[EM] IRV questions...

[James Green-Armytage]

James replying to Bart...

Russ:
>>>Actually, as simple as DMC/RAV is, it may still be too complicated for 
>>>public acceptance within, say, the next 20 or 30 years. If so, I guess 
>>>we're left with Approval.
James:
>> 	What exactly do you mean by that? That approval is the only method that
>> is both immediately adoptable and better than the status quo? I'd
>suggest
>> that IRV fits that category as well, especially if it allows for equal
>> rankings. I still think that I'd rather have ER-IRV than approval.
Bart:
>
>I'd probably take ER-IRV(whole) if offered, but it doesn't seem to be on 
>the table anywhere.  

	Yes, ER-IRV(whole) does not seem to have a lot of strong supporters at
this time. It's kind of an interesting method, and maybe it deserves more
interest. It may well be the better method of the two. I tend to assume
that ER-IRV(fractional) is the more likely to be adopted, though, in that
is more in the general spirit of STV and IRV, and more obviously in the
spirit of one-person-one-vote, hence a more conservative proposal.

Bart:
>At least approval has the benefit of some highly 
>credible advocates, much study, and fairly wide use in private elections.

	To me, approval voting remains something of an enigma. It seems to work
well or poorly depending on one's assumptions about voter strategy.
Approval advocates make some interesting promises, but it seems very hard
to tell how many of those promises would come true if it were adopted in
large scale contentious elections (which are quite a different thing from
private elections).
>
Bart:
>I tend to see regular IRV as a step backward, at least in the U.S. or 
>other places currently using Plurality, when considering the strategic 
>implications of adopting it.  Although IRV may seem to be an improvement 
>over Plurality in that IIA violations are reduced by roughly half (in 
>situations not affected by nomination strategy), this doesn't imply a 
>corresponding increase in third-party wins, or in true multiparty 
>competition.  There doesn't seem to be much evidence of either.
>On the other hand, adopting IRV would likely displace or delay more 
>productive change.

	That is a possibility. The opposite is also a possibility. It's an
interesting question to wrestle with, but I think that I am a bit more
optimistic about it than you, at least at this point in time.
>

James:
>
>> 	I think that Condorcet methods in general will have a better chance of
>> becoming a public issue as IRV is used more often. If there are multiple
>> cases of real elections where the IRV winner differs from the Condorcet
>> winner, then I think that this could be used to make a case for
>Condorcet
>> (or perhaps CWO-IRV) that large numbers of people will be able to
>> understand.
Bart:
>
>Not likely to happen.  Where voting and nomination strategies prop up 
>the two-party system, any possible IIA violations will be masked.  

	Well, it's an interesting question. At some point I'd like to get a hold
of Australian data, British data, etc., and take a look. If there was a
non-IRV-winner CW, I imagine him or her as a centrist. In the US, I
imagine some sort of 'compromise party' running, such that it gets about
10% of the first choice vote, but is preferred by Republican voters to
Democratic candidates and vice versa. 
	I don't see exactly why this wouldn't happen. After all, John Anderson
ran on a centrist platform in 1980. Although Reagan won with a majority of
the vote, it's possible that a significant people whose actual preferences
were Anderson>Reagan>Carter voted for Reagan. Thus, it's possible that
Anderson was a sincere CW, and that this would have been revealed in an
IRV count.
	In general, I don't see a strong nomination disincentive for centrist
candidates to enter the fray. Do you? If so, what is it?

Bart:
>It 
>might be possible in local elections such as San Francisco, but 
>unfortunately SF's three-rank limit will have a similar affect.

	Yes, that's a rather ugly limitation.
>
Bart:
>I have a hard time accepting IRV as a first step in the direction of a 
>better system.  Even if it were true, it would be hard to show that it's 
>worth the delay compared to adopting the system you really want.
>
	Well, in order to figure out whether that's the case, I first have to ask
what is the system that I really want. 
	First of all, I'm pretty sure that approval voting isn't the system that
I really want. If it was, then I can see how IRV wouldn't be attractive as
a stepping stone.
	The system that I really want, at present, is a combination of cardinal
pairwise for executive elections, a Smith-efficient proxy system
(preferably a cardinal pairwise tally or some other anti-strategic
safeguard) for direct-issue voting, and a relatively high-magnitude STV
tally for legislative elections.
	As to the proxy system, yes, I don't think that IRV is a necessary
prerequisite, although I don't think that it would get in the way at all,
either. I think that the best way to ease into the proxy system is to
start with little or no legal bindingness of results, and then gradually
increase bindingness as trust in the system increases. I think that the
tally should certainly be Smith-efficient from the beginning. There is no
need to mess around with IRV here because it would be an entirely new
institution, without the precedent of runoffs or plurality, and because
it's easier for people to see the need to compromise on issues than
candidates.
	As to the STV system, I think that IRV can only help its chances of being
adopted.
	As to cardinal pairwise for major executive elections, I expect that it
is simply too ambitious to jump straight to that from plurality and
runoffs. I expect that there will need to be some kind of transitional
system. There are other possibilities besides IRV, of course. These
include WV (with a Smith-efficient tally such as SD, SSD, beatpath, ranked
pairs, river), and CWO-IRV. I like these significantly better than IRV,
but they are not quite my first choice. 
	Is it possible to jump directly from runoffs to a good Smith-efficient WV
method? In some places, perhaps it is. And I'd be happy to see (and
participate in, when practical) efforts toward this end. I don't mean to
say that Condorcet advocates (or approval advocates) should put down
Condorcet and only promote IRV. If there is an opportunity for Condorcet
somewhere, I say go for it!!! My point is that I'm happy to see IRV make
gains as well. I don't see IRV efforts as delaying Condorcet efforts. The
USA is a big country, and it's quite possible for us to use IRV in some
places, Condorcet in others, and perhaps approval in others. Each of
these, in my opinion, would contribute to the greater good of shaking up
the status quo, advancing public understanding of alternative voting
methods, and moving us toward a more authentic democracy.

Sincerely,
James