[EM] Voting Theory and Populism

[Raph Frank]

On Sat, Oct 18, 2008 at 1:09 AM, Greg Nisbet <gregory.nisbet at gmail.com> wrote:
> 1: Approval (slightly ahead of condorcet)
> 1: Condorcet
> 2: IRNR
> 3. IRV
>
> = Where would Range fit in, just out of curiosity? Of the things that are
> listed, I completely agree with this.

I was just listing the ones he gave.  I would place it ahead of approval.

However, I am not sure that the benefits over approval are worth the
added complexity of score/range at least initially.

I think condorcet, range and approval would in most cases be
functionally equivalent.  They should all mostly elect the condorcet
winner in practice.

If you can get any of them, then go for it.  I think any of the 3
would be a major improvement.

Another one that is good is approval + top 2 run off.  In fact,
Warren's sims put it ahead of Range/Scorevoting.

It replaces the plurality first round of top-2 runoff with an approval
first round.

Basically, the 2 most approved candidates make it to the 2nd round.
The voters can then vote for one or the other.

So my list is something like

1++: Approval + top-2 runoff
1+: Range
1-: Approval
1--: Condorcet

2: IRNR (pending more info)
3: IRV

> CPO-STV (or maybe Schulze-STV) are obvious improvements, but with big
> costs in complexity.  I do think that vote management is a weakness of
> PR-STV (I wonder if Schulze STV would stop parties bothering to try).
> Also, the district sizes need to be reasonable (say 5+).  In Ireland,
> there are 3.86 seats per constituency on average, which I think is to
> low.
>
> =Why have constituencies at all?

Complexity both for voters and counters.

It isn't possible for voters to rank 1000 candidates (or at least they
won't bother).  If you had 200 seats in the national legislature, then
that is likely the number of candidates.

Also, counting a PR-STV (or one of the more complex versions) would be
very hard to do at a national level while maintaining security.  It
would almost certainly require e-voting,

Abd's ballot imaging proposal would help here.  This allows paper
ballots and computer counting.  Basically. after the ballot boxes are
opened, all ballot are photographed using digital cameras (each party
would make its own copy).  Everyone can then use image recognition
software to work out the results and make sure the official count is
correct.

There was a thread on having national PR-STV (and other PR) elections
last month or maybe August.

The logistics of actually carrying out the count were largely ignored,
but we did look at the issue from the voter's perspective.

Anyway, my suggestion would be to have candidate-list based
completion.  For example, each voter gets a local ballot.  It lists
just the local candidates.

You then rank them in order of your choice and then pick a candidate's
list as your completion list.  Your ranking is used and then your
candidate's published rankings are used to complete your ballot.

This means that every voter gets a full ranking.  However, the first
few rankings (where most of the power is in PR-STV) are entered
directly by the voter and there would be write-in spaces in order to
vote for non-local candidates.

> = Now you're getting it, you have to compromise with non-voting-theorists
> here. I'd say go for the voting method. That will break 2 party domination
> and if there isn't a majority party anymore, gerrymandering will crumble/be
> exposed.

Well, it depends, a district with 60% republicans are likely to still
vote republican ... at least for the first few elections.  Hopefully,
the public would vote out politicians who support gerrymandering.

The problem is that districting reform is a hard issue to solve.  At
least, automatic redistricting is unbiased, but could be somewhat
random.

PR is just a better solution as it allows the voters to somewhat
control their own districts.

> = We see it as logical, that's true. Would a random person see it that way?

One option is to point out the similarities to ballot questions that
modify state constitutions.  Some states allow 2 to be run that
contradict each other, if both pass, the one with the most Yeses wins.

>
> Maybe rename approval to "direct voter appointment".  The voters
> directly appoint the winner.
>
> =that certainly seems like it would address some of the problems. The way
> you are presenting it looks pretty logical.
>
> E.g. the ballot would be
>
> Do you want to appoint A. Adams as Governor? Yes ( _ ) No ( _ )
>
> Do you want to appoint B. Brown as Governor? Yes ( _ ) No ( _ )
>
> Do you want to appoint C. Collins as Governor? Yes ( _ ) No ( _ )
>
> Do you want to appoint D. Davis as Governor? Yes ( _ ) No ( _ )
>
> The one with the most Yes votes wins.
>
> =you just might be able to convince people of this~

It is the key reason that approval isn't unfair.  By not approving
someone, you are effectively voting "disapproved".

Also, it doesn't discriminate against any candidate or voter.

> You could also add a rule that if any candidate gets a majority, then
> all candidates without a majority are excluded.  The unexcluded
> candidate with the most Yes votes wins.
>
> =interesting idea, I'll look into it.

It might cause weird effects though with strategic blanks ... but
probably not.  In fact, in any real election, I would hope that the
most approved candidate always get more than 50% approval.

> =Schulze is frequently decisive and arguably pretty good. I see no reason to
> demote it.

Well, you could say that there are a few condorcet methods and you
think Schulze is the best.

> You could note that there are sometimes tie breaking rules for
> circular ties, but that is pointless unless they accept that a
> condorcet winner should win in the first place.
>
> =Is the CW criterion good is part of the massive CW vs Range topic.

Yeah, I know.  But deciding which of the CW compliant methods is best
is not important unless you have convince the person that condorcet
compliance is a good thing.

There are 2 things to 'get' about condorcet methods.  The tie breaking
rules tend to be reasonably complex, but even if they weren't the
reason why one is better than the others is even more complex.