else really nasty, if you elect the CW when there
is one, and choose from the Smith set otherwise. So if you resolve not to
violate the sincere Smith set under any circumstances, then you don't have
to worry about that other stuff either. However, that is easier said than
done. Strategy in Smith-efficient methods (especially margins) can
potentially not only violate sincere Smith, but violate these other things
as well. Whereas IRV will routinely violate Smith and Condorcet, it is
virtually assured that it won't do this other stuff, and it is virtually
assured that it will elect a CW with >1/3 of the first choice votes. 
	Do you see what I'm saying? Of course my goal is to keep the result
within the sincere Smith set. Basically, I'm just saying that as we
transition from IRV to a Condorcet system, we potentially lose a certain
amount of safety from really nasty results that IRV affords even with a
devious electorate, and so we have to assess our particular Condorcet
proposal to make sure that it won't do anything really nasty under
strategic conditions. Choosing WV instead of margins is probably a big
help, but we should examine WV very carefully just to be sure.
	I keep saying that my burying argument applies more strongly to margins
than to WV. I hope that you appreciate this. I've given examples where the
last choice of 90% of the voters has a significant chance of winning in
margins. I'm not sure if WV is likely do anything really nasty as a result
of the burying strategy; all I'm saying is that it is a question that
needs to be asked. This is a good direction to look in next: what are the
worst results that WV can produce under heavy voter strategizing? What are
the strongest CW's that can be successfully buried in WV?
	Do you understand what I'm saying? Please try. And if you don't, please
ask me to clarify, rather than assuming that I don't know what I'm talking
about. This dialogue won't be productive if you constantly assume the
worst about my level of understanding and spasmodically hurl insults at me
(like "imbecile"). I'm not an imbecile, and if you can't recognize that
fact, please don't waste my time by addressing any further messages to me.
I'm trying to engage you in a calm, serious discussion regarding specific
issues that I am generally curious about. I intend this to be a process of
discovery, rather than just a process of argument. I hope that you can
join me in such a discussion.

Sincerely,
James


>If you reply to this part one, instead 
>of waiting for part two, please try to make your separate replies to
>parts 
>one and two each only half as long as a reply to both would be. I
>hesitate 
>to reply in two parts, if it could result in two full-length replies from 
>you.

approx. 928 words new text this message
approx. 1126 words new text in your message
>