[EM] Condorcet strategy and weighted pairwise method

James Green-Armytage jarmyta at antioch-college.edu
Tue Jul 20 14:41:44 PDT 2004


This is James Green-Armytage replying to Dave Ketchum.
>
>
>The method cannot avoid introducing pain.  While I have the option of 
>refusing to do ratings:
>      I cannot do that intelligently without understanding the option.
>      Since other voters could use the feature, I need to understand what 
>they could do to me with it.
>      If I get in deep enough, I start wondering how default ratings 
>interact with truncation (I see that you do describe one variation - 
>bullet voting).

	Sure, any method that is more complex than plurality, two-round runoff,
or approval is a challenge to voters in terms of understanding exactly how
their vote will be processed. Even IRV is relatively challenging for
voters to understand, if they've never dealt with stuff like that before.
Any Condorcet method is challenging, as well... to fully understand it,
voters need to learn about pairwise comparisons, the Condorcet paradox,
and the particular cycle-breaking technique. My method probably adds an
extra layer of complexity on top of that, I admit, because the
cycle-breaking technique is relatively complex.
	However, the happy part is that, in my opinion, you don't need to be
strategically sophisticated to cast an effective vote in weighted
pairwise. If people rank sincerely, and rate sincerely, the result will be
a good one. I think that there aren't a lot of situations where a
sophisticated vote will have vastly more "power" than a sincere vote, if
you keep the definition of power in probabilistic terms, such that you
assume that voters will not know what the direction of cycles will be if
they arise. 
>
	Anyway, I concede that this is a relatively complex method. Whether the
complexity is worth it depends on your situation, I think. Will the voters
be unwilling to learn the rules a complex method, and be hostile to a
method that they can't understand? If so, weighted pairwise might not be
the best bet. At this point I don't know how soon I can hope to see
weighted pairwise. 
	I think that the first question for me is whether it is superior to
rankings-only Condorcet in an electorate that is not hostile to complex
methods. That question is important to me in itself, because I like to
peer into the future and imagine what voting methods people will be using
in a much more democratically-advanced society.
	As for what to promote if people are hostile to complex election methods,
I don't know exactly... That's a whole 'nother question, one which sort of
ties your hands as a voting methods designer.
>
>> And my answer to the question
>> is, sure, it's a small price to pay.
>
>Let's see:
>      "small" requires comparing benefit vs cost, and I remain suspicious 
>on both ends.

	Okay, well, specific arguments as to this are welcome. Again, I think
that the only substantial cost is getting computerized systems with an
interface that allows you to enter numbers. Once the software is produced,
the cost of providing copies to the various precincts is virtually nil. 
	I've already talked about the benefits. First, it has all the benefits of
a Condorcet method, since it is a Condorcet method. Second, it resolves
cycles in a more meaningful way than rankings-only Condorcet, because it
takes into account strength of preference. Third, it may prevent strategy
from getting out of hand in Condorcet, which I'm still convinced is a
significant issue.

>      Argue paper trail another day.

	There's actually an argument against paper trails for voting machines?
>
>With IRV, voter's first choices can get summed by state but, when a loser
>is 
>recognized, voting patterns are needed to do the substitution.  These
>patterns 
>could have been sent up by the precincts (added load - NY has a few
>million 
>ballots for governor), or could be requested as needed by state
>(computers 
>with this data must stay available).

	With IRV, you can put the FULL results for a precinct on one or two
DVD's. You can then physically carry the DVD's via courier to the central
tally location, at which point all the DVD's will be fed into a big
computer, which will look at the ballots in full and give you a result.
	If you don't want to wait for the courier to find out the winner, maybe
you could transmit an unofficial version of the full result from each
precinct via a secure internet connection, tally an unofficial winner at
the central location, and then wait for the couriers to come and enter the
DVD's to double-check the result and make it official.
>
>With your method...?
>Brian is almost certainly doing the equivalent of a precinct, where
>almost 
>anything would be doable - and the doing could be useful to his audience.
> 
>I am leaning on the problems of multiple precincts, such as states.

	Right. I can't see any reason why, if you really wanted to, you couldn't
just take the sum of precinct matrices to find state matrices, in my
method. Again, I think probably the way to do it is to have a
straightforward pairwise matrix, which is summable, plus a ratings
differential matrix, which is also summable, as far as I can imagine...
these two summable matrices together will tell you the winner.

my best,
James


P.S. the proposal for weighted pairwise is currently at 
http://fc.antioch.edu/~jarmyta@antioch-college.edu/voting_methods/weighted_pairwise.htm






More information about the Election-Methods mailing list