[EM] Why I think IRV isn't a serious alternative 2

[Abd ul-Rahman Lomax]

At 06:54 PM 12/25/2008, James Gilmour wrote:
>Abd ul-Rahman Lomax   > Sent: Thursday, December 25, 2008 8:32 PM
> > > At 09:55 AM 12/25/2008, James Gilmour wrote:
> > >Abd, you are a great wriggler.
> >
> > Thanks. I'm not a butterfly to be pinned to your specimen board.
>
>Abd, I don't want to pin you or anyone else to a specimen board.

"Wriggler" implies resistance to being "pinned down."

>I just don't think it advances a discussion about major public
>elections to bring in arguments that MAY have some validity in a 
>totally different context.

I happen to think the opposite: "Major public elections" fail because 
we *don't* use what we know how to use on a small scale. Refusing to 
look at this, which is essentially what James seems to be advocating 
here, isn't helpful.

In general, poor voting systems result in increased alienation 
between citizens and government. It's so common that most of us seem 
to think of it as perfectly normal. Yet it isn't that way in 
small-town direct democratic government, or even nonpartisan 
representative government, normally, when the scale is small. I 
believe that it's possible to bring the situations closer together.

A small step, not enough by itself but certainly an improvement, 
would be voting systems that collect the kind of information that 
citizens share when they, formally or informally, negotiate solutions 
for common support. And that is *only* a Range ballot, actually, it 
*allows* accurate representation of relative preference strength. 
Find a way to pin the relative strength to absolute, commensurable 
strength, and it would be ideal, but my sense from what I know about 
voters is that it will be *in effect* closer to this, and when 
special elections are involved, where the voter has to invest some 
significant time to vote, has to have a certain level of caring about 
the result to be motivated to vote, it will, even with relative 
expression, move toward what absolute utilities would show.

In other words, voting systems should not be seen abstracted from 
general theory about how humans decide how to communication, 
coordinate, and cooperate. The ideal voting system for large-scale 
application would simulate this as closely as possible. My discovery 
has been that runoff voting is an important part of this, even 
though, with a good system, it shouldn't be necessary for most elections.

>   And small direct democratic situations,
>run-offs and write-ins are all completely different contexts from 
>that in which the discussion about the political acceptability of
>strong and weak Condorcet winners was set.

The issue was pure horse pucky. It's impossible to discuss with any 
deep knowledge the acceptability of Condorcet winners without 
examples of the specific context. It's absolutely true that there 
might be a problem with a "weak" Condorcet winner, because of 
truncation. This would be detected by a majority support requirement. 
(It's possible to set the necessary threshold below a majority, where 
the Condorcet Criterion becomes the Majority Criterion, particularly 
where voting patterns -- which could be defined -- make it a 
practical certainty that a face-off would award the election to the 
Condorcet winner.)

Now, a 5% first preference Condorcet winner with a majority support 
shown on the ballot? If it's voluntary ranking, and a majority of 
voters have ranked the candidate, not deeply, but highly? I don't 
think there would be any problem *at all*! At least the *voters* 
wouldn't be complaining! What I've seen with advanced voting system 
is that *losers* complain! Even when the result was patently fair, 
the best possible result given the votes.

> > No small community which understands the system
> > will use IRV.
>
>Then we in the UK must have a lot of small communities that do not 
>understand IRV, because, as I said in reply to one of your
>earlier comments, lots of our smaller communities use it for their 
>internal elections.

How small? How much choice do they have? And, most importantly, since 
they've seen it used in large scale elections, and it's been around 
for over a hundred years, how much have they considered alternatives? 
These are political elections. I was actually referring to 
*nongovernmental communities.* In the current environment, we are 
seeing some such communities adopting IRV. But that is in the context 
of a political campaign, and, in addition, these are communities 
where repeated balloting is considered impractical. This isn't 
"small" as I was using the word.

When you have a two-party system, which means that, in nearly all 
elections, particularly with large districts, one of two parties 
prevails, IRV works, in a sense. It prevents the spoiler effect, 
largely, in this context. Which actually negatively impacts the power 
of third parties, it appears.

Parties don't own voters. However, when there is a simple checkbox on 
the ballot which allows the voters to vote straight party ticket.... 
the ownership increases.

> > "Write-ins" are a U.S. practice, if I'm correct, we are quite
> > attached to them.
>
>Yes, I know the first and I understand the second.  I don't think 
>there is any need for them in public elections, but they are part
>of the scene in the USA and so must be accommodated in any proposal 
>for practical reform if it is to gain political acceptance.

Of course. But apparently James doesn't understand the need for them. 
They are *usually* moot. They are a safeguard against certain 
possible failures of the system. Normally, the failure has to be 
fairly large for a write-in campaign to have a chance of succeeding. 
But it happens. The write-in option gives additional power to the 
voters, not less. And some voting theorists apparently don't give a 
fig about "additional power to the voters," they imagine that they 
know better than the voters what the voters need. Voters who don't 
need write-ins don't use them!

It's like additional votes with Open Voting (Approval), or with 
Bucklin, or, for that matter, with IRV. Most voters don't need them, 
and that most voters don't use them, sometimes, has been used as an 
argument against the more advanced systems. But that neglects the 
point: Plurality usually elects the best candidate. Tweaks or reforms 
are only needed to deal with the exceptions, and a few percent of 
votes is often enough to fix an exception.

IRV, in any case, with nonpartisan elections, apparently *almost 
always* elects the same candidate as Plurality would if people simply 
vote for their Favorite. This certainly deserves study, I have a 
theory as to why it happens this way, it's not rocket science.

> > >Incidentally, my personal view is that there should be no provision
> > >for "write-ins" at all in public elections.
> >
> > Yes. You are English.
>
>NO, I am not English.  I was born in the UK and I am a subject of 
>Her Majesty The Queen (there are no citizens in the UK), but I am
>not English.

The distinct matters to you and to some, not to me, here. You are 
expressing a view that might be expected to be roughly typical for 
those from your environment. You could say the same thing about me 
with respect to write-in votes.

However, what I've seen is something that is *not* common here. 
Write-in votes allow a rough satisfaction of the Condorcet Criterion, 
somewhat Range-modified (which is good!), by top two runoff!

That is, if a true Condorcet winner, with significant preference 
strength, is eliminated in the primary, a write-in campaign has a 
very significant possibility of success, and this is even more true 
if an advanced method is used for the runoff, where voters can add 
write-ins without harm, should they be worried.

Consider France in 2002. Suppose the runoff method were Bucklin. The 
voters would already know that their favorites (most of them) were 
*not* going to win. They have already expressed in the primary their 
favorite, they've taken their stand for their party, so that true 
party strength is known. In the runoff, as it happened, there wasn't 
really any risk at all that Le Pen would win, so they'd have been 
quite free to add write-in votes for Jospin. I'd have predicted that 
Jospin would have won, a far better result, overall, than Chirac, if 
write-ins had simply been allowed. Unless Jospin actually withdrew: 
"Please vote for Chirac, not for me." And I don't see why he would 
have done that.

But with Bucklin, they could have their vote for the write-in, 
Jospin, *and* their security vote for Chirac. Some of the Le Pen 
voters would add a security vote for Chirac. But Jospin would have 
won, by a large margin, in the first preference votes. I'd predict 
that it would have been a landslide.

So are write-in votes "useless?" No, they are a possible solution to 
some specific problems. And they are used that way in the U.S., in 
real situations.

> > You are here, though, talking about
> > American elections. Almost everywhere here it is required by law that
> > write-ins must be allowed, we respect the sovereignty of the voter.
>
>Yes, I know it's US law, so roll with it  -  until you have a voting 
>system that makes it irrelevant.

It's impossible for voting systems short of Asset to make it irrelevant.

(However, very good systems may make it less than "necessary." I 
still see no reason to *prohibit* it; properly done, the 
administrative cost is very low.)

>   (In the UK, the nomination
>process for all public elections requires written confirmation of 
>the candidate's consent to his or her nomination, as do many
>organisations for their internal elections.)

So it does in California for state offices. Here in Massachusetts, 
for local offices, all that is necessary is that the candidate be 
eligible for the office and that the candidate be clearly 
identifiable. Turns out that the rules expect an address to be added 
with the candidate's name, though I didn't verify that by looking at 
the actual code.

In any case, reasonable restrictions *are* fine, as long as they do 
not unreasonably impede the right of voters to freely choose whom 
they vote for, unconstrained by what may be far too cumbersome a 
process for some situations. (Sometimes the need for a write-in 
candidacy appears in the last few days before an election. A 
candidate dies, for example.)

> > >   If I am not
> > >prepared to declare myself as candidate and be nominated in the same
> > >way as all the other candidates, I cannot see any reason why
> > >anyone should take me seriously.
> >
> > You are thinking about yourself. What about the voters? What are
> > their rights? Here, you are intending to deprive *voters* of their
> > right to free choice.
>
>Of course, I am think about you.  You might have many good reasons 
>why you did not wish to be elected to public office, either at
>that particular time or ever.

No, that's concern about a "candidate," not about the voters. If I 
don't want to accept the office, I can decline it. My rights have not 
been impacted by those votes. As I've said, I'd provide for the 
candidate who wins under write-in conditions, who has not already 
agreed to the office, the right to at least suggest a replacement to 
a legislative body that represents the people. Actually, personally, 
I'd give them the right to *appoint* a replacement. In fact it might 
be a good idea to allow this in general, for elected office holders 
to, even before elected, name an alternate.... It's better than the 
office being vacant, probably, or subject to appointment by some 
possibly corrupt official, witness the brou-ha-ha in Illinois over 
the replacement of Senator Obama. Why not allow *Obama* to name his 
replacement? Senators already function through delegation, a lot 
(delegation to staff). Obama might know someone who would make an 
even better Senator than himself! I can't see any reason not to trust 
a Senator who isn't being removed for malfeasance to name a replacement.

>   What right have I and some other voters to make you the winner 
> without even consulting you and
>letting others know about our views and of your consent by 
>nominating you along with all the other candidates?

What has happened is that the public has *asked* the person to serve. 
What's wrong with that? Do I have some intrinsic right not to be asked?

"Nomination" refers to a process which typically takes place long 
before the election. Definitely, to appear on the ballot, I should 
consent. I'll even agree that consent is better, but not for the 
protection of the *candidate*, by your entirely novel argument that 
it is somehow harmful to the candidate to be elected, but for the 
voters, to protect them from supporting someone who will decline.

(Candidates may also decline because they don't want to split the 
vote, and this could turn out to be as a result of a bad situation, 
the candidate would actually gladly accept if elected. I'd say the 
voters should be as free as possible.)

>   Even if we accept
>that voters should have "free choice", with that voters' right to 
>free choice goes responsibility, firstly to the write-in target
>(who is not a "candidate" as he or she has not been nominated) and 
>secondly to all the other electors.

The fact is that write-in campaigns for candidates who don't want the 
office have little more than a snowball's chance in hell. This is 
utterly specious and moot. And no harm has been shown should such a 
candidate accept the office. The situation might be that there were 
two candidates on the ballot, and, last minute, one of them dies. 
It's very unlikely that such a campaign would get off the ground 
without the consent of the candidate, but it might happen under some 
odd circumstance, the write-in *target* we will call her, is 
travelling in deepest Africa at the time, and those close to her 
think she'd accept. So they do it. If they succeed, and she declines, 
nobody is worse off. However, it does depend on the rules for 
deceased candidates..... if a special election is held, it might be 
better to go ahead and vote for the deceased candidate, and let this 
woman run in the special election, if she cares to. If she could 
possibly win as a write-in, she'd almost certainly win the special 
election. So it's really the same situation: she has the opportunity, 
and she can decline to accept it.

> > You and many others, by the way, dislike of
> > free democracy is common among some voting systems theorists
> > and activists.
>
>You have jumped to several unjustified conclusions here.

Perhaps. Prove it by widening your consideration a bit, to see why 
such a claim might be made.

>However, my voting reform campaigning has been within a system of
>representative democracy, and the discussion to which I was 
>contributing was also in the context of representative democracy.

*Of course.* However, the alternatives are not only "representative 
democracy" and "direct democracy." And there are many forms of 
representative democracy, some more "representative" than others. 
Some actually could shade into direct democracy, keeping the benefits 
of representation while at the same time being *more* "direct."


>So
>alternative systems of democracy, whatever their merits, were hardly relevant.

You do not unilaterally decide what is relevant. I'm claiming that 
the practices and traditions of direct democracy, which go back 
thousands of years, *are* relevant to representative democracy and to 
the methods we use to create representation. "Election" is actually a 
poor way to create representation, certain assumptions are 
incorporated that apply little to our time. In particular, it leaves 
many and even most people unrepresented, some not at all, some only 
through poor compromises.

PR systems, as actually implemented, don't do a lot better, mostly 
forcing people to choose between very limited choices. But better PR 
systems are quite possible and, James, it was an Englishman who 
suggested the tweak to STV that would truly make PR almost perfect. 
Lewis Carroll, who also wrote, early on, some of the most detailed 
examinations of STV that came from his time. He was *way* ahead of his time.

>We have managed to make some significant
>improvements to the voting systems we use in our representative 
>democracy in the UK and I am hopeful of seeing some more.

What defines "improvement"? How do you know that you have actually 
improved a system or have merely made it safer for the existing oligarchs?

>   But the
>replacement of our system of representative democracy with some 
>other system of democracy will not be achieved in my lifetime, no
>matter who campaigns for it.

Actually, there is a fundamental error here. It is not necessary to 
replace existing system. They actually work quite well enough. The 
important reform is the direct organization of the people to 
generate, for themselves, reliable advice as to how to vote.

With the right advice, even Plurality works quite well enough.

>I therefore prefer to concentrate my remaining energies on achievable goals.

Sure. Me too. My work with voting systems is a device, a means of 
approach. My real concern is how large numbers of people seek and 
find consensus, and consensus is powerful. Find it, you can do almost 
anything. Without it, everything is a struggle, you might win, you might lose.

I think I know how to do it. Asset Voting happens to be a device that 
I did *not* invent which could lead to the usage in public elections 
of the principles I'm promoting for nongovernmental, advisory 
organizations. I don't know whether Asset is a realistic possibility, 
it might not be. But FA/DP, the delegable proxy, free association 
equivalent, is, I'm certain of that, the only question is how long it 
will take. I predict: it, or something better, will happen. It 
*might* happen in my lifetime, it's possible. And it might not. But 
there are far too many trends converging on it for it to not happen 
at all, unless some major disaster intervenes.

> > >  If my "friends" think I would be the best person to do the job,
> > > they should come and tell me and
> > > persuade me to stand, nominate me, and then campaign like fury to
> > > get me elected.
> >
> > However, what if you were all supporting a candidate, and after the
> > deadline for registration, that candidate dies.
>
>UK election law has provisions that cover that eventuality. For 
>local government councils, the election is cancelled and a new
>election must be held within 35 days of the date of the original 
>election.  (I haven't checked the rules for Parliamentary
>elections, but they'll be similar.)

Sure. You can make rules that *substantially* satisfy the need behind 
write-in votes. However, if those rules are absent? Sometimes they 
are, you know. Further, top two runoff is *much* safer if write-ins 
(which might mean registered write-ins, which answers your major 
objection about consent) are allowed. There are very good reasons not 
to put this third candidate on the ballot. It's *good* that a 
write-in candidate normally can't succeed. It tests preference 
strength. Otherwise one of the top two is *usually* good enough. 
Results don't have to be perfect.

Ideally, a majority should be required for any binding decision; 
that's democratic tradition in direct democracies, which includes 
assemblies of peers, going way, way back to the earliest English 
examples. But with public elections we sometimes make a compromise, 
and top two runoff postpones the compromise, usually, to the runoff. 
So it is *less* of a compromise.

Top two runoff has *long* been considered a major election reform, 
and it has been.

> > Or there is some huge
> > scandal and he becomes unelectable. Why shouldn't you and your
> > friends be able to mount a last-minute write-in campaign.
>
>In the UK, no candidate may withdraw after the close of nominations, 
>so this is a theoretical possibility.

But a candidate cannot be forced to serve.

>I don't know off-hand
>how frequent such post-nomination problems have been in the UK.  We 
>certainly have had situations where a nominated candidate has
>withdrawn and been replaced, but again, I couldn't give 
>numbers.  Most of our problems seems to get sorted out (or exposed) before
>we get to the nomination stage.

It's happened in the U.S., and recently, that a candidate was tainted 
seriously by immediate scandal and became unelectable. (And it's also 
a testament to the persistence of incumbents that candidates have 
been re-elected *after* such a scandal, one that theoretically should 
have made them unelectable, they certainly wouldn't have been elected 
in the first place if something like that had been very public about them.)

When a candidate becomes unelectable after the ballots have been 
fixed, this can simply award the election to the major opponent. 
*Unless* write-ins are available.

> > Write-ins have been used to preserve the power of the voters against
> > the power of legislatures or city councils to decide how voters 
> should vote.
>
>Making sure that the legislatures and city councils are properly 
>representative of the voters will give better and more effective
>projection to the voters than any provision for write-ins.

Sure. Greater good is better than lesser good. So?