Lucien: No Disagreement on Results

[Mike Ossipoff]

Hugh R. Tobin writes:
> 
> Mike Ossipoff wrote:
> > 
> > We're in complete agreement about the results with sincere &
> > truncated rankings in the example. We agree that with sincere rankings
> > Clinton wins, with or without the compulsory contrary half-votes.
> > 
> > And we agree that, with the truncated rankings, Dole wins with the
> > compulsory contrary half-votes, and that Clinton wins without those
> > compulsory half-votes.
> > 
> > Mike
> > 
> > --
> 
> I suggest that "sincere" and "truncated" are likely a false dichotomy in 
> this example, as I think Ossipoff's response to my commentary on the 
> referenced example also suggests.  Truncation as a strategy would not 
> make sense for the Dole voters, so we should take their votes at face 

I said it was insincere, not that it was strategic. Those aren't
the same thing. Voters who prefer Clinton to Nader might refuse
to vote for Clinton on principled grounds, due, for instance
to criminal allegations against Mr. & Mrs. Clinton, or because
of complete lack of confidence in Clinton's honesty by a Republican
who refuses to vote for someone whose honesty he doesn't believe in.
Or, instead of principle, it could be a gesture, a statement. Similar
to principle, but not quite the same thing. And, if there were
lots of candidates, laziness, or being in a hurry could be the
reason for the truncation. All of these would be insincere, in
the sense of failing to express a sincere preference, but none
are strategic, in the sense of being done in order to change
the result of that election.

> value.  The results then do not suggest any defect in the system from 
> counting half votes.  I elaborate in a response to Eppley.

Most truncation won't be strategically-motivated, but it
can still take victory away from a Condorcet winner (an alternative
which, when compared separatetly to each one of the others, is
preferred to it by more voters than vice-versa).

And I've already talked about the violations of majority rule,
and the failure to get rid of the lesser-of-2-evils problem
that result from the compulsory contrary half votes.

Even if, as you discuss elsewhere, a circular tie is a natural
one, majority rule is still violated by the compulsory contrary
half-votes, and voters are still made to regret that they didn't
vote their compromise 2nd choice in 1st place--the LO2E problem.

The problems in the previous paragraph are enough, even if
protecting a Condorcet winner isn't important to you. 

We agree that an offensively strategizing voter, in Condorcet,
would use falsification instead of truncation. But I've observed
strategically-intended truncation in an organizational pairwise
election (where no particular circular tie solution had been
established). If we'd been using Condorcet, & if everyone understood
its consequences, that guy might not have truncated. But doing so
wouldn't usually worsen the result for him, and so maybe he would
anyway, just as a statement, or out of principle, as I said. Also,
he might still do strategically-intended truncation if he doesn't
understand the consequences of the count rules.

By the way, the reason I say it was strategically-motivated
truncation was that it was obvious that the compromise was likely
to win, by beating everything else, and the guy said, "I don't have
to vote for one of those", one of which was the compromise.

> 
> I also think the emphasis on the term "compulsory" clouds the issue 
> presented by this example.  Clinton's win in the example derives from the 
> Dole voters (unaccountably) failing to vote the half-votes between 

I can account for that. The Dole voters know that most likely 
all they'd accomplish by casting false preference votes, half
or otherwise, would be the election of someone worse than would
be elected if they didn't falsify.

By the way, the 3-candidate election is a simple convenient metaphor,
and is useful for that reason, but actually, unlike Regular-Champion,
Copeland, or MPV, Condorcet couldn't actually have a problem in
a 3-candidate race. So these discussions, though they're about
a 3-candidate race, for convenience, become important only
when there are more candidates. 

The reason why there couldn't be a genuine problem in a 3-candidate
race is that in Condorcet the middle voters have no reason to
vote a 2nd choice, since if 1 extreme has a majority it won't
make any difference what middle voters do, and since if neither
extreme has a majority, then Middle is Condorcet winner, and
is needed by each side more than it needs them.

But this isn't true if we have the compulsory contrary half-votes,
because refusal by Middle voters to vote a 2nd choice won't have
deterrent effect. Without the compulsory contrary half-votes,
it would have an absolute deterrent effect.

You said that Clinton wins because Dole voters unaccountably
failed to use the half-vote option between...

> Clinton and Nader, not from the mere fact that the zero option is 

> provided to them.  The whole tiebreak system is "compulsory", and the 

As I said, I've accounted for that failure by Dole votes to
falsify.

> question is what formula to use in making the best of a bad situation -- 
> the circular tie.  For example, using the margin of defeat may not be the 
> best system, and it does not allow any option, but it does not compel the 
> voters to do anything.  The rule I propose would not attribute any false 

Implying that another method does compel voters to do something?

> preferences, because the effect of counting a sincere voter's half votes 

How can you say it doesn't count any false preferences when it
counts you as preferring A to B, and preferring B to A??


> under that rule cannot cause an election result that the voter considers 
> worse than the result that would have occurred had his equal rankings 
> counted zero.  Whether to allow any particular option to voters as to how 

It sure can. When false preferences are compulsorily counted for
truncating voters, the truncation has the effect of order-reversal,
which would tend to backfire badly in Condorcet's method. Oh wait--
it wouldn't backfire anymore, would it? Because the compulsory
half-preferences take away the deterence. Not only would it not
be deterred, but it would be an automatic result of truncation--
half of a reversed preference vote.

By taking away the non-drastic anti-order-reversal
strategy, that compulsory half-votes forced upon the Clinton
voters let Dole win by order-reversal, when otherwise the
order-reversal would have been deterred.

> their votes will be considered for tiebreak purposes depends upon what 
> advantages are gained, compared to the detriments, including increased 
> complexity.  For the sincerely indifferent, non-strategic voter, I cannot 
> see any reason to value the zero option.  Is the only benefit of the zero 
> option to give voters, in rare circumstances, one method strategically to 
> "punish" other voters for the way they are believed to have voted?

First, I don't propose not having votes counted on your behalf that
you didn't vote as an option. It's the natural default assumption.
It would be difficult to justify a default assumption that falsifies
preferences & inventes un-voted votes.

So one benefit of not doing that is that it's more democratic not
to modify someone's ballot without their permission.

There's another benefit though: I've talked about the lesser-of-2-evils
problem, and some majority rule standards, and have posted criteria
for measuring them. The reason I propose Condorcet's method is
that it does an incomparably better job in those respects than
other methods do. That's a benefit. The compulsory (or default)
contrary half-votes would wipe out that benefit.

As I always say, it's a question of what we want from a voting
system. I want to get rid of the lesser-of-2-evils problem, and
I believe that many others do. I believe that if more than half
of the voters indicate that they'd rather have A than B, then
if A or B is chosen, it should be A instead of B. Condorcet, without
the compulsory or default compulsory contrary half-preferences
never violates that standard unnecessarily, and, as I've shown,
it gets rid of the lesser-of-2-evils problem to a degree that
no other method does. Neither of those things could be said
if the compulsory or default contrary half-preferences were
added.

But, as I said, the contrary half-preferences seem harmless
as an option, since people could just regard it as a form
of weak false voting, if they're inclined to opt for false
voting. I have no serious objection to it as an option.

But let's not make falsificatin the default assumption.


Mike


> .-
> 


--