[EM] The Coming California Single Seat Election

[Alex Small]

Donald Davison said:
> I think both Approval and Condorect would fail this election.
>
> Approval would have a leading candidate, but without a majority of the
> voters.

Not necessarily.  The field will probably narrow to a half-dozen or so
contenders in the next few weeks.  Most people would vote for one or a few
of those contenders (we can debate later the optimal number to approve)
plus their favorite porn star, child actor, widow of Frank Zappa,
thong-selling software engineer, smokers' rights advocate, etc.  The
fringe candidates would become irrelevant, and it would be a race between

1)  a shamelss social climber
2)  an actor with no qualifications
3)  a smut peddler who cares
4)  an unremarkable lieutenant governor
5)  a conservative state senator who's too smart to win

Put me down for 1, 3, and 5.  Plus the smokers' rights advocate and
thong-selling software engineer.

> Condorect will need about ten thousand pairings, most of which will have
> too few lower choices in order to yield valid results.

I don't know what you mean by "most of which will have too few lower
choices to yield valid results."

The ballot would NOT require voters to separately vote on ten thousand
pairwise contests.  The ballot would look just like an IRV ballot, where
voters would rank either all of the 135 candidates or as many as they want
to rank, truncating the rest (I'm still not sure whether I'd rank Frank
Zappa's widow above or below the baseball commissioner).

However, unlike an IRV ballot, the data storage would be much easier. 
Normally I don't think IRV's data storage requirements are anything to
harp on (I used to think that way, but I've been dissuaded).  But in this
case the data storage requirement would be of the order 3*10^42 tallies. 
Most of those would be zero, so it would actually be more efficient to
store each ballot separately.  By contrast, with Condorcet the voting
machine would use each ballot to add to 9045 separate tallies, much less
than storing 8 million separate ballots in memory, and then separately
addressing each one to do the statewide count.

So the simplicity advantage definitely goes to Condorcet here.  Now,
simplicity is not the only advantage to consider, but you're the one who
harped on ten thousand pairwise contests, so to the extent that simplicity
does matter to you Condorcet or Approval should be superior to IRV.  Now,
if you want to argue that other normative factors outweigh simplicity
that's fine, we can have a lively debate.  But please don't try to suggest
thta IRV would be simpler than Approval or Condorcet in a race with 135
candidates.

> There's very good chance that the so called `Condorect Winner' will be
> the Pluralty Winner.

It may turn out that the Terminator will be preferred to all others by a
majority of Californians, and that he will get the most first-place votes.
 He's currently consuming publicity the way a black hole consumes dust and
radiation in its vicinity, so he might beat up on the other candidates
like a robot beating a liquid metal monster.  In that scenario he'd also
probably be the IRV winner.



Alex

P.S.  The black hole analogy is particularly apt because black holes
destroy information, at least according to my friends who study black hole
thermodynamics.