[EM] Not obviously untrue

MIKE OSSIPOFF nkklrp at hotmail.com
Tue May 31 21:08:50 PDT 2005


After I'd said that I wasn't going to read your postings, you sent one to my 
e-mail address.

I didn't expect to have to do this, but I've blocked you from my e-mail 
address. And now, in case you have another e-mail account that I haven't 
blocked yet, let me now make it clear to you that I don't want any more 
e-mail from you.

I'll comment on this posting, a copy of which was sent to my e-mail address:

I'd said:

    This isn't in reply to any subjct-line.

James says:

Obviously untrue.

I comment:

In what way obviously untrue? I'd been discussing MMPO's criterion 
compliances. The discussion wasn't finished. For several days, I'd intended 
to comment on MMPO's criterion failures. As I said in my posting about that, 
any proposed method's criterion failures need to be addressed, and MMPO is 
no exception.

Maybe you think that it was necessary to find out about those criterion 
failures from you. :-)

If I read and reply to the postings of people I've said that I've posted my 
last reply to, then where's my reply to Markus's posting that continued the 
subject line of the previous discussion?

James continues:

Funnily enough, your statement above provides evidence against itself.

I reply:

How is that? I knew that you'd posted because of the subject line.

You continued:

(It clearly indicates that you read my "minmax is not a good public
election method" post, and that you were concerned with it in the writing
of your message, which addresses the same salient points.)

I reply:

The "salient points" are three criterion failures: ICC, MMC, and CL. But 
those are common knowledge, not something that could only be commented on in 
reply to you. Anyone discussing the merits of MMPO would mention those 

James apparently thinks that he was the only one who knew that MMPO fails 
ICC, MMC, and CL   So, if I commented on those criterion-failures, then I 
must have found out about it by reading James' posting. :-)

One of the nice things about declaring that I won't reply to James is that I 
don't have to read his postings. Reading the postings, so that I can reply 
to them, is about half of the unpleasantness of the overall reply task. 
There really is no reason to read what I don't reply to. It's a nice relief 
to not have to display James' postings on the screen.

Commenting on MMPO's failures of ICC, MMC, & CL means that I wanted to 
address MMPO's well-known criterion failures. It doesn't prove that I read 
James' posting.

I'd said:

MMPO can fail ICC, by means of a fratricidal majority cycle. But
candidates in a clone-set are presumably rather similar. So how come, in 
that ICC failure scenario,  they can muster more votes against eachother 
than againsts their genuine opponent?
Likewise with MMC. That majority prefer eachother's candidates to all the 
others. And yet they contribute to greater vote totals against eachother 
than against someone they all like less?

James says:

Unless you mean to imply that the majority faction should use equal
rankings, your question makes no sense.

I comment:

You say that, but apparently can't justify what you sayj.

I didn't say that the scenario is impossible. I said it isn't likely. My 
wording implied that it's odd, and I stand by that.

James continues:

A cycle in MMPO that thwarts a mutual majority can happen as a result of
bad luck

I comment:

Improbably bad luck, sure.

James continues:

and/or well-coordinated strategy on the part of the minority

I reply:

Though I didn't read your posting, I read other people's postings with the 
same subject-line, and so I can tell you that what you're saying has been 
answered. You're repeating an already-answered statement.

You apparently want to say that the minority are sophisticated and organized 
enough to engineer a strong majority cycle, but that the majority are too 
unsophisticated and disorganized to be able to equal-rank. Now you've been 
answered twice. But I don't expect that to stop you. But that's ok, because 
I don't display your postings, and you aren't going to e-mail any more of 
them to me.

I'd said:

Condorcet Loser? It must be a peculiarly popular Condorcet loser who has
the fewest people who prefer someone else to him. Not only is it unlikely 
a Condorcet Loser, but suppose one did. It would be unaesthetic to elect a 
Condorcet loser, but how disastrous would it really be to elect a
candidate over whom fewest people prefer someone else? Would Hitler, Bush, 
or David Duke win MMPO?

James says:

Who knows.

I comment:

Anyone would know that. Someone thoroughly disliked isn't going to be the 
least unpreferred candidate. Someone so disliked that he is Condorcet Loser 
isn't going to be the least unpreferred candidate. You can write an example 
of that happening, but that doesn't mean that it will happen.

As with the majority-cycle scenario, I don't claim that the CL MMPO winner 
scenario is impossible in the sense that an example can't be written.

James continues:

It's not clear how this question supports your argument.

I reply:

What can I say?

James continues:

Yes, failure of MMC and CL might be pretty rare in MMPO. But why on earth
should we chance it when methods that pass MMC and CL use the same ballots

I reply:

I've said this many times, but I'll repeat it for you again: Different 
methods sometimes give different results. That's why we call them different 
methods. And, because they give different results, they often meet different 
criteria. Therefore, one chooses which criteria are more important. 
Sometimes the choice of methods is influenced by how easily or likely will 
be a criterion-failiure.

I've talked about the likelihood of MMPO failing ICC, MMC, CL, and GSFC. 
MMPO meets FBC, WDSC, and SFC. With enhancements MMPO strictly meets SDSC, 
ICC, MMC, GSFC, and Strong FBC.

Without enhancements, MMPO meets different criteria than wv and methods 
related to wv. Remember, James, that your Cardinal Pairwise doesn't meet SFC 
without an enhancement. MMPO, in plain form, meets SFC. It's a question of 
which criteria you want to meet.

FBC and SFC  vs ICC, MMC and CL, when MMPO's failures of those latter 3 are 

FBC is the most basic guarantee to reassure the timid, and SFC is the most 
ambitious guarantee for strategy-free voting, under the conditions for which 
that can be guaranteed. Those make a good combination. Even unenhanced, MMPO 
looks good in coimparison to Cardinal Pairwise.

But, with enhancements, MMPO has everything else beaten, including Cardinal 
Pairwise with enhancement. AERLO, ATLO, and cycle collapsing don't make MMPO 
meet Condorcet Loser. But when you have a method that meets everything else 
that's important, one could inelegantly gain CL compliance by method 
definition: Add to the definition of enhanced MMPO a provision that says 
that if a Condorcet loser wins, then s/he is deleted from the ballots and 
the election repeated.

Will that CL protection cause MMPO to lose FBC and Strong FBC. Maybe, 
strictly speaking, and, if so, I wouldn't want to add it. I'd rather keep 
FBC & Strong FBC. But if CL were more important to other people, then of 
course that trade-off could be made. Of course, even then, an FBC failure 
that only happens when it involves eliminating a Condorcet loser who wins 
MMPO is one that will be very very rare, and probably not very important 
even if it happened.

James continues:

and are just as easy to promote?

I comment:

You've got to be kidding. MMPO is incomparably easier to introduce, define, 
promote, and implement than Cardinal Pairwise.

James continues:

If we're able to get beyond IRV, then why
not use SD, Smith//minmax, beatpath, ranked pairs, river, etc.?

I reply:

1. They aren't nearly as easily and simply introduced and defined to the 
public as MMPO. They don't even come close in that regard.

2. They don't meet FBC. MMPO's FBC (and, when enhanced, Strong FBC) gives to 
the more timid voters a guarantee that they have shown that they need.

As I said, Australian voters often bury their favorite, because they're 
using Plurality strategy. And no one can assure them that IRV won't make 
them regret that they didn't do that--because IRV can make them regret that 
they didn't do that.

As I said, I recently observed someone voting in an Internet presidential 
poll, and that person agreed that Nader was more honest than the Democrats 
and had better polices. But that person ranked all the Democrats over Nader. 
Lesser-of-2-evils. With Cardinal Pairwise I couldn't absolutely guarantee to 
that person that that person could never regret not burying Nader lke that. 
With MMPO I could absolutely guarantee that.

That's a guarantee that voters need, as has been shown by experience.

Aside from all that, though LNC isn't important to me, the fact that MMPO 
meets it will help in discussions with IRVists.

James says:

There is no need to implement MMPO instead of SD. SD is just as easy to

I reply:

I consider SD a good proposal. But not as good as MMPO. For one thing, SD's 
definition mentions cycles. If you haven't talked to people who are 
completely unfamiliar with voting systems, then you don't know the degree to 
which something like that puts people off.

For another thing, the pairwise count methods other than MMPO, such as wv or 
margins, need a lot of preliminary definitions. I e-mailed an RC definition 
to a newspaper editor once, and he wrote back and said that no one at the 
paper understood it. I'd thought that RP's brief definition would be 
understood by everyone. But I'd left out all the preliminary definitons 
without which someone new to voting systems won't understand most 
pairwise-count methods.

It's necessary to say:

1. X beats Y if more people rank X over Y than rank Y over X.

2. If one or more candidates are not beaten, then they win and the count 

3. Otherwise a "circular tie solution" is used to choose the winner. 
Circular tie solutions will be defined after these preliminary definitions.

4. For the purpose of that circular tie solution, an instance of one 
candidate beating another is called a defeat.

5. For the purpose of that circular tie solution, if X beats Y, the strength 
of that defeat is defined as the number of people ranking X over Y.

[end of preliminary definitions (I hope)]

These preliminary definitions are what I left out when I sent my brief RP 
definition to that editor. No wonder they didn't understand RP, without 
these definitions.

Do you still think Cardinal Pairwise and wv are as easy to introduce and 
define to the public as MMPO is?

James continued:

, it's Smith efficient, MMC efficient, CL efficient, and Condorcet

I comment:

Different methods meet different criteria. The lack of some criteria have 
worse effect on people's voting than does the lack of other criteria. 
Experience suggests that FBC is more necessary, in that regard, than Smith, 
MMC, CL, ICC, or CC.

James continues:

I don't know how to quantify IC failure, but I know that MMPO's
IC failure is worse than SD's IC failure. Trading Smith, Condorcet, MMC,
and CL for FBC and later-no-harm is a lousy trade.

I reply:

Without enhancement, Cardinal Pairwise fails SFC. MMPO meets SFC without 

So add that to the unenhanced trade.

And, enhanced, the trade is: FBC & Strong FBC vs CL.  And CL failure is 
easily avoided, when a method is good enough to justify adding a special 
rule to avoid CL failure. How easily can you avoid letting Cardinal Pairwise 
fail FBC and Strong FBC?

Mike Ossipoff

Express yourself instantly with MSN Messenger! Download today - it's FREE! 

More information about the Election-Methods mailing list