[EM] STV+AV

[Raph Frank]

On Thu, Feb 2, 2012 at 12:41 PM, Jameson Quinn <jameson.quinn at gmail.com> wrote:
> Yes, this works. One downside is that, unlike STV, a hand count becomes
> quite untractable.

I think the random selection method for doing surplus transfers may
still work somewhat.

Approval voting is inherently harder to hand count than plurality and
this would have the same relationship to PR-STV as approval has to
pluarlity.

Fractional votes are also harder.  You could have a system where they
are allocated at random too.  The randomness would be stratified
random sampling so less variation.

If Meek's method was used, then computer tabulation would be required anyway.

I like the ballot imaging proposal for allowing complex tabulation,
with "counting" referring to the process of converting the paper
ballots to an official record of the ballots.

The big weakness is that it can weaken the secret ballot.  Voters can
"sign" their ballot by ranking minor candidates in a particular order.

Even with only 3 ranks, the fact that equal ranks are allowed means
that voters could encode some ID into their ballot, based on how they
voted on their 3rd rank.  However, there is slightly more risk, since
the 3rd choices might be relevant.