[EM] Why I think IRV isn't a serious alternative 2

[Abd ul-Rahman Lomax]

At 12:46 PM 12/26/2008, Dave Ketchum wrote:
>We have a nominee list with much of the formality you describe.
>
>Then we have write-ins, with very little formality.

Too little, probably. I know of a case where a write-in should have 
won the election, by law, but the clerk didn't count the votes. I've 
described it before, here. The problem has to do with recognizing and 
identifying the write-in. "Write-in" doesn't necessarily mean 
"unregistered." It is legal to prohibit votes for candidates who 
haven't registered. Registration requirements are different than 
ballot position requirements.

Ballot position often requires fairly long notice, petitions, or the 
formal recognition of a candidate as the candidate of a recognized 
political party. Write-ins recognize the fact that this process 
sometimes fails us.

The two-party system, plus Plurality elections, is *like* top-two 
runoff, when the parties are roughly balanced. But sometimes they 
name candidates who are too far from the center, where both 
candidates are too extreme for most voters. When the extreme faction 
within a party, motivated by high preference strength, can overwhelm 
the centrists within the party, which doesn't take a lot, this can 
happen. If it happens with both parties at once, it's like TTR 
failing to find the compromise candidate. The *system* experiences 
center-squeeze.

Sometimes in that case, there is an independent candidacy or a third 
party steps in. These occasionally win elections. Write-ins 
occasionally win elections. I've never seen serious harm from 
write-in votes, though theoretically they can cause a spoiler effect. 
That effect is *worse* when the candidates are on the ballot.

Write-ins screw up the nice neat calculations of voting systems 
experts. What's a "sincere" vote if voters can write in their true 
favorite? Voters, by not doing that, are *already* being strategic in 
voting for their Favorite among those on the ballot. A good system 
will allow them to write in the favorite and still participate fully 
-- or *almost* fully -- in the rest of the election.

But it can be proper to require registration of write-in candidates 
-- which should be easy, it is just to identify them and to confirm 
that they accept the responsibility if elected.

Asset Voting, I expect, will lead to a veritable explosion of 
candidates, ultimately. And registration would, then, be even more 
important. Counting of write-ins could be automated if candidates 
have numbers, possibly even with an error-correcting code 
incorporated, while still allowing a hand-filled ballot. This would 
have the additional advantage that writing in identifiable 
information, other than a legitimate code, could void the ballot (as 
it is supposed to, but write-in votes currently don't void a ballot, 
even if the voter writes in the voter's name, in some places. Other 
races might be on the same ballot....)

>James frowns on such, saying that the UK properly demands more 
>formality in dealing with the needed exceptions to normal nomination.
>
>I agree that present write-ins are too informal, nominations are too 
>formal to cover all needs, and UK thoughts might help us with doing 
>something to fill the gap.

As I mentioned, San Francisco, I know, requires registration of 
write-ins. I don't know the exact requirements, I should look them up.

They used to allow write-ins on runoffs as well, the California 
default. But the last runoff election they held was in 2004. I think 
it was only for that election, write-ins were prohibited, and it went 
to the California Supreme Court, there was a write-in candidate, 
registered, who might well have won -- possibly. The Court ruled that 
the constitutional provision requiring write-ins in all elections 
didn't apply to a runoff, except by default. Runoffs, they reasoned, 
were part of the same "election," and, since the voters could have 
voted for this candidate in the first election, they had their 
opportunity. Considering parliamentary precedent, it was poor 
reasoning. Runoffs are a new election with special rules for ballot 
access, intended to make the finding of a majority likely. The 
original election failed for lack of a majority, but was used to 
select the top two to be featured on the ballot.

If the first and second election are considered one election, then 
why not consider the total vote important? (It is then like every 
eligible voter having a half-vote in each election. It becomes a bit 
like runoff Bucklin, then.)

The voters, as a result of the first ballot, may recognize the value 
of a write-in that they did not see before. To prohibit write-ins, 
then, reduces the flexibility of voters in dealing with unusual but 
important situations. It institutionalizes center-squeeze, thus 
making it impossible to fix the most common, and known, failure of 
top two runoff. I rather doubt that much of this was considered by 
the California court, usually the level of expertise shown in court 
records misses the most cogent arguments when it comes to voting systems.