[EM] Lomax reply, 3/14/12

[Abd ul-Rahman Lomax]

At 03:53 PM 3/14/2012, MIKE OSSIPOFF wrote:


> >I admit that that is a mess--when my
> >optional-conditionality-by-mutuality algorithm definition
> >is in three widely-separated postings. At least I should re-post the
> >corrected pseudocode in
> >one posting. Should have already done that before now. Will within 
> a few days.
>
>While there may be value for this in terms of working on improved
>methods, as to theory, as to possible public implementations, not
>method that is so complex to explain has a prayer of seeing
>application outside of specialized societies where they are willing
>to tolerate that.
>
>[endquote]
>
>How many people have seen, or asked to see, the computer program for 
>vote-counting
>in our current elections? How many people in IRV jurisdictions have 
>seen or asked to see,
>or understood the count program for IRV?

Mike, that's irrelevant and you know it. People know how the vote 
will be used to determine winners, with Plurality or Top Two Runoff. 
They may not know every detail, but they would probably get it very 
close to correct. (i.e, what is the basis for majority? Some might 
not get this right.)

IRV has been explained, and most people, again, understand it if they 
are paying attention. The rules can be simply stated.

Some people don't care, for sure. But when a new voting system is 
proposed, people will want to at least think they understand. They 
certianly did not understand all the details when IRV was implemented 
in San Francisco, the voter imformation pamphlet lied to them. That's 
a different problem!


>People are told how IRV works, but they don't have to see the software.

I wasn't asking for the software, I was asking for the rules which 
would be used to create the software. The algorithm, if you will, and 
if an algorithm is complex, the various implications of it will be 
even more obscure. Mike, this is a political issue. Your method might 
be theoretically superior, but as a first reform, forget about it. If 
public process can be set up that, say, studies election method 
performance in simulations, and if the recommendations of the 
committee formed are trusted, maybe. But getting that is quite 
difficult, all by itself.


>AOC conditionality can be described in terms of what it does for the voter.
>
>A conditional approval isn't counted unless it is reciprocated.

What that means isn't obvious. But I assume you'll explain:

>It can be said in more detail, but a little more wordily:
>
>Call a ballot's unconditionally-approved candidates its "favorites".
>
>A ballot on which C is favorite is called a C-favorite ballot.

You have not defined "unconditionally approved." How is that shown on 
the ballot? I could guess, but I'd rather not!

>For each pair of candidates, C and D, the number of ballots on which 
>D, but not C is favorite,
>and which conditionally approve C must at least equal the number on 
>which C, but not D is
>favorite, and which conditionally approve D. Otherwise enough 
>C-but-not-D-favorite ballots' conditional
>approvals of D are ignored to achieve the above-described parity condition.

Mike, most people's eyes will be glazed over at this point. I have a 
habit of reading stuff like this with "stupid eyes." I cannot 
immediately understand what you have written. Probably because it 
doesn't "make sense." That is, a series of facts about the method, 
all new, are being presented without the *significance* being known. 
This is about pedagogy, Mike, and polemic, the same thing, really.
Now I'll put in some effort. Realize that most people will not. And 
they will dislike that the information is being presented this way, 
and they will not trust it.

Okay, I think I get it. *If* the chicken dilemma is found to be 
damaging results, it might even be useful. I still find the *meaning* 
of this, i.e., the actual effect it will have on voter behavior, 
obscure. It seems to me like the "conditional approvals" being 
counted are dependent upon the behavior of other voters. I find that 
highly suspect. My "conditional approvals" are being deprecated. It 
might be fair, because if I don't like that, I can fully approve. But 
I'd still want to be able to rank my approvals....

>But people will understand that, in examples like the one below, 
>it's good if the voter can
>make an approval conditional upon reciprocity:
>
>(If you haven't been on the list lately, you might not have seen 
>this "Approval bad-example":

Probable.


>Sincere preferences:
>
>27: A>B
>24: B>A
>49: C
>
>The A voters should approve B, and the B voters should approve A.

Why? Mike, *that depends on preference strength.* *You* may not have 
been following long-term discussions on these lists....

Okay, let's assume that the B>C and A>C preference strengths are 
solid. First of all, this is a race where A and B would, in most 
situations, cooperate; one of them would drop out, it is a very close 
election and by both running they are risking the election of C.

The scenario posits C who would win Plurality hands-down if this is 
the situation. A and B are close, relatively speaking.

Mike, you are showing a situation which demonstrates the power of 
runoff voting. Runoff systems resolve this just fine. There is 
majority failure, C is way ahead of A and B, but the leader between A 
and B will go into the runoff with C. It works perfectly, in fact, 
this will be very likely to elect A.
How about Bucklin/Runoff> You have presented a scenario where the C 
voters equally detest A and B. This kind of division of society, with 
this class of voter being *almost* a majority, is not something I've 
seen in real life. You have the A voters and B voters divided, with 
no specification of preference strength. This is the kind of voting 
system study that I've argued against for years. It has a value, but 
it is purely created to show a criterion failure or the like. Whether 
it is realistic or not isn't even addressed, often.

IRV handles this situation, of course. IRV was *invented* to handle 
this, the problem is it breaks down badly elsewhere. Now, I've looked 
at a lot of IRV elections, and I've never seen one that looked like 
this. The problem of clones (and to some extent A and B are clones, 
as to the C voters' view) is not just in voting systems, it damages 
campaigns. A and B need to cooperate to beat C.

If this is the "chicken dilemma," it's been made up. What this 
situation means (if we interpret it realistically) is that C is 
likely to win. Period. A whole lot of reality has been truncated. 
There will be write-in votes. There will be voters whose voting 
patterns don't make sense. C is within the noise of winning, whereas 
A and B voter behavior has to be about perfect.

Do we know of any real-life example of the Chicken dilemma?

How would Bucklin handle this? Do the A and B voters know the risk? 
If so, they would be likely to vote their preferences. C voters, 
would they be aware of the danger that A would win? After all, they 
also have a strong preference, as this is stated. In fact, some of 
them will prefer one of A or B. Voters are *not* identical, they 
resemble each other *statistically.*

In any system that awards an election based on plurality, C will be 
almost certain to win. Even IRV, with real voters, C's awfully 
likely. Some A and B voters will truncate.

I've never seen an IRV election that shifts preferences as 
drastically as required to accomplish the defeat of C. What is 
normal, in fact, is that the additional votes from eliminations have 
*no effect* on relative standing, in nonpartisan elections.

And if this is a partisan election, it is *really, really weird* that 
A and B are duking it out!

Runoff voting was designed to fix this. Vote splitting, among 
candidates where one of them could win if not for the presence of the 
other, will typically cause majority failure.

If this were Bucklin runoff, it might well make sense for the A and B 
voters not to trade approvals. But they would be risking that C bumps 
over the majority line.

>But what if the A voters
>approve B, and the B voters don't approve A? Then B will win, and 
>the B voters will have
>successfully taken advantage of the A voters' co-operativeness and sincerity.

People are far more alike than you might realize. If A voters betray, 
B voters also betray, they betray equally, more or less. So C wins.

That is why politicians try to avoid situations like this!

Now, look at this election if the ballot is a Range ballot....

>That's the co-operation/defection problem, or the chicken dilemma.

A false dilemma, that assumes people are playing a game different 
from what they actually play, and that society is as neatly divisible 
into factions like this. Most people won't sweat this at all!

>If you're an A voter, you'd be glad to hear that you can give a 
>conditional approval to B, an
>approval that is conditional upon reciprocity.

This is doing something with the election process, making it a goal 
in itself..... I'm not thrilled. I'd want to see how the method 
performs in simulations.

But it can be difficult to model strategy. There is a cost here, the 
cost in canvassing complexity. I'm not convinced I'd approve it.


>So, what AOC does isn't complicated to tell. People would understand 
>why they'd like it.

I'm still not convinced I really understand it. I could probably 
explain it, though, i.e., how the counts are modified. What I don't 
get is why this is really necessary. It's obviously devaluing 
information from the voter, based on some assumption that... what? 
That voters have not been properly reciprocal? But that would seem to 
assume that the A>B and B>A preference strengths are the same. They 
will not be, in general!

I think this algorithm could damage overall social utility. In fact, 
with sincere votes, it's obvious that it *will.*

The question would be whether it balances out the damage from 
strategic voting (which, because the votes are not "maximally 
sincere," does damage S.U.) I'm pretty strongly suspecting, no, it 
causes further damage by removing a strategic voting effect that may not exist.

>In any case, remember that I don't suggest AOC for a first proposal, 
>partly because the simpler
>plain Approval is simpler, and partly because AOC is to 
>computation-intensive for an easy, convenient
>handcount. At first, till a count-fraud-proof computer count can be 
>guaranteed, only a handcount
>is acceptable. The benefits of the best and most sophisticated 
>method are nil if count-fraud
>changes the result.

Well, fortunately, we agree on this. And, likely, it will be up to 
future generations.

>I don't know whether GMAT &/or MMT is suitable for handcounting.

I lose the abbreviations.

>By the way, though Bucklin was used with a handcount, ER-Bucklin, 
>with the MMC-preserving delay that I spoke
>of, is incomparably more computation-intensive than ordinary 
>Bucklin, and therefore, almost surely unsuited to
>a handcount. And, without that delay, you lose MMC compliance.

Not sure what you mean. ER-Bucklin can be hand-counted, and was (it 
was often ER in lower ranks than first). Your "delay" may well 
introduce problems. I don't know what you mean, in fact.

>You asked about what I meant, regarding that delay:

Glad I did!

>Suppose that, at your 3rd rank position, you've ranked 5 candidates. 
>Say that in round N, they get votes from your
>ballot. The delay provision that I speak of (and which is in the 
>electowiki definition of ER-Bucklin) says that
>your votes to your 4th ranked candidates won't be given any sooner 
>than they would be if you'd ranked your 5
>rank-3 candidates in separate consecutive rank positions. In other 
>words, in this example, your 4th ranked
>candidates don't get their votes from you until round N+5.

Gosh, people can make things complicated. Just effing count the 
votes! How in the world did ER-Bucklin become so complex? I, naively, 
assumed that it was *Bucklin* with Equal Ranking allowed. Who tacked 
all this absolutely hopeless crap onto it?

>If you'd ranked those candidates in consecutive rank positions, then 
>one of them would get your vote in round N.
>The 2nd would get a vote in round N+1....and the 5th would get your 
>vote in round N+4. So only in round N+5
>would your ballot then give to your next candidate.

I could probably actually understand this if I suspected it were worthwhile!

This is utterly damaging to social utility, as I see it. I see 
Bucklin as practically using a Range ballot, with an analytical 
method that slides down the approval cutoff until there is a 
majority. If voters vote sincerely, it's obvious that messing with 
the counting messes with the basic principle. Now, maybe, somehow, 
this compensates for the problem with majority-seeking in general 
(social utility optimization can violate the majority criterion).

But, you should know, I dislike overcoming a majority preference 
without the voters being explicity asked if it's okay! Or, at least, 
having expressed that, as by unconditionally approving, by a majority 
as well as a plurality, a candidate.

>As I said, that preserves Mutual-Majority-Criterion compliance, but 
>it greatly increases the labor of a handcount,
>almost surely making handcount infeasible.

And it also makes the voters dizzy when they try to understand the 
effect of their vote....


>So then, when you rank 5 candidates at rank 3, receiving your votes in round
>N,your 4th ranked candidates don't get votes from you until round 
>N+5. At that time, all of your
>4th-ranked candidates receive your votes.
>
>So, of the Approval election vote-management options that I've 
>proposed, the only ones suitable for a handcount
>would be MTA, MCA (ordinary, non-conditional), and maybe GMAT &/or 
>MMT (someone else might be able to answer whether
>GMAT or MMT would be handcount-suitable).
>
>At least for now, a handcount is the only reliable way to avoid count-fraud.

I agree. That's why paper ballots should be used (even if they are 
printed by machine -- and the voters should have that printed ballot 
in their hands for inspection, before they deposit it in the ballot box).

Sorry about not putting more effort into understanding MMC. I should, 
at least, understand the criterion itself. I'll look at that, thanks 
for your effort.

>Some people are very worried about fraud on the part of some voters. 
>What we should really be worried about
>is count-fraud.

Indeed.
(to be continued.)