Validity of Vote (SWC)

[Mike Ossipoff]

Sure, those who vote are self-selected, since they're the ones
willing to vote. But surely that's unavoidable in any election.

In the same sense, EM is, itself, self-selected. Some members of
of ER indicated that they'd rather have recommendations from
the group interested in discussing relative merits of methods,
instead of having us discuss the topic on ER. So recommendations
were asked for from the EM group.

So I claim that a vote in EM on sw methods, for the purpose
of a report to ER, a report on the rankings, their results
by various proposed count rules, and why the various count
rules produced the results that they did, is valid. It
needn't result in a single recommendation, to be chosen byk
some particular favored count rule. But, as I said, it could
be helpful to ER if we were to hold a 2nd balloting between
the winners by the various proposed count rules--but never
mind that now. Right now what we need is rankings from EM
members of some of the methods that have been propsed to
EM, and which are claimed to be good methods to propose
to the public.

***

Sorry I left the WB/Condorcet proposal out. It was inadvertant.

***

We should be able to change our rankings, shouldn't we? For 1 thing,
now that I've been reminded proposals l'd left out, I'd like to
revise my ranking to include them. Besides, if, after several people
have voted, it becomes clear that it would have been better for you
to vote differently, then it would be good to be able to. 
A "visible & re-vote-able" election does much to get rid of
strategy dilemmas, no matter what count method is used--not that
there's likely to be a need for that with some count methods.

***

My new ranking of methods:

1. plain Condorcet
2. Smith Condorcet
3. WB/Condorcet
4. WB/Smith//Condorcet
5. Approval
6. Saumur's circular tie solution by lot.

***

My reason for ranking the WB/Condorcet methods below the
ordinary Condorcet methods is that I'm concerned that 
it wouldn't be desirable to seem to be endorsing the point-count
standard to the public, because it could send a mixed message.
But because WB/Condorcet's & WB/Smith//Condorcet's results
are just as good as those of Condorcet & Smith/Condorcet,
I've ranked the WB/Condorcet methods next.

My reason for not including the Schwartz methods (which were
discussed on a different list) is that the Schwartz set is
more complicated to define than the Smith set (and that's
always a problem for a public proposal), and that in public
elections the Schwartz set will be identical to the Smith
set anyway, due to the large number of voters, and the consequent
unlikliness of pairwise ties.

***

I've "truncated", by not ranking all of the alternatives. I didn't
do that strategically--in fact, in the count rule that I prefer,
truncation can't possibly result in strategic gain. The reason
for my truncated ballot is merely that all the methods that I
consider undesirable are, for all practical purposes, equally
undesirable, it seems to me. So there'd be little point ranking
them according to gradations of undesirabililty.

If I thought one of the methods that I consider really undesirable
were Condorcet winner, then I'd have a practical reason to rank
it--but that doesn't mean that I would do so, if I don't consider
the methods I haven't ranked to differ significantly in merit.

***

Why did I rank solution-by-lot over Copeland & Regular-Champion?
Because a solution by lot seems better to me than a solution that's
influenced by how many candidates the various parties can afford
to run. 

***

Why Approval over circular-tie-solution-by-lot? In Approval
a majority who vote A over B automatically ensure that B
won't win, without ever having to vote A over anything they
prefer to A. Not true with solution-by-lot.

Approval is deceptively simple, not providing for rankings, but
it's surprisingly good, compared to most methods.

Besides, the simple Approval method would probably be easier
to sell to voters, a simple extension of Plurality, than would
election by lot. A proposal to use lots would be very vulnerable
to criticism.

***

Mike




--