Demorep1: Truncation. Approval.

Mike Ossipoff dfb at bbs.cruzio.com
Sat Dec 7 00:45:30 PST 1996


DEMOREP1 at aol.com writes:
> 
> A Condorcet major truncation example--
> A 334
> B 333
> C 330
> CB   2
> 
> 2 of 999 voters made a second choice.   B wins a glorious victory and a
> mighty mandate to be El Supremo by beating A 335 to 334 and by beating C 333
> to 332.  Plain Condorcet supporters apparently would say this is a perfectly
> OK result.  What would a public focus group think about such a result ?

But you forgot to tell us what's wrong with that result. As I
always say whenever one of you 2 repeats that particular statement
for the millionth time, if you need a comromise to beat someone
worse, and you refuse to vote for that compromise, then no method
can help you. Sorry. Who's dissatisifed with that result? The A
voters? If they like C better than B then whom should they blame
for the fact that they didn't vote C over B? Likewise with the bulk
of the C voters.

> 
> I repeat--- the major competitors to plain Condorcet are top 2xN runoff (N=

> Number to be elected, in common use in nonpartisan elections and in partisan
> runoff primaries in 10 southern states), approval voting and instant run-off.

And Plurality, which is the method used for U.S. partisan elections.

>   All such 3 methods will probably produce a majority of all votes (MOAV)

What does a "majority of all votes winner" mean, and how do those
methods "produce" it? As I always say at this point in the repetition,
Runoff & IRO "produce" a fake majority winner. An alternative that
is preferred to 1 other alternative, and is called a "majority winner"
for that reason, even though a majority prefer someone else to
it--a majority whose wishes are being violated by that "majority
winner".

I've just been told that I repeat things endlessly. I wonder why.
I repeat things endlessly because I endlessly reply to people
who repeat the same many-times-answered arguments endlessly.
Maybe you're tired of my endless repetition & my endless replies
to endlessly recycled junk. Yes, and so am I, and I'm not going
to continue doing it. I just wanted to reassure you about that--
you who told me that, & anyone else who is tired of the endless
repetition.

This is the last, or close to the last of these repeated
replies.

> winner (who of course may not be the Condorcet winner) or winners (where N
> >1).
> 
> The realistic main competitor to head to head is plain approval voting
> despite its distinct possibility in picking a candidate who might be

Though Approval, of the well known methods, seems to me to be next
best, after Condorcet, there are less-well-known ones that are
between Approval & Condorcet in their merit: Bucklin, IRO-1,
& Stepwise-Plurality, for instance. I'm just talking about 1-balloting
methods in this paragraph. The good 2-balloting methods aren't well-
known.

I don't know what you mean by "realistic competitor", but Plurality
& Runoff are the winners so far. And IRO seems to be Condorcet's
main competitor among electoral reformers. Approval doesn't have
that many advocates, so I don't know why you're calling it the
main competitor. Approval's good, & simple, however.

[...Approval could pick...]
> everyone's last or next to last choice if there were rankings.
> 
> With such majority winner alternatives, I would suggest that the public will
> hardly go wild for plain Condorcet with its possibilities of plurality
> winners (as with B above) and its possibilities of circular ties and its

More endless repetition: Any kind of winners is possible if you
postulate ridiculous enough, stupid enough, voting. Even with the
best methods.

Possibilities of circular ties. I repeat again that the possibility
of circular ties isn't a consequence of what method you use.
Collectively intransitive preferences are possible, and have nothing
to do with the method. Why do you keep repeating these things?

> possibilities of order reversal strategies (no matter how risky to the
> candidates) (such as A telling some of his/her supporters to vote for C as a
> second choice to get into a circular tie-- followed by B telling some of
> his/her supporters to vote for A to defeat A's strategy-- followed by etc.

Nonsense. Order reversal is never needed in Condorcet as a defensive
strategy. 

You're concerned that offensive order-reversal will be used
"however risky". Saying that isn't the same as showing that
will indeed be used. I've never known order-reversal to be
used in a rank-balloting election, except on your ballot
in our methods vote, when you ranked IRO, but not Condorcet,
later saying that IRO is worse than Condorcet. And even your
order-reversal wasn't strategically intended, but more likely
emotionally motivated.

As I said, and as Tobin (whom you quote here) also said, there's
no perfect method. Does Condorcet have, in principle, under
improbable conditions, strategy need? Sure. Do you know a method
that doesn't have a problem? As I said, it's necessary to put
this in perspective, with respect to gross, commonly manifested
problems of other methods, such as your Condorcet-Approval
bastardization. I've shown that your proposal has big strategy
problems, and they aren't subtle, & they won't be rare.
You never replied to that. That's what you typically do. Ignore
things that you don't have an answer to, and merrily, blissfully
continue repeating your assertions.

And though you have that in common with Don, I want to appologize
for lumping you together with him in a previous message, because
there are certain ways in which you aren't at all like Don.

> etc.). (See Mr. Tobin's comments about my 2-1-2 IRO fails example).  
> 
> Can a plain Condorcet supporter guarantee that there will not be any
> strategic voting attempts made if polls show first choices to be such as
> 34-33-33, 49-49-2 or 49-26-25 (or with 4 candidates--- 27-26-24-23,
> 49-49-1-1, 49-18-17-16) ?  

If voters are sophisticated & strategically-inclined, and if it's
known that some voters would like to steal the election by
offensive strategy, the problem is that, in Condorcet, they
can only succeed at that with the help of their victims. Somehow
I doubt that they'd receive that help. You can't assume well-
informed offensive strategizers, and ignorant sappy victims.
If that's a known possibility, & if voters know about strategy,
as they'd have to, for it to be a problem, then the strategy
would be thwarted.

Hugh said the Middle voters would have a problem,  but not as
much as the would-be order-reversers. As I said, it's a game of
chicken in which the Middle voters have the upper hand.

As I also just said in a long message yesterday, if the voters
were sophisticated & strategically-inclined enough to have
order-reversal a possibility, then, if others didn't like that,
they could adopt added features that would thwart it. Or they
could just use defensive strategy. Again, don't judge Condorcet
against the standard of perfection,or Arrow's impossible criteria,
etc. Rather, compare it to the heavily-flawed other methods, like
your Condorcet-Approval, with its big-time strategy problems.

> 
> Plain Condorcet elections for offices without incumbents will obviously
> produce lots of candidates who will be happy to play the strategic voting

Obviously? You're the only order-reverser I've heard of so far.

> game (especially if the winner can appoint the losers to some high salary
> public office if the losers tell their supporters to vote their high next
> choices (especially second or third choices) for such winner).  Would this be
> a form of candidate bribery ?

With any method, people unlikely to win could ask their supporters to
vote for someone more likely to win. Yoo-hoo...It happens all the
time now. With Condorcet there'd be no need for it. But regardless
of the method, a crook could throw an election, though it isn't
clear why his voters would co-operate.

> 
> The obvious first question for public focus groups is- do you want an
> election method for executive and judicial officers that elects a majority
> winner or a minority winner ?  
> It takes no great genius to guess the public's probable response.

Then they won't want your proposal, which encourages--forces--
people to vote insincerely, and would thereby be as likely as
not to elect a minority winner by a forced, insincere majority
vote.

One way to talk about getting rid of the LO2E problem is to
say that a majority group can get its way without falsifying
their preferences. In Condorcet they can, but not necessarily
in your method. Your addition to Condorcet would screw up
majority rule.

> 
> Plain Condorcet supporters would apparently ask the question-- do you want an
> election method for executive and judicial officers that elects the least
> evil of the candidates ?

Maybe you were out to lunch when I talked about standards, such
as majority rule.

> 
> I repeat again-- rankings (1, 2, etc.) show only relative support and not
> absolute support (such as on a +100 percent to -100 percent scale).   Many
> candidates are totally unacceptable to many voters (especially for higher
> power offices- U.S. President, Governors, Mayors).  
> 
> Rankings for such unacceptable candidates are basically false.  
> Example- Voter D votes
> 1. John Q. Moderate
> 2. A. Hitler
> 3. J. Stalin
> 4. G. Khan
> Voter D regards candidates H, S and K as unacceptable.  It is obviously
> possible that a vote for H or S will elect H or S (combined with votes of
> other voters- especially the first choice voters for H or S).  If H or S is

No that isn't obvioius. If the JQM voters list GK 2nd that won't
change the fact that JQM beats everyone, as he would in a real
election. We don't have to go into order-reversal offensive &
defensive strategy again here, do we?

> elected, then H or S will claim a people's mandate to be a "strong" ruler
> (read killer- tyrant).
> 
> It is bad public policy (rather than using some much worse derogatory
> adjectives) to have a candidate who may be unacceptable to a majority in a
> yes/no vote to be able to be elected using plain Condorcet.

I've mentioned how your propsal does, with regard to majority rule.

> 
> Thus, if head to head is to have a chance to be adopted as an election method
> (especially against plain approval), then I suggest that it must be combined
> with a majority yes/no vote. 
> 
> Here again, I suggest that executive and judicial officers be nominated for
> and elected in nonpartsan elections and that such officers have no
> legislative powers whatsoever (based on the very long and bloody history of
> killer tyrants having combined legislative and executive and judicial powers-
> e.g. Hitler who had supreme legislative, executive and judicial power in
> Germany after 1934, Saddam in Iraq currently, etc. etc.).  
> 
> Election reform goes along with separation of powers reform.
> 
> .-
> 


-- 




More information about the Election-Methods mailing list