Slashdot Mirror
An Analysis of Various Election Methods
An anonymous reader writes "David Cobb talked about Instant Runoff Voting (IRV) as the best choice in electoral methods in his interview here, but is it really? The folks over at electionmethods.org seem to think it isn't. They favor Condorcet voting, which is another ranking style method using simulated one on one elections. Here is an evaluation of various methods, including IRV and Condorcet."
Voting systems are one of those things people will ALWAYS disagree on, because the set of "reasonable" desirable properties that most people would like in a system are contradictory, as shown by Kenneth Arrow.
If you must spend all votes, then the system becomes non-repudiable, which as I mentioned in another post, is a very serious problem with approval voting.
In straight approval voting, what stops the guys that take your ballot from marking their candidate of choice on your ballot?
The reason that it can be true that 1+1 > 2 is that very peculiar nonzero value of the + operator
The person who is would win a one-on-one vote against for every other candidate wins, if such a person exists.
"Never attribute to malice that which can be adequately explained by stupidity." -- Hanlon's Razor
The attractive thing about runoff elections is they make it more viable to have more than two parties. Unfortanetly, the two major parties have stacked things to make it difficult for a viable third party to establish itself.
The Bolachek Journals
What you're proposing is a modified version of Range Voting, which has its proponents.
"Never attribute to malice that which can be adequately explained by stupidity." -- Hanlon's Razor
He has made a simulation that is open source.
So hack away. Look here and here.
This is somewhat like the Borda voting method, except that in the Borda method, you must give N points to your favorite candidate, N - 1 points to your second favorite, and so on - the number of points is fixed.
The problem with your method is that everybody is going to throw their points at one candidate - their favorite. The problem with the Borda method is this scenario: Suppose you have high school band members voting on where they want the band trip to be. The options are Chicago, Toronto, and Myrtle Beach. The situation is this: 45 bandies want Toronto over Myrtle Beach, 45 prefer Myrtle Beach over Toronto, and 10 loonies prefer Chicago (which is such a bad idea, by the way). Each person lists their three choices in order - first place votes are worth 3 points, second place 2 points, third place 1 point.
All the Toronto-wanters decide that to screw the Myrtle Beach crowd, they'll vote for Myrtle Beach in third place, with Chicago in second, even though it is a crappy place for a band trip (because they shouldn't have to worry about Chicago getting picked). The Myrtle Beach-wanters do the same thing. The result is that 180 points go to each Myrtle Beach, Chicago, and Toronto.
Then the Chicago loonies vote for Chicago in first place, putting Chicago over the edge. Chicago wins, and 90% of people hate the band trip.
Most election methods operate under the assumption that the popular vote is what matters. In America, that simply isn't the case. What matters is which candidate will most accurately reflect the needs and desires of the nation, not only of its population centers.
Additionally, a charismatic candidate can sweep the popular vote by carrying a handful of major cities. Popular voting in America implies that only the inner city vote matters, which disenfranchises the rural voters - you know, those who produce oil, wheat, beef, milk, chickens, pork, corn, soybeans, potatoes, and other things that you like to have in your life.
Quite simply, the Electoral College is a very effective compromise that has kept our Presidential elections mostly sane for more than 50 iterations. It ain't broke - don't try to fix it.
Approval Voting is very easy to explain: vote for ("approve") as many candidates as you wish (no ranking), and the candidate with the most votes wins. There it is in one sentence.
Approval voting requires no new voting equipment. It could be implemented very quickly once a consensus is reached, and it would truly revolutionize our political system, giving minor parties a much fairer chance than they now have.
One caveat: it will not work well in US Presidential elections as long as the Electoral College in place. Then again, neither will any of the other alternative election methods. Oh well.
I watch Brit Hume on Fox News
The number of ballots should equal the number of people who cast ballots, though, should it not? And would this not provide the number you're looking for?
You can never go home again... but I guess you can shop there.
As the webmaster of ElectionMethods.org, I am thrilled to see this link on slashdot. Please tell your friends and relatives too!
I would just like to clarify a couple of points. We believe that Condorcet voting is the best system if properly implemented. However, as you will see at our site, the proper implementation gets very technical. Therefore, we realized a long time ago that Condorcet is simply not practical for actual implemention on a large scale in the forseeable future. It's just too darn complicated.
However, Approval Voting is very simple. It's the same as our current plurality system except that the voter is allowed to vote for more than one candidate (no ranking). When people first hear about Approval Voting (myself included), they think it is defective because it does not allow you to rank the candidates (as in IRV and Condorcet). But this is misleading. IRV lets you rank the candidates, but it does not properly count your preferences. Technical analysis shows that Approval Voting is a surprisingly good system given its extreme simplicity. And it requires no new voting equipment. It could be implemented very quickly once a consensus is reached to do so, and the only objection I can see is to protect the two-party duopoly.
Think about it, folks. We could revolutionize our political system by simply letting voters vote for more than one candidate. This will have a far more profound effect than term limits or campaign finance reform, for example.
What effect it will have cannot be predicted exactly, of course. Perhaps the Republicrats will still remain dominant for a long time, perhaps not. But it's definitely worth a try, perhaps starting at the local level.
Oh, one more caveat. You must realize that *no* alternative voting system can make the US Presidential election fairer for minor parties as long as the Electoral College is in place. Trust me: it just can't be done. That's why I'm for aboloshing the EC. Unfortunately, many of my fellow conservatives are dead set against that, and it requires a Constitutional Amendment.
I watch Brit Hume on Fox News
A missing criterion that Arrow left out was proportional representation. Condercet fails that test miserably, especially in multi-seat elections.
Where in every district 35-45% of the electorate was out-favored by 55-65% of the electorate, Condercet would give each seat to the 55-65% winner in each election, completely disenfranchising the minorities.
Rank the candidates in your order of preference.
There is your sentance. Condorcet voting indicates that you vote a preference for each possible combination, however this can be simplified to just ranking them in order because it satisfies all of the possible combinations. For example:
Choose A over C
Choose B over A
Choose B over C
Choose B over D
Choose D over A
Choose D over C
Is exatly the same as saying:
1. B
2. D
3. A
4. C
But ranking is easier for people to understand.
Approval Voting (or any other alternative election method) can be used with the Electoral College in place, but then it can't help minor parties get a fair chance.
Think about it. Suppose your state uses Approval Voting and selects Nader. Now, the spoiler effect is just transferred to the national level, where Nader can spoil the race in the EC. Your state "wasted" its electoral votes on Nader. Most people will figure this out in advance (or be told) and won't let it happen.
I watch Brit Hume on Fox News
For one thing, the results don't mirror each other across the diagonal like they should (e.g. in one intersection for Kerry and Bush it said A=151,B=152,NP=9 while in the other intersection it said A=112,B=191,NP=9).
Another problem is that the legend says that "A" votes are for the person in the column headers, but if you read down the Nader column, every single item lists a higher "B" value (for the person in the row) but the results claim that Nader is undefeated.
OK, so if we assume that they got A and B backwards, we then notice that in some boxes when "A" is the winner it is colored blue, while in other boxes when "B" is the winner it is colored blue but the win-loss summary at the bottom matches the colors, not the values.
So, it looks like there is a very big problem with the program they are using to either calculate or display the results.
..wayne..
In case you didn't explore the site fully, this page explains their arguments against IRV. Personally, I find them very compelling.
What you say here leads into their arguments:
Yes there are contrived conditions where you can show that some mathematically disproportionate fraction of the populace would be "happier" with a different candidate, but look at the reality of voting in the US. 90-99% of the voters split their votes relatively evenly between the two major parties. The rest split them fairly unevenly between the remaining minor contenders.
As long as the minor parties are quite minor, IRV will just provide more interesting protest votes. And there is probably some value in that, but it isn't enough. As soon as a party or candidate becomes big enough to challenge the main two, the spoiler effect comes right back into play.
But please read the above linked page for a much clearer and more thorough explanation.
Rome wasn't bilked in a day.
Actually, there are lots of works (e.g., the ones by Kanneman and Tversky) showing that people DO follow preferences inconsistent with simple ranking.
Simple ranking assumes:
* transitivity (A > B and B > C implies A > C -- in preferences frequently that is not the case).
* complete ordering (no incomparability, that is I must distinguish between all options).
Condorcet will allow both restrictions to be lifted, but of course we know it is not perfect,
because there cannot be a perfect voting method...
So you write the number of approvals on each ballot. At that point, poll workers have to get into removing your preferences to insert theirs (which is what they'd have to do under the current system anyway).
Admittedly, the math required to do this might be beyond the grasp of the average American voter...
Daniel
Hurry up and jump on the individualist bandwagon!
With Condorcet, if four million put X over Y, and five million put Y over X, that comes to a million Y over X. This is why it's called a pairwise system...instead of one election, there are X * (X-1) elections tallied, where each person was in a hypothetical race with each other person.
Sometimes there will be a clear winner (If A won every hypothetical race), usually not. At that point what happened is called circular voting. A is better than B is better than C is better than A.
So what do we do? Well, first, we throw away D, who didn't win any virtual elections. He didn't beat out anyone, so we drop him. This doesn't accomplish anything except make the rest of the math easier, because the votes that had him in first, then C, A, and B, were already counted, unlike in IRV.
And we repeat that...if E only beat D, now he's gone. And so on.
Honestly, by this point, we'd be pretty much done with any election in recent history, but let's take our hypothetical and figure out if A, B, or C won.
So now we just have a single loop. So what we do is find the virtual election with the smallest margin of victory, and just throw it out. And we do it until someone wins all remaining virtual elections.
It gets rather more confusing if we have A beats B and C, C and B beats D, C beats B, and D beats A, and other such crazy results. But the math is worked out and doable.
IRV, the system with issues with circular ballots, is just a way to let people 'safely' vote for third parties while entrenching the two party system, anyway.
If corporations are people, aren't stockholders guilty of slavery?
What's wrong with "If one candidate beats all the others in a head-to-head, that candidate wins"? I'm pretty sure most people would understand that idea.
The difficult part, IMHO, would be convincing a politically-motivated media to run their fact files and commentary on why such-and-such a method would meet the goal(s) of a fair election, so enough people actually understood what was going on that the general population would accept it (the two not being the same thing at all).
The problem with Condorcet, for example, is going to be explaining how they break a tie -- not really an issue if you're used to a duopoly, but rather important here! What you need to make it successful is a system that is sound, which will stand up to critical examination from the few, but that can be summed up in a nutshell so the many understand how to vote. Fortunately, most systems meet the latter criterion: you say "list the candidates in order of preference", "vote for the guy you want", "tick all the guys you'd be happy with", etc.
BTW, if anyone hasn't looked, the linked site (electionmethods.org) is very well done. As a mathematician and someone who cares about elections, I found the page on technical evaluations of the various methods most interesting. The kind of criteria it presents for a good system -- the one-liner sound-bites -- are the sort of thing that should be fed to the general public, with the accompanying reasoning available for those critical enough to examine the details.
If you disagree, post your argument. (-1, Overrated) isn't your personal censorship tool for views you don't like.