footer
The software utility, eVote and the eVote clerk, injects true democracy and deliberation into our real-world landscape.

Thursday, February 07, 2008

Range Voting

Some in possession of big mathematics insights have insist that range voting is fundamentally better than ranked voting. If you're interested, see http://rangevoting.org.

Range Voting is the best of any known voting system, where "best" means that it produces the greatest overall happiness/satisfaction. This holds true even if people try to game the system by doing things like ranking their top choice as 100% and all other choices 0%. (In fact, it's rational to do this if you know that your top choice is polling just barely behind your #2 choice.)

4 Comments:

  • At 10:22 AM, Blogger BROKEN LADDER said…

    "Ranked" voting is not a voting method. Condorcet, Borda, Bucklin, and IRV are but four ranked voting methods. There are scores of others, and an infinite number of hypothetical ones that you could create.

    Range Voting is superior to essentially all known ranked methods.

     
  • At 12:24 AM, Anonymous Greg said…

    Under Range Voting, voting for anyone beyond one's first choice jeopardizes the chances of the first choice. In practice, such a system will lead to bullet voting, and we'll be back we're we started with plurality.

     
  • At 2:17 AM, Blogger BROKEN LADDER said…

    Greg said:

    Under Range Voting, voting for anyone beyond one's first choice jeopardizes the chances of the first choice. In practice, such a system will lead to bullet voting, and we'll be back we're we started with plurality.

    That's just a false and deceptive argument that IRV proponents use all the time. Imagine you're a Nader supporter back in 2000, and you've decided to vote for Gore so that you don't "waste" your vote. If you were suddenly given the ability to score the candidates, why on Earth wouldn't you give the highest score possible to Nader, if he's really your favorite? And now say you are one of the Nader fans who decided that helping Gore to defeat Bush was important, but not as important as showing your support for Nader. With Range Voting, you could give Gore points to help prevent the greater evil of the major candidates (from your perspective), without sacrificing your desire to show maximum support for Nader. Thus there is no credible doubt that Range Voting would have fixed the spoiler effect in that election, while also giving Nader a dramatically more accurate level of support.

    Or imagine you're voting in a primary situation, where Obama, Clinton, and Edwards are pretty much tied according to the polls, and your honest scores would be Obama=10, Edwards=6, Clinton=0. Your argument says you would want to bullet vote, giving Obama a "10", and everyone else a "0", because if you help Edwards, you lose 4 points of satisfaction compared to Obama. But, imagine Clinton wins because you didn't help Edwards -- then you lose 6 points of satisfaction compared to what you could have had. That's a bigger loss! And the decision gets even simpler (overwhelmingly obvious) if your Edwards score is more like an 8. Plus, what limited real world data we have about Approval Voting elections shows bullet-voting was not a problem, aside from the obvious mathematical perspective.

    This has been explained to Greg at length, so I cannot see any good excuse for his deceptive argument.

     
  • At 2:31 AM, Blogger BROKEN LADDER said…

    A more precise look at Greg's views, in light of the fact that he's an IRV supporter:

    Greg has been engaging in an extremely common type of bias regarding
    strategic behavior in voting. The flawed argument basically goes:

    1. With Range/Approval Voting, voters will want to bullet vote (even
    though it's strategically stupid), out of fear that a vote for a
    less-preferred choice will hurt their favorite.

    2. With IRV, voters will want to be sincere (even though it's a simple
    and generally advisable strategy to use favorite-betrayal).

    == Examples of #1 ==
    Sincere scores: Nader=10, Gore=7, Bush=0
    Rational strategic vote: Nader=10, Gore=10, Bush=0
    [Note it would be incredibly stupid for this voter to bullet vote for
    Nader out of fear that also supporting Gore would hurt Nader, since
    Nader almost certainly won't win anyway.]

    another..
    Sincere scores: Obama=10, Edwards=7, Clinton=0
    Rational strategic vote (assuming all 3 are tied in the polls):
    Obama=10, Edwards=10, Clinton=0

    [Greg is suggesting this voter wants to bullet vote for Obama, since a
    vote for Edwards risks a loss of 3 points of utility. But..HELLO!
    Giving Edwards a 0 risks a loss of SEVEN points of utility if Clinton
    wins, while the voter could have had his second choice by not bullet
    voting.]

    == Example of #2 ==
    Voter who prefers Nader>Gore>Bush

    Voter considers

    P: Probability that situation will arise in which Bush will win, while
    an insincere switch to Gore>Nader>Bush will cause Gore to beat Bush.

    Q: Probability that situation will arise in which Nader will win,
    while an insincere switch to Gore>Nader>Bush will cause Gore to win
    instead.

    Voter realizes that

    P is statistically much larger than Q.

    Voter (if he is rational) raises his favorite _viable_ candidate as a
    general rule.

    == Moral ==
    It is biased and/or naive to think that Range Voting will inspire
    widespread use of an obviously bad (anti)strategy, while IRV will
    inspire naive angelic honesty in spite of a reasonably obvious
    generally applicable strategy.

    *THEREFORE*
    It is disingenuous of RV doubters and/or IRV supporters to make this
    claim when it is so clearly wrong.

     

Post a Comment

<< Home