[EM] MultiGroup voting method

[Abd ul-Rahman Lomax]

At 09:52 AM 4/6/2007, Juho wrote:
>Here's one method for PR multi-winner elections. [...]
>
>Candidates (or parties) are free to form any kind groups. Some
>typical groups are parties and regions. The groups are allowed to be
>hierarchical and to overlap. Also mandatory groups can be covered
>within the method (all candidates might e.g. be mandated to represent
>one of the regions).

Is Juho aware of the use of Asset Voting that I have proposed? It 
seems far simpler and more flexible than what Juho has proposed, with 
similar effect. It also has overlapping "groups." The groups are 
created by the candidates when they assign their Assets. They can 
choose to reassign votes in precinct blocks, approximately, so, if 
they have enough votes, they can create regions that are connected in 
some way, with a single representative. Where they don't have enough 
votes, they can reassign the votes to someone else, and, if enough of 
those amalgamated votes come together, they can still form regions 
smaller than the whole state, but more commonly would create 
state-wide constituencies.

A popular representative, perhaps the leader of a party in the state, 
could create relatively small districts. A political party that was 
small and spread thin, but with enough loyal voters, state-wide, to 
gain a quota of votes, would gain representation state-wide. A party 
with even less support than that, could cooperate with other similar 
groups to create a seat that represents more than one party, 
presumably with similar agendas or interests.

It takes no complex, top-down, imposed system. All it takes is 
allowing votes to amalgamate intelligently, i.e., under the direction 
of a trusted candidate. And my position has been that if we can trust 
a candidate to exercise our vote in the legislature, which is what 
representatives do, we can trust the candidate with what is actually 
a lesser task, that of choosing who is to represent the votes involved.

(I might trust someone absolutely, but also know that this person, 
for whatever reason, is incapable of serving. Maybe they are 
terrified of public speaking, maybe they have other responsibilities 
that would prevent it, etc.)

With Asset Voting, even if I don't trust an individual, I can assign 
fractions of a vote to a virtual committee.... Asset, in its 
Fractional Approval Asset Voting incarnation, takes a very simple 
ballot, it is just like a plurality single-winner ballot, but I can 
vote for as many as I like. My vote is distributed equally among all 
those I vote for. These votes are, as in all Asset methods, available 
to be assembled to create winners at the discretion of the candidates 
holding them.

If it weren't so common, I'd be nonplussed by those who object that 
this gives the candidates too much power. Those candidates, if 
elected, are going to have *greater* power when they are serving. 
They will amalgamate votes, at their discretion, to create winning 
resolutions. What's the difference?

The fact is that Asset doesn't need to consider party affiliation at 
all. Parties can continue to function, and I assume that some 
candidates will only distribute Assets within their own party. But 
that is actually up to them, and they are responsible before the 
voters for what they do.

Asset Voting wastes no votes. If we allow the transfer of fractional 
votes, seats can be an *exact* quota. Normally, slop might result in 
a missing seat. However, I see no harm in having a legislature with, 
say, 30 seats, but it might be 31 if the candidates manage to 
assemble all the votes. (It might be less if they really can't agree, 
and this problem is widespread.) As a voter, if my chosen candidate 
failed to distribute a significant number of votes, I'd really not 
like it and I would be likely to not vote for that candidate again.

It is also possible to use a ranked ballot to reassign seats, with 
exhausted ballots (none of the candidates ranked on the ballot are 
chosen immediately) being assigned, then, to the candidate listed 
first, or to more than one if more than one is listed first. Then it 
functions as Asset. We could call it STV-Asset. I'm just not sure 
that the gain in direct voter power is worth the complexity of the 
ballot. To me, it would be a fish bicycle. Why should I second-guess 
the candidate I trust the most? He or she probably has much better 
knowledge than I of the possible persons to serve....

[...]
>Finding the optimal outcome as defined above is computationally
>complex.

Of course it is. What Juho is trying to do is to automate a process, 
using fixed rules, instead of allowing human intelligence, which is 
far more flexible and which can deal with unanticipated 
contingencies, etc., to allocate votes and seats.

The Asset scheme that I've proposed allows the candidates to create 
districts on the fly. They don't have to, but there is very little 
reason for them not to. All they have to do is to transfer votes in 
precinct blocks. It's possible to define a procedure whereby seats 
are entirely allocated from intact precincts, but at the cost, 
obviously, of precision in representation. Because this loss would be 
small, it could be acceptable, but it complicates things 
unnecessarily, in my view.

Instead, if a candidate is assigning, say, fifty precincts to a 
candidate, and there are some votes left over, a fraction of a 
precinct, that candidate would simply assign the extra wherever it 
might be useful. If I'm from one of those 50 precincts, even though 
perhaps 1% of my vote went elsewhere, I'd still consider the 
representative to be mine. The difference is minute. Precincts aren't 
anonymous, but they are still relatively small units, so its easy to 
get close enough for folk music, as they say.

It wasn't clear to me how Juho's scheme really differs from Asset 
Voting, it seems to function similarly, but he proposes, if I read 
correctly, this or that restriction intending to have this or that 
supposedly desirable effect. I think that trying to get party or 
"group" representation by manipulating the rules is, quite simply, 
unnecessary. As long as voters can freely choose candidates, and 
especially the candidate they trust most, and access to the ballot is 
easy, with perhaps minor candidates being listed on a supplemental 
booklet so that voters can enter a number rather than write in a 
name, it will be the voters who determine what kind of groups are 
represented. Candidates can make various pledges, if they want, and 
voters can pay attention to these pledges, or not. Candidates could 
pledge to reassign votes according to a candidate list (and this is a 
special version of the method we have called Candidate List), but I'd 
make that optional. Quite simply, I don't want to say "I trust you," 
and then, "Well, no, I don't trust you, you must do A and B and, if 
necessary, C." What's the point of having a representative if you 
have to instruct him or her? We don't allow instruction of 
representatives in Congress, why should be allow it in the process of 
choosing representatives. Rather, a good system must properly require 
voters to use their intelligence *in choosing whom to trust.* It's a 
much simpler decision, actually, rather than having to consider 
electability -- which is not relevant to Asset --, party affiliation 
--, it's up to the voter to determine if this is important or not, -- etc.

I didn't invent Asset. Warren Smith did. But it is such an obvious 
scheme that I think we should consider it clearly and openly and only 
introduce complexity if, in the light of this, it seems necessary. 
Rather than considering what we'd want representatives to do, then 
trying to create Rube Goldberg schemes to force them to do it. I like 
to be free and I want my representatives to be free as well. If I 
don't trust them, I should vote for someone else.

(And, one will note, under Asset it is possible for any citizen to 
become an "elector," by receiving votes, which might even be as 
little as one vote, and, because electors are not anonymous, it is 
also possible to allow electors to vote in the assembly, though not 
to routinely participate in deliberation. Thus we could possibly have 
solved the problem of scale in democracy, without losing the security 
and practicality of representative democracy. For such a simple 
scheme, that's quite a possibility!)