two bit ratings

[Forest Simmons]

On Fri, 5 Oct 2001, Bart Ingles wrote:

> 
> 
> Forest Simmons wrote:
> > 
> > Meanwhile, how can we make the best use of our limited equipment?
> 
> I don't suppose anyone is turning blue while waiting for my answer.  :)
> 

Approval is definitely the best use of plurality ballots and ballot
processing equipment.

And so far, nobody has made a convincing case that it would be worth it to
graduate to "two bits" (i.e. two bubbles, two ovals, or two other
characters to scribble on, next to each candidate's name). 

Two bits would allow four levels of expression. Is that important to the
general public or not?  If not, then we should stick with the best one bit
method, Approval.

If it is important, then a ballot that has "- 1" after each name, could
allow four levels of expression as follows:

The voter is allowed to change the "-" to a "+"  and/or to change the "1" 
to a "4" .

The candidate receiving the greatest number of "+" symbols (any detectable
attempt at modifying the "-" is considered a "+") wins the election.

The intensity of approval or disapproval indicated by the choice of "1" or
"4" is for psychological and informational purposes only.

The voters get all of the nice properties of Approval working for them
together with oodles of expression by sacrificing a tiny bit of
simplicity.

For official election results the equipment only has to be able to read
the first bit, i.e. detect if the "-" has been altered.

The intensity information can be compiled later from the archived ballots
by volunteers from interested parties. 

IF this two bit method has a better chance of being adopted than Approval,
why not push it?

How big is the "IF" ?

All the FairVoteOregon folks (that I have talked to) pushing IRV claim
that they just want to be able to vote their favorite strictly above their
compromise. As long as that doesn't hurt their compromise in the case when
their favorite has very weak support compared to compromise, they are
happy.

That's what IRV does for them.  And because IRV does that for them they
are willing to push it in spite of its many faults.

They are even satisfied with a version of IRV that only allows you to rank
three of the candidates (because existing equipment allows room for only
three ovals, and they don't know that three bits translate into eight
levels).

My two bit medicine would allow them to vote their favorite strictly above
their compromise without any of the horrible side effects of IRV.

"But the expressive bit is purely symbolic," you say.

That seems to be what they want.  When their favorite has no chance of
winning, they still want to vote her strictly above compromise whom they
are banking on. What point is there in doing that?  Symbolic support. 

When support for favorite begins to rival support for compromise, some of
them consider the symbolic support so important that they will continue to
rank favorite above compromise even when that would decrease the chance
of at least one of them winning.

Other IRVies turn pragmatic at this stage and resignedly vote compromise
above favorite.  Clearly the two bit method would not require this of
them. 

In fact, with the two bit method they could continue to distinguish
favorite and compromise with the symbolic bit while maximizing the chance
that at least one of them would win the election by marking them both "+"
in the instrumental bit.

In the rosie stage where your favorite has a clear advantage above
compromise, no method (neither Approval, nor IRV, nor Borda, etc.)
discourages you from voting favorite strictly above compromise.

In sum, at any stage in the evolution of relative strength of favorite and
compromise, the two bit method allows voters as good or better support for
favorite (relative to IRV).

The two bit method is uniformly better than IRV.

Approval is uniformly better than IRV except in the all important (to
IRVies) symbolic expression category.

The other slogans that IRVies are obsessed with are "majority rule" and
"one man one vote."

Martin Harper gave a good "vote shuffling genie" procedure for counting
Approval votes that might satisfy them on "one man one vote."

The special status of 50%+ majority in the case of three or more
candidates may be a tough superstition to root out from IRVies.

I have argued elsewhere that the approval winner is probably better for
the average joe than the majority winner in the rare case that the two
turn out to be different (almost never in public elections with typical
horse race coverage). 

[When different,the Approval winner is acceptable to more people than the
majority winner. Trading in someone who is acceptable (or better) to 61%
of the people for someone that 51% like better, deprives 10% of the voters
of their minimal satisfaction so that some of the others can have extra
satisfaction. That might be OK in a society where satisfaction gets spread
around, but not in a dog eat dog society like ours.]

Approval is the best we can do against the "tyranny of the majority" in
single winner elections.

This argument should strike a sympathetic chord in anybody who sees
Proportional Representation as a solution to the "tyranny of the majority"
problem.

Many IRVies support IRV because they see it as an approach to PR, unaware 
that there is a better approach through Approval.

That concludes my argument in favor of the simplest two bit method with
a possible psychological advantage over Approval.

Is there a better one?

Would trying for a head-to-head "beats all winner" before falling back on
the candidate with the greatest approval, significantly improve this in
the eyes of the public?

If so, I would be happy to go along with it.


Forest