[EM] PR in student government

[Abd ul-Rahman Lomax]

At 11:20 AM 4/17/2007, James Gilmour wrote:
> > Abd ul-Rahman Lomax> Sent: 17 April 2007 15:50
> > The ballots could also be counted sequentially, as needed. I dislike
> > this, because I think every vote should be counted, even if
> > supposedly "moot." If I went to the trouble to cast it, it shouldn't
> > be tossed in the trash!
>
>This is an understandable "social choice" interpretation of ALL the
>information on ALL the ballot papers.  But that is not what STV-PR is
>about, and certainly not where it came from.

Mr. Gilmour may not have understood what I wrote. I did not intend to 
indicate that the lower ranked votes should be used to elect winners.

>   You vote for your first
>choice.  Your second preference is a contingency choice, to be brought
>into play only if your first choice is already elected and cannot
>proportionately represent you as well, or has so little support that
>he/she has no prospect of election and is excluded (eliminated).  And so
>on.  And of course, originally it was "your vote", i.e. your whole vote,
>that was transferred.

The system asks me to rank contingency choices. I'm saying that the 
information provided shouldn't be left uncounted. It's entirely 
another matter if it is used to elect. What I'm saying is that there 
is a social value in counting all the votes, there is a cost to 
obtaining this value (voters have to fill out the ballot in order to 
generate the source data), and those lower ranks matter.

Now, where an environment requires voters to rank all candidates, 
it's a bit of a different matter. Still, it would be valuable 
information to know that such and such a candidate ranked *last*.

"My vote" includes all the information I enter on the ballot. "My 
electing vote" is extracted from that information and is used to 
create a winner.

I didn't claim that this information was "what STV-PR is all about." 
It is primarily a method for creating a proportional representation 
assembly. The information I'm talking about is not directly relevant 
to that goal. But, I assert, it should still be made available. If it 
is determined that public funding shouldn't be spent on that, then 
ways could be provided for private funding (such as through 
nonprofits or media) to cover the costs of counting. If nobody is 
sufficiently interested to count all the votes, that's another 
matter..... but I think there would be interest.

(I have proposed that ballot images should be public record, and that 
this could solve many security issues. It does raise, potentially, 
problems with vote-buying, allegedly a serious issue -- I doubt it 
--, but those problems already exist. Essentially, incumbents or 
those who are connected with them, in general, already have access to 
the ballots themselves, and either vote-buying is moot -- in which 
case the vote buyers wasted their money, as they might deserve -- or 
it was effective, in which case those who benefited from it may have 
preferential access to the ballots. And it's possible to handle 
ballots and ballot imaging in a way that would make reliable 
identification of ballots sufficiently difficult. If ballot images 
are public, then anyone can count votes; and part of the proposal was 
that ballots be serialized -- after being cast -- so that it becomes 
trivial to combine counts from many people to produce reliable and 
verifiable overall counts.)

> > If I was a candidate for office, and it turns out that many people
> > voted for me, but not at a high enough preference for me to be
> > elected, I'd hate not to know this! The result might actually be
> > encouraging. Or not, depending on what is in those buried votes....
>
>The problem of excluding a "Condorcet winner" is unavoidable in STV-PR
>so long as we give an absolute undertaking to every voter that under no
>circumstances can a lower preference count against a higher preference.

That's right. But I was not addressing this issue at all. I think 
that Mr. Gilmour did indeed misunderstand my comments.

>Most proponents of STV-PR regard that undertaking as extremely
>important, and that view is, in my experience, shared by the
>overwhelming majority of the electors with whom I have ever discussed
>STV.  Once you change that solemn undertaking to save a "Condorcet
>winner" from exclusion, you open the door to tactical voting which is
>otherwise impossible in real STV public elections, i.e. with large
>numbers of electors whose preference patterns you cannot possibly know.
>This exclusion rule makes STV-PR non-monotonic, but that is not
>generally regarded as important and certainly nothing like so important
>as ensuring that a lower preference can never count against a higher
>preference.  Also, the non-monotonic effect cannot be exploited by
>either the candidates or the voters, so it is of no practical effect.
>It would be nice, but we cannot have it all  -  at least, not all at
>once!!

I wasn't proposing *any* change in the method. Only in procedures and 
practices *around* the method. From what I understand, it is rare 
that complete ballot information is available, and an additional 
undesirable result is that it becomes impossible to do any analysis 
of the election to determine how changes in voting methods would 
affect the results.

(Actually, I've seen some people propose that as a desirable result, 
i.e., allegedly riots would result if people found that some flaw in 
the election method produced a poor result. So these would conceal 
the truth in favor of public order, a position I have always found to 
be short-term questionable and long-term just plain foolish. A public 
order that depends on hiding the truth is unstable.)

I've argued that Range ballots could be used, allowing candidates to 
be rated rather than ranked, *without* changing the election method. 
The purpose would simply be to study the results, with a 
consideration, perhaps, to reforms down the road. Range ballots 
contain full ranking information (I'll neglect the issue of identical 
ratings for the moment), but the reverse is true only to a very small 
degree. In an election with a very large number of candidates, we 
could theoretically infer ratings from rankings, because we can 
assume that candidates exist on a spectrum, and that spacings are 
roughly even, if there are enough candidates. But voters wouldn't 
want to fill out such a ballot, and it's likely that many of the 
candidates would be unfamiliar....

Because such a system would not change the legal aspects of 
vote-counting, it could be tried on a small scale even in a 
large-scale election. It's simply a different ballot configuration, 
it shouldn't bias the results in any way. And, of course, all the 
ballot information would become relevant. Range, like Condorcet, does 
not neglect any votes.

I will not address the wisdom, or lack thereof, of this alleged 
"absolute undertaking" with the voter that their second-place choice 
will not interfere with the election of the first. For one thing, 
it's possible to design ballots that allow the *voter* to choose 
whether or not this could be the case. Election system designers have 
a tendency to consider that they know better how to amalgamate 
preferences than do the voters themselves.... but this is a complex 
issue, and I'm not taking it up here.

(But I will note that the idea of such a promise is actually the 
major argument made for IRV here -- single winner. And most of us 
consider that method defective. Again, I've written, the defects in 
IRV become much less objectionable when the method is multiwinner, 
because many more of the lower-ranked votes end up being considered. 
Single-winner, they could be doozies!)