[EM] Smith PC doesn't have any serious problem. Specify pairwise count first. PC dfn.

MIKE OSSIPOFF nkklrp at hotmail.com
Wed Jan 28 05:27:02 PST 2004

Smith PC doesn't have any important faults. As other wv versions have been 
proposed, we've tended to be perfectionists, and so Smith PC has gone out of 
favor. But it's still a very good method.

Smith PC's disadvantages:

It doesn't meet GSFC or SDSC.

Others will name other criteria, but defensisve strategy need is my voting 
system concern, and so those 2 defensive strategy criteria are, for me, the 
things that separate Smith PC from SSD, BeatpathWinner/CSSD, MAM, & SD.

I agree that Smith PC may very well be the right compromise between 
simplicity & merit.

The advantage of Smith PC over plain PC is that, because Smitih PC meets the 
Smith Criterion< it also meets condorcet loser> this computer"s upper case 
characters aren"t working again>

Smith PC meets SFC & WDSC. SFC is worth a lot, it seems to  me. PC meets SFC 
too, but Smith PC is less vulnerable to criticism because it passes 
Condorcet Loser. I agree with Anthony Duff that Condorcet Loser isn't 
important. It could be used as emotional criticism against PC. The only 
reason to propose Smith PC instead of PC is to avoid having to deal with 
those criticisms. But we don't know if that Condorcet Loser criticism will 
materialize in an enactment campaign.

The Libertarian Free State Project uses PC. They wanted something especially 
briefly defined, because, all else being reasonably equal, a brief 
definition is easier to justify. Maybe we should listen to them for public 
proposals too. There's so little experience that it's impossible to say.

I have to report this: Around '93 or '94 I proposed Smith PC on a newsgroup. 
I was accused of trying to teach them set theory. Then I changed the 
proposal to PC, and the same person who'd made that accusation congratulated 
me for finding a winnable proposal.

Maybe we get too involved in the search for ultimate merit, and maybe that 
takes us too far from what's most proposable & winnable.

On the other hand, at the other extreme, there are some people who are about 
to propose MAM for public elections, and I haven't tried to convince them to 
change it to Smith PC or PC, because I figure if they like it, then maybe 
the people they propose it to will like it too. When people want to propose 
a method, that means it makes sense to them, and that's promising for its 

Also, people do a better job of explaining and advocating what they like. So 
if those people like MAM, then I hope they succeed with it. But if there's 
much objection to complexity, then one should immedaitely go to something 

Duff's point was well-taken that we put too much emphasis on the circular 
tiebreakers, because those are what distinguish all these methods we 
propose. It's quite true that the important message for public advocacy is 
the pairwise-count itself, and the election of a CW if there is one.
I'm not saying that CC compliance is enough. But the pairwise count 
principle is a good first step when introducing these methods. The circular 
tie solution is what gives the method further properties and advantages, 
beyoned CC, but maybe the pairwise-count should be the up-front offering.

As was suggested, that should be the main offering, and the circular tie 
solution should be offered as a footnote.

My PC definition starts by saying "If anyone is unbeaten, then they win and 
the count ends."

So that definition starts out by looking for the unbeaten candidate if there 
is one. That will make plausible sense to a public audience.

Now, BeatpathWinner and MAM can be defined without mentioning electing the 
unbeaten candidate if there is one. But maybe that should be what comes 
first anyway, so that people will know at the outset that we're talking 
about a pairwise count method that will always elect someone who has no 
pairwise defeats. Yes, when we just give BeatpathWinner & MAM definitions, 
the pubilc audience isn't getting that important introductory orienting  

I emphasize that SSD starts as PC does, by saying that if anyone has no 
pairwise defeat they win.

SSD also, among the deluxe public proposals, is the one that doesn't mention 
cycles, even obliquely.

There was discussion about how to word pc> to me, the simplest definition 

If anyone is unbeaten, they win and the count ends.
Otherwise drop the weakest defeat. Repeat till someone is unbeaten.

I claim that dropping the weakest defeat is by far the most natural thing to 
do, and that that will be more obvious to people than speaking of the 
candidate whose greatest defeat is the least.

Mike Ossipoff

Scope out the new MSN Plus Internet Software — optimizes dial-up to the max! 

More information about the Election-Methods mailing list