# [ER] Cumulative Approval Voting

Mike Ositoff ntk at netcom.com
Thu Oct 1 19:04:20 PDT 1998

```>
> Dear Bart Ingles,
>
> You wrote: >  Approval is logically a group of separate
> >YES/NO races, one for each candidate.  The total number of votes for
> >each race is equal to the number of voters in each race -- including
> >both yes and no votes (i.e. 100).

Approval can be regarded as a point system in which we can give
any candidate 1 or 0 points. Plurality is a peculiar point system
with the inexplicable rule that we can only give a point to 1
candidate.

>
> Donald: And, the winner is??   What determines the winner?
>      Is it the one with the most Yes votes? If so, I must point out to you
> that all the other candidates together had more Yes votes than the winner.

Yes, Don, sometimes there isn't a majority winner. With rank-balloting,
we can devise the best count rules to do the best we can in that
regard. In a multicandidate race we can at least recognize the
majority that says A is better than B, and honor their wish. GMC.
Votes-Against is the method that does that. But we can also accomplish
much with a simpler method. I'm not saying that Approval is as good
as VA, but I'm saying that Approval is so good that insisting on
something better is perfectionism. Perfectionism is fine if we
can afford it, but it can be a mistake when we can't afford it.

> If you are to have six separate Yes/No races, you cannot merely pick one
> race of six and say that this race will decide the election.
>      Consider this point: The reason Approval Voting is being considered in
> an election is because we did not have a majority of the total votes on the
> first count. We then go to some method seeking a majority winner. Now, you
> are saying that we do not need a majority of the total of the votes.
>      Consider this next point: If we did have a majority on the first count
> of the ballots, will Approval Voting always confirm that winner? The answer
> is NO! (Run-Off and Condorcet will always confirm the majority winner of
> the first count.)
>

As I said, to give away a majority candidate's win in Approval,
his voters must believe that he has fewer votes than the disliked
candidates that they're trying to defeat by voting also for the
compromise. That's a big mis-estimate. As I said, this differs from
FPP in that it takes twice as many mistaken voters to give it away,
and if it happens it isn't an equilibriulm, a stable situation
that can go on & on.

No matter how you define Approval Voting, we are forced to face the
> majority of the total something - like votes or races or pairwise contests.
>      What you have written above for Approval Voting, could also apply to
> First Past the Post when it has no majority.

I've clarified important differences between Approval & FPP

>
> You wrote:
> >This is similar to the way separate but competing ballot initiatives are
> >handled in California:  If two or more conflicting measures pass, the
> >one with the greatest plurality becomes law, and the others are defeated
> >(wholly or in part, depending on the way they were worded).  The only
> >difference is that each ballot measure must first get a majority in its
> >own race.
>
> Don: No, it is not the same. The voters did not vote Yes/No on each
> candidate. Each voter only had two votes, and Approval Voting regards each
> vote as having equal weight. This is hardly six separate Yes/No races.
>      Besides, this is not the best way to handle two or more competing
> ballot initiatives. The voter should be allowed to rank competing
> initiatives and then have some method, like run-off, to decide the winner.

Ideally it would certainly be best to rank them, if the rankings
were counted in a good way, and advocates of Margins & IRO aren't
going to ever agree on that issue.

>
> You wrote:
> >Actually, if you want to add the votes of all the races together, the
> >total is 600 in this example.  It doesn't really matter whether you use
> >100 or 600 anyway, since the method neither demands nor guarantees a
> >majority of any total.  I suppose this makes it no more and no less
> >valid than FPP in that respect.
>
> Don; Actually it does matter if you use 600 for 6 candidates and 100
> voters. This would mean that each voter made six choices and that all
> choices were being used. The result would give the six candidates exactly
> 100 Yes votes each - a six way tie. How does Approval Voting solve ties?
>      I will agree that Approval Voting is less valid than FPP. In FPP the
> one vote is the most preferred choice of the voter. Approval Voting does
> not allow the voter to have a most preferred choice (if the election

Nonsense. Approval allows you to only vote for 	1 if you so choose.
It gives you more freedom than FPP, and you can use it as FPP if
you choose. FPP's problem is that it doesn't give that freedom.
IRO is FPP with a moving vote, but still nothing but glorified FPP.

> reaches the point of using the method).
>      I would like to suggest that Approval Voting can be improved by giving
> each voter the same number of votes as there are candidates and let the
> voter use the votes any way the voter wishes to use them - like in
> Cumulative Voting.

That's been discussed. It's equivalent to FPP. Your best strategy
is to give all your votes to the candidate with greatest strategic
value.

>
>      We could call the method "Cumulative Approval Voting".
>
>      This dosen't mean I support the method of Cumulative Approval. While
> the change should give us some winners with majorities, it will not give us
> a majority every time.
>
> Regards,
> Donald
>
>
>      \\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
>      ///                 N E W    D E M O C R A C Y                ///
>      \\\ Home of Citizen's Democracy   http://www.mich.com/~donald \\\
>      /////////////////////////////////////////////////////////////////
>
>
>
>

```