[EM] MJ: Worse Chicken Dilemma than Approval or Score, elaborate bylaws, computation-intensive count.

[Michael Ossipoff]

I'd said:

>> ...and it [SFR] isn't available in MJ, for the reasons that I described in
>> my previous reply to you.
>
Jameson replied:

> Yes it is. Because with approval-style votes, MJ gives approval results.

No. Not with a different count-rule.

I'm just guessing, but you seem to want to say that, though
Score-style SFR won't work in MJ, Approval-style probabilistic SFR
will work. Ok, let's look at what would happen:

Suppose that you want to do probabilistic SFR in MJ. You want to
probabilistically give N points to candidate X. So, with a probability
of N/max, you give X max points instead of 0 points, as you would in
Approval, to probabilistically give N points to X.

What will be the result?:

Depending on what N/max is, and depending on the sizes of the
factions, on on how other factions vote, X's MJ score might be 0, or
max, or some inbetween amount that you and your faction have _no_
influence on.

In other words, probabilistic SFR doesn't work in MJ, just as
Score-style SFR doesn't work in MJ (as I showed in previous postings)

Jameson, you really need to better say what you mean. You need to
better specify whatever strategy it is that you want to suggest for
MJ, in order to achieve SFR.

I've told you why Score-style SFR won't work in MJ, and I've just now
told you why Approval-style probabilistic SFR won't work in MJ. If
there's some other strategy that you think can achieve SFR in MJ, then
you need to actually specify it. And then, don't forget to furnish an
example to show that your strategy works, and how it works.


> if it's possible under approval, it is possible under MJ.

Certainly not. MJ isn't counted as Approval is counted. I've just told
you what would happen if you attempted Approval-style probabilistic
SFR in MJ.

> And in scenarios like the one you gave, where the median of the unified
> minority candidate (C) is known (0 in your case)

Sure, C is known to have a median of 0, provided that A voters and B
voters add up to a majority, and 0-rate C.

> , it doesn't require votes
> of max or min; it can be done just as well with votes of min or min+1

...in order for A voters to help B to beat C. Certainly. The problem
is that if the B voters don't reciprocate, and give 0 to A, then B
will win by defection.


>> SFR could be done unilaterally, or could be done by agreement--an
>> agreement that doesn't depend on trust, but only on the other
>> faction's self-interest.
>>
>
> As in MJ.

No. I've told why Score-style SFR, and Approval-style probabilistic
SFR, won't work in MJ.

You can't have MJ with Score properties. You have to choose between MJ
and Score.

>> Several people at EM have discussed and demonstrated why Approval soon
>> homes in on the voter median, and then stays there.
>
>
> Did you even read that page? Because that's a non-sequitor response to that
> page, as far as I can tell. You're just repeating prior assertions.

You'd said something to the effect that median does well, or the
median does well. I assumed that you meant that the median candidate
does well in Approval and Score. But apparently you were temporarily
changing MJ's name to "median". Ok, that's fine.

So you're saying that a more sophisticated discussion at a website
shows that MJ does well, whatever that means.

That's nice, but I've shown here that MJ doesn't do SFR at all.

How regrettable that you're unable to quote those highly sophisticated
arguments here, from that website. So then, you're saying that that
website's more sophisticated arguments show that MJ can do SFR? Or, if
that isn't what you mean, then do you want to tell us what you mean?
If you don't want to, that's ok.

Yes, by all means, if you want to, do quote for us those more
sophisticated arguments that show that SFR can be done in MJ, or that
the Chicken Dilemma is less serious with MJ.

But "handwaving" at a website just won't do.

...a vague statement that some website's sophisticated arguments show
that MJ does well, whatever that means.

>
>>
>> In fairly recent postings, I've told some reasons why the Chicken
>> Dilemma won't be as much of a problem when looked at over time (as
>> opposed to in one single particular election) in Approval or Score.
>> But sometimes one wants to avoid the Chicken Dilemma in one particular
>> election. That's when SFR is more important. But it's helpful in
>> general too--and unavailable for MJ.
>
>
> Wrong.

So you've claimed. I've told why Score-style SFR won't work in MJ.
And, in this post, I've told why Approval-style probabilistic SFR
won't work in MJ.

>> > In practice, in MJ both factions could rate each other's candidate at 1
>> > (the
>> > second-from-bottom rating). This would mean that any further defection
>> > would
>> > be risky
>>
>> No it wouldn't. If the A voters rate B at 1 (out of 100), and the B
>> voters rate A at 0, then here are the MJ scores:
>>
>> A: 0
>> B: 1
>> C: 0
>>
>> (...for the reasons described in the post before this one, the post
>> that you're replying to)
>>
>> B wins by defection.
>
>
> This defection is dangerous: if both sides do it, C wins.

Exactly. That's a necessary condition to have a Chicken Dilemma.
Because bilateral defection is so dangerous, the A voters, being more
co-operative, feel compelled to not defect. And that's why they're had
by the B voters.

> And it is not a
> temptation as with score or approval: unlike score or approval, it is
> impossible for defection short of that required to give C a chance, to give
> A or B an advantage.

You need to re-word that, to better say what you mean (provided that
you mean something and know what you mean).

Of course defection by B "gives C a chance" if the A and B factions
both defect. That's why there's a Chicken Dilemma. That's equally true
in MJ, as you yourself agreed in some abovequoted text.

>> Sure, if the A voters and the B voters both give eachother's candidate
>> a point, then the winner will be A or B. But that just means that
>> there isn't a problem if no one defects. The Chicken Dilemma is about
>> what happens when someone _does_ defect.
>
>
> Yes. When some ONE. Not when some entire faction, as in your example.

It's customary, when speaking about such problems as the Chicken
Dilemma, to speak of there being two "players". That doesn't mean that
there are only two voters in the election. It means that the A and B
factions are each collectively referred to as a "player". That's a
convenient simplification.

Are you saying that, in a large election, MJ doesn't have a defection
problem if only one voter defects?

> You don't understand MJ or CMJ.

I have no idea what CMJ is. I've been talking only about MJ, because
it's a popular proposal.


> They both have "tiebreaking" procedures that
> would naturally give the right result.

Of course. MJ needs that. Do you remember when I said that MJ has
elaborate bylaws?

>> For one thing, tie-proneness isn't
>> considered a good property.
>
>
> In CMJ, the "tiebreaker" is an integral part of the process, such that the
> tie is broken before it even exists. There is no sense in which CMJ can be
> called tie-prone.

Again, I have no idea what CMJ is. But of course, it goes without
saying that when the needed tiebreakers are added to a tie-prone
method, then it can be called "not tie-prone".

When MJ gives the same median score to two candidates, as in the
example I discussed, and if you wouldn't flip a coin--you forgot to
tell us what you'd do instead. But it doesn't matter. It comes under
the term "elaborate bylaws".

>> For another thing, A's win probability
>> will be 1/2, even if A has many more voters than B has.
>
>
> Wrong.

Jameson, when you say, "Wrong", you should then tell why it's wrong :-)

Ok, if, in the situation that I described, where the A and B factions
both rated eachother's candidate at 1, and they both ended up with
equal MJ scores: If you wouldn't flip a coin to choose between those
two equal-MJ-score candidates, how would you choose between them? In
MJ. We're talking about MJ, as it's popularly proposed, not CMJ,
whatever that is.

>> If you want to talk about co-operative trust (as you were doing),
>> then, in Score, each faction could agree too trustingly and ethically
>> give eachother's candidate max minus one. Then, they're helping
>> eachother nearly maximally against C, and yet whichever of {A,B} has
>> more voters will be the winner.
>>
>> That's another thing that won't work in MJ.
>
>
> Um, yes it will.

Um, if both the A and B factions give the same non-0, non-max, rating
to eachother's candidate, and if neither A nor B has a majority voting
it 0 or max, and if the C voters give 0 to A and B, then A and B will
both have the same MJ score. For reasons that I've already told.

In other words, even if the A faction is larger than the B faction, A
and B will still have the same MJ score. In other words, the
co-operative strategy that I described doesn't work in MJ.

Um?

Mike Ossipoff