[EM] new working paper: "Four Condorcet-Hare hybrid methods for single-winner elections"

[Kristofer Munsterhjelm]

James Green-Armytage wrote:
> 
> Dear Election Methods Fans,
> 
> I've been working on a paper entitled "Four Condorcet-Hare hybrid 
> methods for single-winner elections", which I'd like to submit to Voting 
> Matters sometime in the near future, and I'd really appreciate your 
> comments and feedback.
> 
> Here is a link to the current draft:
> http://www.econ.ucsb.edu/~armytage/hybrids.pdf

Okay, here's feedback! :-)

Regarding most strategies being burial or compromising: I seem to recall 
that in your previous paper, that was the case for most methods, but not 
for Hare (IRV) and top-two runoff. For the sake of completion, you might 
want to say that although other strategies are possible, IRV and 
Alternative Smith do better (are more resistant) than the other methods 
even when those are included. Thus the readers know it's not just a case 
of the resistant methods "reorienting their weak spots" away from burial 
and compromising.

If you want to define HRSV and HRSN more formally, one way of doing so 
would be to set a threshold on the area under the strategic curve. It 
might be incomplete since you only consider a limited number of 
candidates, but that's not a problem since HRSV/HRSN aren't strict 
criteria in any case - we don't know if the systems suddenly become much 
more vulnerable at say, C = 100 (or 1000, or ten million).

I was going to suggest including ANES simulations, but that could make 
the strategy part of the paper too cluttered. Perhaps mention that ANES 
(which are notable because they're based on real data) also show that 
the Condorcet-Hare methods resist strategy well.

-

Some other observations: it seems that adding a Smith constraint (Smith, 
or Smith//) limits the vulnerability to compromising, and that having 
the base method satisfy LNHarm greatly limits vulnerability to burial, 
since the base method is then immune to burial.

If that's right, Smith-constrained methods based on LNHarm methods could 
be interesting. They would either show great strategy resistance, or 
they would show that LNHarm is not enough (e.g. since Smith can't pass 
LNHarm, what it does let through may line up wrong). So comparing 
something like Smith/Plurality or Smith/DSC to the Condorcet-Hare 
methods could give more information either way. I may do it if I ever 
get around to implementing your strategy tests.