[EM] Reading rangevoting.org/VotMach.html

[Abd ul-Rahman Lomax]

At 03:23 AM 7/30/2006, Dave Ketchum wrote:
>Agreed that for a special election for filling one office, ten levers
>could be used for each candidate - but who cares.  General elections are
>what counts.

What is commonly overlooked in these discussions is that Range is a 
method of vote counting. The resolution is actually a detail. As has 
often been stated, Approval is a kind of Range Voting method: it is 
counted in a way that is exactly equivalent; blank equals Range zero 
and a mark equals Range one. The votes for each candidate are 
totalled and the candidate with the highest total count wins.

Extending this by adding one more possible vote could actually give 
90% of the Range buck at only a small cost in ballot space. 
Basically, the vote for each candidate becomes trinary instead of 
binary. The way that I would configure it would be that blank would 
be 0.5. Each candidate would be listed on the ballot as if he or she 
were a question, with Yes and No options. No would be zero and Yes 
would be one. Pursuing discussions that have taken place on the Range 
list, to avoid the election of an unknown candidate, a special rule 
would require the winner to have a certain minimum level of Yes votes.

This is still Range. Range 3, as we usually measure, Approval being Range 2.

>Our many columns and rows seems like a lot, til you try to arrange a
>ballot neatly.  Cannot be done if there are enough candidates and enough
>offices.  So you squeeze - has happened that a loser went to court
>complaining that he lost because of the poor quality of the squeezing.

The same problem exists, of course, in any election with too many 
candidates. I agree that it is a problem, but it is a technical 
problem rather easily soluble. The core of it is the inflexibility of 
voting machines, which, in my view, were a bad idea before their 
time. Voting machines make sense when a vote total must be available 
immediately, as in an assembly where every member has an electronic 
device for entering a vote. But for counting public elections, they 
open the door to undetectable fraud, and the cost of counting a 
public election by hand is only a small fraction of the real cost of 
the election, which mostly consists of the labor invested by voters.

>So, the RV promoters would have us install extra voting machines.

Well, I'm not exactly an RV promoter, but I do agree that RV deserves 
careful consideration. As I have written elsewhere, RV is actually an 
optimal decision-making system, compatible with game theory, 
translated into a voting method.

I have a sense that the strategic voting considerations often 
proposed as arguments against Range are a red herring. It has been 
said that hard-core partisans will bullet-vote, and it seems to be an 
assumption that this is harmful. However, if we normalize the vote, 
as I did in the examples above, where the most-positive vote is one 
and the most-negative vote is zero, the most that any voter can give 
to any candidate is one vote. So a bullet voter gives one full vote 
to the favorite and nothing to any other candidate. Yet every other 
voter (who fully votes, i.e., gives maximum rating to at least one 
candidate) contributes the same full vote to his or her favorite(s). 
So all the bullet-voter has done is to abstain from all contests not 
involving his or her favorite. This is a loss of power for that 
voter; my sense is that the loss of power is more significant than 
the alleged gain by refusing to contribute to the totals for 
non-favorite candidates.

In the end, I think the math would show that the expected outcome for 
each voter is maximized if the voter votes sincerely, that is, if the 
voter assigns expected election values to each candidate. If the 
voter really thinks that all candidates but the favorite are equally 
bad, then the voter is certainly free to so vote.

>   No
>sale, for then there would be trouble making sure all the voters got to
>all the machines.

Again, this is a voting machine problem, a very strong argument for 
going back to paper ballots, particularly paper ballots that can be 
scanned. Since the equipment necessary for scanning is lying about, 
essentially free, the conversion cost would be minimal. Note that 
voters are already able to cast paper ballots, since I presume they 
don't mail out the machines to absentee voters.....

As I've written many times, voting machines were a bad idea from the 
beginning. If a voting machine could not handle, say, two or three 
positions per candidate, then it was already primed to have 
difficulty with just about any electoral reform, since most of the 
proposed reforms give more opportunity for additional candidates to 
run without damaging outcomes, so we can expect candidate counts to 
increase. The voting machine argument, in the end, is against just 
about all election method reform, not just Range. It inhibits IRV or 
any Condorcet method that allows more than two ranks. (Condorcet 
reduces to Approval if only two ranks are allowed.)

>HAVA is demanding new voting machines.  With proper planning and
>procurement these could have whatever capabilities are useful.

The proposals -- and laws -- for new machines just about drive me 
crazy. Paper ballots are cheap, can be scanned with cheap or free 
equipment, and provide inherent audit trail. However, they don't make 
the mfrs of voting machines rich, and there is no paper-ballot lobby, 
nor a significant pencil lobby, nor will there be until voters wake 
up and realize that the system is eating their lunch.