[EM] IRV vs Plurality

[Abd ul-Rahman Lomax]

At 07:10 PM 1/11/2010, Dave Ketchum wrote:
>The possible excitement tangles with the secrecy laws - reporting in a
>manner that identifies how ANY ONE voter voted needs preventing
>(needed protection of voters).

There is a problem with ranked ballots and true write-in votes: a 
voter may identify the ballot specifically and clearly. The voter 
writes in their own name. In last rank, that has practically no 
effect. With plurality, the write-in shows, but it then makes the 
ballot moot, generally. (If it's unique, it's moot!) Vote-coercers 
don't want the person to vote for themselves! And a mark in any other 
place than the write-in place invalidates the ballot, and any writing 
in the write-in -- any at all -- generally invalidates a vote for any 
other candidate.

Because Asset Voting shares this problem in a different way, my 
solution was to consider that true write-ins are not allowed. Rather, 
there is a candidate pamphlet that is printed with the name of all 
registered candidates. The registration fee is nominal, just enough 
to pay for a listing in the pamphlet with a unique name and a code to 
be used to vote for the candidate. There are no names on the ballot. 
The voter has to find the name (unless the voter is assisted, which 
is allowed often for people with physical difficulties). The ballot 
only has a number on it, marked in a way that would be difficult to 
specifically identify. With computer ballots, it's even easier. And 
there could be a look-up for the number so that the actual name of 
the candidate was displayed.

But it's still possible to identify ballots with very high 
probability. Perfect secrecy in voting is not possible to guarantee, 
for, usually, the risk comes from those with special access to 
ballots. I.e., incumbents. They can set up means to identify voters, 
easily, with that control. I'll note that in some democracies, each 
ballot is serialized on the back and this serial number is recorded 
with the original voting record. You can't tell who voted just by 
looking at the ballot, you'd have to also have list to the registry 
of votes made at the voting place. And when ballots are counted, this 
side is normally kept down.

So the protection that is practical and perhaps necessary is that it 
is kept *difficult* to identify a vote from a voter who does not want 
to be identified.

With fully-ranked ballots with enough candidates, it is also possible 
for a voter to vote, say, for a candidate in first preference and 
then use the remaining preferences to create a unique code, even if 
write-ins are not allowed. In San Francisco, one supervisor race had, 
what, 23 candidates on the ballot? Overvoting in lower ranks is moot 
in first rank and doesn't affect the first preference vote at all 
with IRV. So that's a code with 22 possibilities, even neglecting 
write-ins, for each of the ranking slots, and blanks can be used as 
well as part of the code. So there would be 23^22 possible codes. It 
would definitely be suspicious if many voters used more than a few 
double-votes. Still, what could they do about it? As it is, write-ins 
*must* be allowed in every election by the California constitution, 
and the exception for some runoff elections was a result of an 
argument that the runoff wasn't a separate election, it was merely 
part of the first one.

But that argument was totally stupid! Remember, the argument was that 
the winner *must* get a majority. A majority of votes! A majority of 
*what* votes? If the whole process is one election, it would have to 
be a majority of all the votes cast in the primary and runoff. But 
obviously, that's not it. Rather, it's a majority of votes in the 
runoff. It's a separate election, the only one that counts, if the 
first one didn't find a majority. All the votes cast in the first 
election are moot except for determining who is on the ballot. And 
thus the allowance for write-ins should have remained, and, if the 
method remained plurality, a plurality winner would have been allowed 
in the runoff. Simple. Don't run in a plurality runoff election 
unless as a write-in unless you are willing to risk a spoiler effect. 
You already know, probably, how you stand with the voters. And if I 
were a voter seeing a write-in candidacy, I'd want to know that not 
only do I prefer this write-in candidate, by a substantial margin, 
but also a plurality of voters prefer this candidate. I'd want to 
know, clearly, that there was a center squeeze effect in the primary.

Runoff voting is quite cool for minor parties, it allows them to 
express that first preference with reasonable sincerity, usually.

Give voters a better method in the primary, and there will be fewer 
runoffs and better results with sincere voting.