« Wednesday Samuelblogging | Main | Wednesday Bonus Solidfoodblogging »
Wednesday, March 03, 2010
The Lottery
The voters of Burlington narrowly voted to repeal their four year old experiment with instant runoff voting. As I tweeted earlier, I find that disappointing (though not entirely surprising).
Some people complained that the system didn't work as advertised or is patently unfair. I disagree with both of those charges.
I think pretty much every objection to IRV really stems from a cultural experience we have in the US. In the rest of the world it seems to have more acceptance, perhaps because voters elsewhere aren't trapped by the hegemony of a two-party system and are used to things like proportional representation in parliamentary systems. Here we expect binary choices, whether it be ballot initiatives like the one that got rid of IRV yesterday or tossing a coin between the Democrat and Republican.
In the last two mayoral contests IRV did precisely what it was supposed to, but voters weren't really prepared for how the mechanism cranks through a multiparty ballot. Really, the results aren't all that unexpected: the socialist candidate wins in a town that has consistently elected such people?
It's just that the process is unfamiliar and perhaps a bit more complicated than our usual one-off ballot schemes. So much like with global climate change denial or charges that NASA faked the moon landings, there's enough "weirdness" for people to latch onto that flies in the face of their experience and thus they can be easily convinced they've somehow been sold a bill of goods.
Seems like the 2005 election was cool for people, but 2009 freaked them. The eventual winner was in 2nd place after the initial round (nobody ever comes back in a second traditional runoff, of course), beating out the "leading" Republican and a Democrat who polled better than either them in head-to-head matchups (which accurately reflect a multiparty IRV election, right?), so clearly this system is bogus (I'm not sure that's a strawman). Whatever.
There was enough concern by voters, mostly in the conservative/Republican wards, that the city has returned to its old system. Now if a candidate gets 40% of the vote, they win--otherwise there's a runoff between the top two vote getters. Yeah, sounds like majority rule to me!
That seems weird because we do have a general distaste for plurality. In our state we require a majority for the Governor to be elected, lest the decision reside with the Vermont House of Representatives (Dean narrowly won post-civil unions against two opponents).
And remember how some people called Clinton illegitimate because he didn't win 50% of the popular vote in a 3-way race (his large Electoral College majority notwithstanding)? The US Constitution requires a majority of Electoral Votes to select a President, not just getting more than the other candidates, otherwise there's an almost-instant runoff in the US House right after the ballots are counted.
So it's not like we're against the idea of multiple rounds in an electoral process. But for some reason a goodly number of folks don't trust an approach that has it all happen at once instead of requiring more campaigning and another expensive election wherein fewer voters turn out. From where I sit, going into overtime seems more clunky and unfair than getting it all done with on election day.
If I'm not mistaken, most distaste for IRV comes from Democrats and Republicans. Makes sense because this does threaten the established order.
But let's think about this from an outside, alternative vote perspective, which in the end could help the two dominant parties. If I'm not all that happy with the Dems' nominee because he's not progressive enough or whatnot, maybe I'd prefer to vote for the Green. But I'm stuck in something akin to the Prisoner's Dilemma.
Should I fear enabling a Republican's victory, my absolute last choice even after Satan himself, because I've diverted a vote away from my second choice, I have to gamble and either make a protest vote that could have unintended consequences or just throw my lot in yet again with "lesser evil". And speaking of gambling, here's my buddy Thoreau for the eleventy millionth time:
All voting is a sort of gaming, like checkers or backgammon, with a slight moral tinge to it, a playing with right and wrong, with moral questions; and betting naturally accompanies it. The character of the voters is not staked. I cast my vote, perchance, as I think right; but I am not vitally concerned that that right should prevail. I am willing to leave it to the majority. Its obligation, therefore, never exceeds that of expediency. Even voting for the right is doing nothing for it. It is only expressing to men feebly your desire that it should prevail. A wise man will not leave the right to the mercy of chance...
I leave less to chance when I can provide more information about my preferences. This frees me to cast a ballot more consistent with my conscience because I can indicate not only my first choice but also what I'd rather see as an outcome should the question still be up in the air when the majority's positive will is not yet fully expressed.
Thus in 2008, I might have felt more empowered to work for McKinney's campaign to get as many 1st choice votes as possible, vote for her myself and then put Obama down for 2nd. In 2000, of course, Green voters could still have voted Nader without necessarily throwing Florida's Electoral Votes to Bush.
I suspect if more people took a risk on non-duopoly candidates in an IRV environment, independents and alternative parties could make a better showing for public financing, inclusion in debates, and even win significantly greater than a couple hundred offices out of tens of thousands in the US. That would lay the ultimate foundation for more choice at all levels of government, making it more representative of our diverse constituencies.
So yes, I lament Burlington's capricious decision to regress to the plurality threshold. No democratic process is going to perfectly reflect the desires of the electorate, but a system that allows for more nuance better approximates the collective will. This is a little setback for statewide and nationwide electoral reform--it would be nice to point at a successful implementation as we try to expand its presence--but hey, old habits die hard and tearing down the walls wasn't ever going to be easy.
ntodd
March 3, 2010 in Why We Fight | Permalink
TrackBack
TrackBack URL for this entry:
http://www.typepad.com/services/trackback/6a00d8341c525c53ef01310f5bf257970c
Listed below are links to weblogs that reference The Lottery:
Comments
A Condorcet vote would be much better, IMHO.
Posted by: Mike Goldman | Mar 4, 2010 6:08:52 AM
Certainly that has a more elegant mathematical basis and could claim to best represent the collective will, but it's so much more complicated than IRV, which at least has some semblance of our usual plurality resolution mechanisms, that it would be a tough sell out of the gate. I doubt people would trust it, especially since they can't even fathom the different between head-to-head polling and multiround voting.
I'd love to experiment with lots of different methods at the local level, and perhaps state level in the smaller states like Vermont. I think that we could start with IRV, with it's relative simplicity, and then once we're used to alternative voting/counting mechanisms, we'd be able to try some others.
What amazes me is that the pollsters haven't figured out survey models to deal with multiparty/multiround elections. I would've thought they'd have a field day.
Posted by: NTodd | Mar 4, 2010 10:15:23 AM
IRV will be used in our elections for the first time this November here in Oakland (and Berkeley).
Posted by: Mike Goldman | Mar 5, 2010 4:21:22 AM
Excellent! Hopefully your voters won't make the same mistake Burlingtonians did in having the IRV work exactly as it's supposed to.
Posted by: NTodd | Mar 5, 2010 10:43:34 AM
I'll keep advocating Condorcet though, maybe it keeps the Overton window shifted in favor of IRV too. :)
We use Condorcet voting in the Debian project. It's open source and everyone trusts the results. You have to use a method to eliminate cycles, but the result reflects consensus.
With IRV, I'm not sure you get anything like consensus. You slightly open the door to a third party/candidate, but certainly not more than that.
Posted by: Mike Goldman | Mar 6, 2010 1:07:10 PM
I think you can get consensus with IRV if people have the expectation that they're merely expressing preferences and their second or third choices still might not win. Problem is consensus, at least in the Iroquois tradition, is quite different than majority rules and more difficult to achieve, so IRV seems to be a compromise position of sorts.
I absolutely dig the idea of Condorcet and appreciate your support of that. That's why I'd like to see lots of local experiments, even "failing" ones, so people get used to the idea that our binary system isn't necessarily the best. Keep moving that window!
Posted by: NTodd | Mar 6, 2010 1:20:59 PM
Consensus is tied in with the idea of consent, so whenever there is non-consent in the result there is no consensus. Obviously with the FPTP-style election you can still have consensus most of the time, but it is uncertain when plurality candidates win with no majority.
IRV lets you vote ranked ballots, which is good. And most of the time you get consensus for the same reason FPTP does, but that does not mean it produces the consensus result necessarily. When there is no clear majority for one candidate, the Condorcet winner may be the first to be dropped!
Suppose you have candidates A, B, C. Order representing preference, if 40% voted ABC and 40% voted CBA, with the remaining 20% voting for B as their first choice, B represents the consensus choice, but IRV ensures that either A or C must win.
Posted by: Mike Goldman | Mar 7, 2010 10:21:03 PM
It isn't any harder to vote Condorcet than IRV...the only difference is the counting method. So the only argument against the Condorcet method that makes sense to me is that voters are supposedly too stupid to understand. I don't believe that. I think the media is too stupid to explain it, however.
Posted by: Mike Goldman | Mar 7, 2010 10:32:59 PM
http://en.wikipedia.org/wiki/Condorcet_criterion#Instant-runoff_voting
"In cases where there is a Condorcet Winner, and where IRV does not choose it, a majority would by definition prefer the Condorcet Winner to the IRV winner."
Posted by: Mike Goldman | Mar 7, 2010 10:49:54 PM
The argument against Condorcet is that people don't get IRV, which is really straightforward but the results can be unexpected given our traditional voting experience. Condorcet is more sophisticated in its counting mechanism, so if people can't accept IRV, they certainly won't accept something that appears to be smoke and mirrors.
So you experiment with different systems, get people used to the idea that our voting method is not the only kind and not even the "most fair" and eventually we'll have something that enables more diversity and democratic choice.
Posted by: NTodd | Mar 7, 2010 11:46:40 PM
But I don't get IRV, since it fails to provide the solution to the problem of plurality creating a majority opposed to the winner. It solves nothing at all, it only makes the voting more complicated.
Posted by: Mike Goldman | Mar 8, 2010 3:15:19 AM
The single "benefit" of IRV is that it reduces the spoiler role of third party and independent candidates. For instance, a right wing nutjob can then run safely without splitting the vote from a more "moderate" Republican.
Posted by: Mike Goldman | Mar 8, 2010 3:19:04 AM
This isn't about experimenting, it's about what works, mathematically, and what does not.
Posted by: Mike Goldman | Mar 8, 2010 3:25:26 AM
But I don't get IRV, since it fails to provide the solution to the problem of plurality creating a majority opposed to the winner. It solves nothing at all, it only makes the voting more complicated.
No, it creates a winner with a majority. There are always "winners" in the voting systems, the difference is how you define them based on voting and counting method. Since we have a tradition of avoiding plurality in most cases (in BTV the threshold is 40%, though), we accept runoff systems as providing a resolution. IRV simply compresses the timeframe for that, allows people to express preferences up front, and achieves a greater granularity and consensus than a simple FPTP requirement by allowing more players in the game.
The single "benefit" of IRV is that it reduces the spoiler role of third party and independent candidates. For instance, a right wing nutjob can then run safely without splitting the vote from a more "moderate" Republican.
I'd say that is a great benefit. Diversity is important in all things, especially in a democratic system. If you're going to suggest that it's bad because a RWNJ could win an election, we might as well do away with all voting!
This isn't about experimenting, it's about what works, mathematically, and what does not.
But representative democracy isn't about math really. It's about creating mechanisms for expressing popular will, and there are myriad ways to do that. All of them work and it's a question of approximating the will of the people and creating an environment where they feel it has been done so and the winners, however defined, are legitimate.
Democracy in essence is experimentation. We've played with different systems in our history, including the College, proportional allocation of Electors, Town Meeting, caucuses, etc. They function because people accept them, even with grumbling about the process, candidate choice, unpalatable results, etc.
So how do you get people out of the old mode? Not by establishing a new "perfect" system, but by trying new methods and getting people used to them in the crucible of the more local contests where the impact isn't global and people already have greater trust relationships with the people and the process.
I'm certainly willing to accept Condorcet is "the best" method, better than IRV. I'm of the mind, however, that IRV is an easier sell in the short-term because there are already so many efforts underway and a mature movement infrastructure, we can take existing runoff methodology and just show it as a natural evolution, point to lots of examples where it is used in national systems, and then as people grow comfortable with new implementations, and new candidates enabled by IRV's breaking the duopoly logjam who will further advocate for electoral reform, we can settle on something else. Like any other fundamental change, it's a generational process, and we'll need Condorcet proponents to keep pushing, and we'll need other method advocates as well to shift the window.
Posted by: NTodd | Mar 8, 2010 11:43:56 AM
Todd, I'll make a final comment on this thread and leave it there, you know I agree with you that IRV is a step in the right direction but mainly because it provides for ranked ballots which is the exact same thing needed to do Condorcet voting. IRV does not make the RWNJ more likely to win an election, it makes the RWNJ less likely to spoil an election for the "mainstream" Republican. In other words, it preserves the duopoly more than it diminishes it. That having been said, by removing the spoiler role there is less reason for the duopoly to restrict the field of debate and participation. IRV does not lead to a majority result. That is inaccurate. It leads to a countermajority result when it eliminates the Condorcet winner. But it may be the best compromise that can be achieved now.
Posted by: Mike Goldman | Mar 9, 2010 5:22:35 AM
hmmm... .I doubt people would trust it, especially since they can't even fathom the different between head-to-head polling and multiround voting.
Posted by: Nursing cover | Jun 30, 2010 11:29:44 PM




