[Election-Methods] Simple two candidate election

[Abd ul-Rahman Lomax]

At 01:46 AM 12/23/2007, rob brown wrote:
>Well, I once again mention the 2000 election. Someone who likes 
>Nader best, Gore second best, and Bush the least would probably say 
>"Oh cool!  Before, I couldn't say that I liked Nader best because it 
>would take my vote away from Gore who really needs it.  Now with 
>this newfangled Range voting system I can express that.  I'll give 
>100 to Nader, 80 to Gore, and 0 to Bush.  Woo hoo!"
>
>And then Gore loses, just like he did with Plurality.

Actually, that is a 4/5 vote for Gore. If Nader supporters had voted 
that way, Gore would have won. What are you thinking?


>What exactly have you guys solved again?

"You guys," as if we all think alike. Fact is, the general consensus 
in the Range Voting community is to support Approval Voting. Range is 
*theoretically* ideal, though there are also good arguments to be 
made for Approval. Bottom line, though, it is a no-cost reform, 
better than Plurality, it should be a no-brainer. Just Count All the 
Votes. All it takes is deleting a couple of sentences in the election 
code, generally.

>The thing is, even more so that plurality, Range ASKS people to vote 
>"honestly".

Nope. It *allows* people to vote fractional votes. Whether or not it 
asks anything depends on the ballot instructions. I'd certainly 
oppose ballot instructions that made out a Range vote to be anything 
other than a fractional vote, an application of voting power. The 
example above would not be harmful. If a Nader supporter is willing 
to lessen the support for Gore to 80%, that's fine with me. *It is 
the voter's choice.*

>   I didn't feel dishonest at all when I voted for Gore (even though 
> I liked Nader better.  I just felt like I was giving a vote to the 
> candidate that could best use it.  (some, of course, people feel 
> otherwise, and interpret a plurality to mean "I like this candidate best")

Of course, that's an error. It's an action. It's tossing a pebble into a scale.

>Range, however, has a strong implication that you are SUPPOSED to 
>express more "rich" preferences than, say, approval.  After all, why 
>else would they go to the expense of offering the more complex ballot?

Once again, we are not generally proposing Range at this time for 
public elections. However, Range 3 would be fine. The MSNBC polls I 
saw that were Range 3 were far more informative than simple plurality 
(vote for your favorite) or approval (vote for all you would support) 
polls. With full reporting of the results, these polls (-1, 0, +1) 
showed not only net support, but that some candidates had strong 
positives and strong negatives, others showed different patterns, 
with modest support but less opposition, greater net positive. These 
were the polls that showed Ron Paul way ahead of all other 
Republicans. I think they were accurate! On the Dem side, the polls 
showed Obama and Edwards leading. Clinton had positive numbers as 
strong, but lots of negatives, for a net zilch, approximately. Again, 
that reflects my understanding of what the situation was at the time.

Approval is simple, and don't you like the slogan, which I'll repeat, 
just in case.

JUST COUNT ALL THE VOTES.

Geez, you'd think we'd already be doing that!

(There is a knee-jerk reaction that Approval is a violation of 
one-person, one-vote; but, in fact, multiple votes are alternative 
votes, for only one of them is not moot, in the end. That is, either 
only one of them counts, or none do. (If the voter votes for both 
frontrunners, the net result is an abstention from that pairwise 
contest. But how often would voters want to vote for both frontrunners?)

>I know that's an opinion, and is psychological speculation....but I 
>*am * a UI designer, and psychological speculation is what I do. :)

Voters voting sincerely in Range do no harm. Certainly, however, some 
may want to vote standard Range "personal optimization" strategy: 
vote 100% for the favorite frontrunner and 0% for the least-favored 
frontrunner, then vote sincerely for all others *given those votes as 
benchmarks*. I.e., prefer some candidate to the preferred 
frontrunner, also vote 100% for this one, detest some candidate more 
than the least preferred frontrunner, vote 0% for that one. As to 
those in the middle, it matters little where you put them; and I'd 
suggest asking yourself the question: How would you feel if your vote 
caused this candidate to win? If you'd like it, rate the candidate at 
50% to 100%. If not, rate the candidate at 0% - 50%.

But Approval is quite a bit simpler. Arguments over Range are really premature.


>By the way, I'm not about to go on a crusade to implement Range Voting
>for two-choice elections. I don't think the benefit of perhaps
>slightly better outcomes would be worth the effort and cost. I'm just
>responding to Rob's question: Yes, I think there could be _some_
>benefit, sometimes, to using RV in binary-choice elections.
>
>
>Well, if range is implemented, I'd assume it would apply to two 
>candidate elections as well, wouldn't it?

Not necessarily. However, with computer voting, it's not a problem, 
the extra complexity is at no cost. Remember, though, there aren't 
really two-candidate elections if a majority is required. A 
definition of "majority" would need to be required for Range; one of 
the simplest is to treat -- and note on the ballot -- a vote of 50% 
or more is consent to the election.

>My point, however, is that on two candidate elections, where you 
>don't have the potential for cycles and all these other problems you 
>have with multicandidate elections, it makes more clear the 
>absurdity  (in my opinion) of asking people to weaken their vote.

It is wrong to *ask* people to weaken their vote, but it is not 
wrong, at all, for people to weaken their vote, if that is what they 
choose. "Hey if you guys want A or B, either is fine with me, but I'm 
expressing my preference."

>People already have the ability to not vote.  I skip candidates and 
>propositions I don't know much about or care much about.  Here in 
>san francisco, I even get to rank candidates, and will rank 
>candidates equally if that's how I feel.  But if I actually have a 
>preference, I will express it as strongly as I can.

Are you allowed to equally rank in San Francisco? That's unusual in 
implementations claimed to be IRV.

IRV with equal ranking is a *much* better method. But I'd be 
astonished if it's allowed. *Equal ranking is Approval Voting.* -- 
unless the vote gets fractionated, which has quite properly been 
found unconstitutional.

>I think people already feel like their vote doesn't count for 
>much.  Expecting them to reduce it further is nuts, in my opinion.

*Who* is suggesting that people "should" reduce the power of their vote?


>With 3-or-more choice elections, I think the benefits of Range Voting
>would be HUGE.
>
>
>Understood.  Disagree.

Based on what? Hunch?

However, *most* of the benefit of Range comes from the ability to 
independently vote for candidates, i.e., Approval; the additional 
rating levels add more benefit, but decreasing with each additional level.

However, if we have preference analysis, there can be some good 
arguments for either having fairly high resolution -- at least Range 
10 or so, better Range 99 or 100 -- or having a Favorite marker, 
which would allow identification of a Favorite for pairwise 
comparison purposes, while leaving the strategic implications of 
full-strength votes in place.