Primaries Over Parties

Over at Dispatches a comment I made about the open primaries led to this question from Michael Heath.

Are we seeing an effective mutation of the general [election] into a two-stage run-off?

The background of the question is my claim that our changing primary system, particularly open primaries*, are causing a decline in the parties’ ability to discipline candidates and elected officials, because the parties aren’t the ones selecting their nominees; the public is. And if the parties can’t discipline their own elected officials, the public can’t effectively discipline the parties.†

This is a bit paradoxical. The push for primaries as the method of selecting nominees came as part of the generalized demand for more democratic structures and processes in U.S. politics. It is partly associated with the progressive movement, which also pushed for such direct democracy tools as the initiative, referendum, and recall. But it is also part of a “from the beginning” expansion of democratic forms in the U.S., from the elimination of the property restriction for voting, the shifting of selection of the electoral college from state legislatures to the public, to the direct election of senators and universal suffrage. So this is all about the expansion of democracy.

Yet political scientists say democracy is inconceivable without political parties to help organize the public and to offer them coherent policy alternatives. I’ve never found this quite wholly persuasive as an empirical claim, but as a normative claim it may have more certainty. Certainly democracy seems to function better if it is clear whom we can hold accountable. It’s easy to say we hold the individual legislators accountable, but each legislator can be pleasing to his her own district without their cumulative actions adding up to any coherent policy whole, particularly in an era of excessively gerrymandered districts. So this extension of democracy may in fact be undermining democracy.

More specifically, the candidate selection process seems to be a true slippery slope, a continuum on which there is no logical stopping place short of the far end. We began in this country with party caucuses, which were nothing more than a collection of party elites sitting around saying, “How about young Johnny Smith? He’d make a damn good Congressman, eh?” This is where the mythical image of the smoke-filled room came from. The candidate selection process was closed even to most active party regulars. This shifted in the mid-19th century to the convention system,‡ which allowed more party activists to get involved, but still excluded the rank and file. The demands of the rank and file led to the primary (the closed one), where any party registered voter could help select the nominee. This led to the open primary, which led to the blanket primary, which led to California’s “top two” primary, in which all the candidates for any given office, from all parties, are listed together (in a blanket primary, the parties are usually in separate columns), and the top two vote-getters move on to the general election regardless of party affiliation. If two Democrats are the top vote-getters, the general election will be between two Democrats.

That may be the end-point of the continuum, or perhaps eliminating the party ID from the candidates is. I suspect not the latter because most candidates will themselves want that party ID by their name. But perhaps the voters in California will collectively want to take that last step and have a true pure two-stage run-off election. That is, instead of the parties producing nominees for the public to select from, the public selects the nominees and then votes between them.

This is all very bad. First, let’s just admit that the public governs badly. Anyone who’s looked at how the initiative and referendum process has tied up California’s budget knows that. There are sophisticated theoretical and technical reasons that explain why the public governs badly, but that’s its own overly long post, to be saved for another day. Please, for the moment, allow me to just stipulate it.

Second, it’s a system that’s ripe for creating perverse outcomes. Theoretically, the “most preferred” candidate should win the election. For technical reasons it sometimes impossible to define exactly which candidate is most preferred,** but we can say that a good electoral system will not produce a winner who is clearly less preferred. However that is precisely what the “top two” primary can do.

Assume a “top two” primary with four Democratic and eight Republican candidates, in a district that is 40% Democrat and 60% Republican. For simplicity, assume every voter is party-identified, and nobody crosses party lines. Also assume that each Democratic is identically popular as each other Democrat, and each Republican is identically popular as each other Republican.

In the primary those four Democrats will equally split 40% of the vote, getting 10% each (assume just enough difference to distinguish a winner among them). Those eight Republicans will split 60% of the vote, each getting 7.5% of the vote. Two Democrats will win the primary, and so one Democrat will win the general election, even though each of the Republicans is preferred.

Advocates of ever more power to the people never think about these kinds of possibilities, and when brought to their attention they usually just dismiss them. Democracy is treated as absolute, incommensurable, so no other concerns can possibly justify any constraint on the Demos.

Noticeably, as we have shifted more nomination power to the Demos, our polis has grown more dysfunctional. It’s hard to say that this correlation necessarily has any causation, but I suspect it does, particularly in conjunction with excessive gerrymandering. (Nomination by the public encourages pandering–gerrymandering encourages pandering to the dominant extreme.) However the public places all the blame on Congress, rather than on themselves. Most ironically, the public places the blame on the members of Congress, whom they themselves chose and nominated.

Of course most people are placing the blame on other people’s representatives, and are satisfied enough with their own. But they fail to see that a decent nomination system would produce more representatives that we could all live with, that would behave reasonably and legislate rationally, both because they would be likely to be intrinsically more rational people and because the parties would be enabled to discipline them.

It’s anathema to suggest limiting democracy. The tide is clearly in favor of ever more open primaries, and to argue for rolling back primaries is to demand reversal of a two-century long tendency. Nevertheless, we are rapidly approaching–it would appear–the end point of the candidate-selection continuum. Clearly it’s not serving us well. When we get to that end point and we have a wholly disfunctional democracy, what pseudo-solution will the demophiles propose then?

.
.

_____________________________
*Definitions: Closed primary: Voters must be registered to vote under a particular party’s label, and can only vote that party’s ticket. Independent/non-partisan voters (like me) cannot vote in the primary. Open primary: All voters can vote in the primary, whether party-identified or independent/non-partisan. Ind/non-part voters must select only one party’s ballot to vote. In some states, party-identified voters can cross over by selecting a party ballot other than their own. Blanket primary: All registered voters may participate, both parties are listed on the same ballot, and a voter can vote in the Democratic primary for one office, the Republican primary for another office, the Green primary for a third office, etc.

†This is, by now, an old theme in political science, dating back to early/mid 20th century. See, e.g., this famous, if perhaps ill-considered report by the American Political Science Association’s Committee on Political Parties, one of the very few times the APSA tried to speak as an organization on an issue (beyond it’s regular call for more funding for the NSF).

‡Trivia time: 180 points to the first person to name the first U.S. president who was nominated by convention.

**That’s another long post, but one assignment I give students is an election with six candidates, where each candidate’s number of supporters is fixed and invariant, but the method of counting the votes changes. Using simple plurality voting, simple plurality runoff, sequential runoff, approval voting, and a Borda count, five different candidates win, although no voter ever changes his/her vote. In the book I adapted it from, an additional method, the Condorcet method, produces a sixth unique winner. (I don’t use that one in the assignment because it’s so slow and cumbersome, and the students are taxed enough already on this project.) The essential question is, if different methods of counting votes produce different outcomes, how can we say which candidate is most preferred?

If you’d like to play with that exercise yourself, you can download it here.

Advertisements

About J@m3z Aitch

J@m3z Aitch is a two-bit college professor who'd rather be canoeing.
This entry was posted in Political Theory, Vote Early-Vote Often. Bookmark the permalink.

27 Responses to Primaries Over Parties

  1. James K says:

    When we get to that end point and we have a wholly disfunctional democracy, what pseudo-solution will the demophiles propose then?

    More referenda probably.

  2. buddyglass says:

    Living in Texas, primaries are the only way I get vote in a way that actually matters during presidential elections. If one party has a runaway candidate (e.g. the incumbent) I typically vote in the other party’s primary since Texas has open primaries. Then, in the general election, if I can’t get behind either candidate, I just write in whoever I want, since the Republican is going to win Texas anyway.

  3. buddyglass says:

    Err…that should read “get TO vote”. Stupid typos.

  4. Heidegger says:

    buddyglass—you never need to worry about prepositions–the brain will always fill them in when reading, correctly, too, I might add. Not only prepositions–sometimes articles, both definite and indefinite.

  5. Heidegger says:

    buddyglass, another interesting aspect about visual perception and how the brain makes sense of “non”sense.

    “Arocdnicg to rsceearch at Cmabrigde Uinervtisy, it deosn’t mttaer in waht oredr the ltteers in a wrod are, the olny iprmoatnt tihng is taht the frist and lsat ltteer are in the rghit pcale. The rset can be a toatl mses and you can sitll raed it wouthit pobelrm. Tihs is buseace the huamn mnid deos not raed ervey lteter by istlef, but the wrod as a wlohe.”

  6. On the first president to have been nominated by a convention, my guess is the “dark horse,” James K. Polk, but I really don’t know.

    On the substantive issue, I personally prefer the blanket primary , not because I’m a demophile, but because I dislike people knowing whose party primary I’m voting in. Maybe it’s foolish, but here in Chicago, I would feel somewhat intimidated asking my polling agent for a “Republican” ballot. (I’m not a Republican, but might like to vote in one of their primary contests.)

  7. James Hanley says:

    Well, we know that Pierre Cornielle is not a presidential historian. *grin*

  8. Voters are dumb, but they aren’t stupid.

    When there are more than two candidates in the primary and all are, initially, equally well-liked, support will invariably coalesce around only two of them. Which is why the “4 dems and 8 reps” open primary will virtually never result in two candidates from the same party; or, in fact, in anything other than 1 dem and 1 rep. Because, if that seems likely, the other side consolidates its resources. Look north, to Washington state, for my proof.

    More importantly though, let me answer your question: “if different methods of counting votes produce different outcomes, how can we say which candidate is most preferred?”

    The answer is: extensive computer simulation. You assign hundreds computer-coded “voter’s” utilities for several candidates, and see what voting method, on average, results in the greatest total utility for society (or, equivalently, which method misses the best-possible result by the smallest margin). You do it in simulation because a) real people are unreliable judges of their (inherently unmeasurable) utilities, and b) you can simulate millions of elections in a weekend.

    The result?

    Of the methods you mentioned–plurality voting, plurality runoff (not listed on the graphic), sequential runoff (AKA instant runoff or IRV), approval voting, Borda count, and Condorcet–the hands-down winner is approval voting; and a method you did not mention, range voting (AKA score voting), which is closely related to approval, does better still.

  9. Michael Heath says:

    Dale Sheldon-Hess,

    Did any of those types include the option of voting both for a plurality of candidates and against an other(s)?

  10. buddyglass says:

    One problem I see with range voting (perhaps addressed already): it fails to work properly when people don’t understand how it works and try to “game” the system.

    Say I’m used to the system we have now, and I’m not very mathematically minded. I see a ballot with Obama, Hillary, Romney and Palin. I’m a tea party wing nut, so I’m really high on Palin. She gets a 99 out of 99. I’m hate Democrats, so Obama and Hillary both get a zero. Now I’m confronted with what to do about Romney. Obviously I like him more than Obama and Hillary (I mean, c’mon, he’s no stinking Democrat), but I really don’t want Palin to fail to win the election because of my fellow Republicans’ support for Romney. So I give Romney a score of zero as well.

    I just shafted myself, since I scored Romney equivalently to Obama and Hillary, who I actually dislike way worse than I do Romney. If enough people voted like me, we could have (as a group) cost Romney the election, instead giving it to a Democrat (since we failed to indicate any preference for Romney over the Democrat.)

    To some extent, then, range voting would seem to favor the candidates whose supporters actually *understand* how range voting is supposed to work. Perhaps the level of understanding would be equivalent across all candidates and its a wash. But I can’t help thinking that many, many people (on both sides of the political spectrum) would totally “misuse” a range ballot.

    What about a simplified version of range voting that attaches “descriptions” to each score? e.g.

    1 = I am highly enthusiastic about this candidate and would be elated if he were elected.
    2 = I would support this candidate’s election, but he is not among my top choices.
    3 = I cannot support this candidate’s election, but his or her election would not represent the worst possible outcome for me.
    4 = If this candidate were elected I would consider it among the worst possible outcomes.

    Given the above, my theoretical wing nut would score Obama and Hillary with a 4, Romney with a 2, and Palin with a 1.

    You could potentially make it slightly more fine-grained to allow for a larger range. The important aspect for most voters, though, is (in my opinion) the attached “descriptions” for each score.

  11. James Hanley says:

    I appreciate Mr. Sheldon-Hess’s contribution, but I vigorously disagree on several points.

    First, it is not true that voters “will invariably coalesce around only two” of the candidates. Mr. Sheldon-Hess makes no explanation of what might cause that to happen, or what the mechanism causing coalescence might be. As it stands, he is suggesting a magical occurrence–it just happens. I would note the ’02 (iirc) election in France, when everyone was shocked that Le Pen made the runoff. He made the runoff precisely because there were so many candidates and there was a failure to coalesce around two of them.

    Second, while I am a strong supporter of computer simulation in the social sciences (did my dissertation that way in fact), he makes an error both in assuming that you can come up with a general social utility (check the public choice lit) and in thinking that we have the knowledge–particularly in real-time–to assign hundreds of voters utilities that will adequately capture real world voters’ assessments of the candidates. I’m glad he recognizes that people are inherently poor judges of their own utilities, but amused that he believes someone who is not those people can actually better judge their subjective utilities (again, check the public choice lit).

    Third, approval voting is only a good method if everyone wants sincerely, and there’s no reason to assume people will vote sincerely. If people reliably voted sincerely, George Bush would have won at least two more states in 2000, due to Nader getting more votes. Indeed, if voters are not dumb, as he argues, then they will not vote sincerely. When strategic voting is taken into account, the Borda count is far superior to approval voting.

  12. Michael Heath says:

    James,

    I’m not following how Nader’s winning more votes would have won Bush two extra states using the approval voting method. I’m assuming Gore would have won Florida since liberals could have voted for Nader and Gore. Where am I wrong?

  13. James Hanley says:

    Michael,

    I wasn’t clear. I meant sincere voting under the system we actually use, simple plurality voting, would probably have produced a couple more states for Bush. I was only using that to demonstrate that we can’t rely on people to vote sincerely.

    And then, following the thin trail of logic bread crumbs that I tried to lay….since we know people don’t necessarily vote sincerely, we should be wary of using approval voting.

    Hope that clarifies things.

  14. James K says:

    Dale Sheldon-Hess:
    In addition to James Hanley’s points I would note that under Arrow’s Impossibility Theorem the potential for strategic voting renders the collective preferences of the electorate intransitive. With intransitive collective preferences it is impossible to create a coherent collective utility function. There is no possible way to determine which voting system produces a socially optimal outcome because there is no socially optimal outcome.

    buddyglass:
    Since I adhere to Bryan Caplan’s voter irrationality theory of government failure I would in some ways welcome a more complex voting system. A more complicated system would drive away or frustrate less informed or committed voters, and those voters are the most likely to vote irrationally.

  15. James Hanley says:

    JamesK,

    Are you reading my mind? I have a beginning draft of a post on Arrow’s theorem, and here you are scooping me.

  16. Well, we know that Pierre Cornielle is not a presidential historian. *grin*

    Ouch! 🙂
    But truth be told, I’m not even a 17th century playwright!

  17. ppnl says:

    How about a simple list vote.

    1)Voters list the candidates in order of preference.

    2)Each candidate gets a vote for each list that they are on top of.

    3)If no candidate gets a majority the candidate with the least number of votes is dropped from all the lists.

    4)Go back to step two and recalculate votes. Continue until some candidate has the majority vote.

    I don’t see any way to game this system. You don’t need primaries at all. The parties are free to decide what candidate they will support by any means they wish. Privately run elections, flip a coin, read goat entrails or whatever. Voters are free to reward or punish them any way they see fit. This totally removes party apparatus from the election process.

  18. James Hanley says:

    ppnl,

    That’s a sequential runoff system.

    If I recall correctly, every type of voting system can be gamed (but it’s been quite a while since I studied that, and the book that would answer that question for me is missing, I noticed yesterday (I really have to stop loaning books to students)). But some can be manipulated more easily than others, and I think this method is less susceptible to it than many others.

    The drawbacks are that it’s a bit awkward if there are many candidates on the ballot, and a person may end up casting a vote for a candidate he really doesn’t like (if you have to list too many candidates in your column–although its possible to avoid that by allowing incomplete lists). All in all, it’s not a bad system, although my preference is for Borda counts.

    Notably, every single one of these methods is constitutionally allowable, although the majority of Americans wouldn’t believe that on first hearing. The Constitution makes no specifications about how to count votes–the only rule is that each person gets an equally weighted vote. We often hear “one man-one vote,” but a more accurate way to say it is “one person*-equal vote.” That’s not so lyrical, unfortunately.

    ______________________________
    * Either to be more precise or politically correct, whichever you prefer.

  19. ppnl says:

    I prefer sequential runoff because it just feels like a process of consensus building. Who can object to dropping the last place in each round? Borda replaces votes with point counts which I find far less intuitive. It does seem to accomplish the same thing.

    But we are just dreaming. There are so many vested interests in the current system that it could not be implemented even if all elected officials agreed that it should be.

  20. James K says:

    James Hanley:

    I always go to Arrow first when talking about voting. You just got caught in the crossfire.

    ppnl:
    We call that system Single Transferable Vote or Preferential Voting in my neck of the woods. New Zealand uses it in some of our local body elections and Australia uses it as well I think.

  21. James Hanley says:

    Just to make sure there’s no confusion sequential runoff = Single Transferable Vote (STV). At least in the U.S., STV is the popular name, and sequential runoff is the traditional academic term, although STV is becoming more common in academic use, too.

    It’s a good system and I would happily adopt it. It’s the only thing beyond simple plurality that has any political support in the U.S., although that support is still minuscule.

    I personally prefer Borda, though, because I’m a geek. Borda has the advantage of allowing me to weight my vote. It’s a more disciplined version of my own truly preferred method–100 votes for each voter to distribute among the various candidates for each office as she sees fit. 70 for Smith, 10 for Jones, 8 for Gonzales, etc.

  22. Michael Heath says:

    Does Borda contain the weakness of supporting zealotry?

    Take Sarah Palin’s supporters, their zeal is beyond reason. All one has to do is read the book reviews for Going Rogue at Amazon to understand how divorced from reality her fans are, including Ms. Palin’s personal qualities and accomplishments.

  23. James Hanley says:

    Michael,

    Borda doesn’t really, since the number of votes you can give your favorite is fixed. It’s normally something like 5 votes for your first choice, 4 for your second, 3 for your third, etc.

    My “divvy ’em as you like” method would support zealotry, no doubt. Tamping down zealotry certain has its pragmatic value, but keep in mind that from the standpoint of democratic theory what you’re saying is, “you shouldn’t be allowed to express your preference too strongly.”

    Obviously we all agree with that in the abstract, unless we’re willing to condone political assassinations. And I suppose the campaign finance limitations on individual giving are making the same statement (part of why I oppose such limits). But while we obviously can’t give unlimited numbers of votes to each voter, and 100 was a number I just pulled out of my rectum, I think we should find some way to allow voters to express the weight of their preferences to some reasonable degree in the voting booth. If I’d had 10 votes in the last election, I’d probably have given Obama 5, McCain 4, and that jackass on the Libertarian ticket 1.

  24. D. C. Sessions says:

    First, it is not true that voters “will invariably coalesce around only two” of the candidates. Mr. Sheldon-Hess makes no explanation of what might cause that to happen, or what the mechanism causing coalescence might be.

    One important mechanism is that of endorsements. Both in the old-fashioned endorsement statement that we saw Palin sprinkling on the tea party candidates such as O’Donnell and in the back-room money sort we saw anointing GW Bush in 1999 before the first primary.

    Curiously, this means that in effect the California system cycles back to the early days of primaries: the machers of the party get together privately and thin out the field before throwing money and other support to a short list. Strategic planning enters in, of course: if there’s a real shot at capturing both general-election slots then by all means run two in the primary, but otherwise it’s best to dissuade all but one.

    If the election finance system offers little leverage to manage would-be candidates with the carrot of financial support, there’s always the stick. The machers have any number of ways of making their displeasure known, especially since most entry-level politicians are small businessmen (Whitman and Fiorina being notable exceptions) who can’t afford to bet the farm on winning the election and who can be made or broken by a whisper or two in the right quarter.

    It’s not an ironclad system, but as evolution teaches us even a light thumb on the scales can make a big aggregate difference.

  25. My last post didn’t make it. Maybe it was too long? I’ll try to break it up by topic then.

    approval voting is only a good method if everyone wants sincerely, and there’s no reason to assume people will vote sincerely. When strategic voting is taken into account, the Borda count is far superior to approval voting.

    Actually that is completely the opposite of the case, as you can see in this Bayesian regret graph here.

    Here are some sample Bayesian regret values. (Lower regret is better, like a golf score.)
    Sincere Borda: 0.10079
    Sincere Approval: 0.16549
    Strategic Approval: 0.23101
    Strategic Borda: 0.48438

    With 100% sincere voters, Borda is better than Approval Voting. But with 100% strategic voters, Borda is much worse. Once you get over 10% to 15% of the voters being strategic, Approval is better. Since the vast majority of voters are strategic in real life, Approval Voting is effectively guaranteed to be better than Borda.

    It’s pretty easy to see why this happens. Say you honestly prefer Cornell>Vedder>Cobain>Staley, and Vedder and Cobain appear to be the frontrunners.

    With Approval Voting, you would want to vote for Vedder, but not for Cobain (or Staley for that matter). And of course you’d also want to vote for Cornell, since you like him even better than Vedder, and even if Cornell has no chance, it can’t hurt you. Supporting your favorite candidate can never hurt you with Approval Voting. This is self-reinforcing. It means that if voters think X and Y are the two most likely candidates to win, then either one of them will in fact win, or the winner will be a candidate preferred by a majority of voters to both of them. (See our “Pleasant Surprise theorem” for more about that.)

    With Borda, your best tactic would be to rank Vedder first and Cobain last. But if lots of other voters use that same strategy, then the winner would most likely be whoever the majority preferred between Cornell and Stayley. In other words, Borda tends to stop the electorate’s favorite and second favorite candidates from winning — so the electorate generally gets its third or even fourth favored candidate.

    Borda himself said: “”My scheme is intended for only honest men.”

  26. First, it is not true that voters “will invariably coalesce around only two” of the candidates.

    In almost every deterministic single-winner voting method, the basis of strategy is to identify the first and second most likely winner, and then to help the one you prefer as much as possible relative to the other. With ranked ballots, the simple way to act on that strategy is just to rank one of them in first place and the other at or near last place.

    This is true of Instant Runoff Voting.
    http://www.electology.org/analysis/irv-plurality

    It’s also true, in general, for Condorcet methods.
    scorevoting.net/CondBurial.html

    But the more important thing is, regardless of any of that complicated mathematical theorizing, decades of empirical evidence says voters actually do exaggerate — probably just because it intuitively “feels” like a way to give their ballots a stronger effect. We at the Center for Election Science call that the “Naive Exaggeration Strategy”, and here’s some empirical evidence for it, from Australia (which has used ranked voting since 1918).
    scorevoting.net/AusAboveTheLine07.html

    In virtually all ranked voting methods, that naive exaggeration predestines those democracies to two-party duopoly.
    scorevoting.net/NESD.html

    By contrast, most of the nearly 30 countries that use a delayed “Top-Two Runoff” (or “TTR” for short) have three or more viable parties. (France, by the way, uses TTR, not a ranked system.)
    scorevoting.net/TTRvIRVstats.html

    Here is some speculation as to why that happens. (It seems to primarily be related to the fact that TTR gives voters a second chance, so they feel more safe in supporting their sincere favorite in the first round.)
    scorevoting.net/TTRvIRVreasons.html

    Clay Shentrup
    The Center for Election Science

  27. he makes an error both in assuming that you can come up with a general social utility (check the public choice lit) and in thinking that we have the knowledge–particularly in real-time–to assign hundreds of voters utilities that will adequately capture real world voters’ assessments of the candidates.

    I don’t know what you mean by a “general social utility”. Voters have specific utilities, along with internal estimations of those utilities — which aren’t perfect because they are clouded to varying degrees by ignorance and irrationality.

    Voters vote on the basis of their estimations of their utilities. The social utility of a candidate is the sum of all voters’ actual utilities. That is mathematically proven, by the way.
    ScoreVoting.net/UtilFoundns.html

    Utilities don’t have to capture real world voters’ assessments in order to be realistic. There are numerous ways to make fairly realistic utility distributions. You can start with a 1-dimensional “left/right” issue axis, where a voter’s utility is inversely proportional to his distance from a candidate on a line. You can then add more dimensions, and vary your distribution patterns (e.g. Gaussian, bi-modal). You can even use real-world data to do a regression fit, where you attempt to find a utility distribution function which closely matches actual real world ballot data. Curiously, what we actually find is that the more “realistic” you make the utility distributions by adding these extra dimensions and complex aggregators, the closer the personal utilities come to appearing random.

    But the really significant thing is this. Score Voting (aka Range Voting) outperforms all commonly proposed alternative voting systems, regardless of which utility distribution model or environmental parameters you tune (e.g. number of candidates, number of voters, ratio of strategic vs. sincere voters, ignorance levels).

    If while performing the simulations, the relative performance of various voting methods were to change significantly depending on the utility function was used, then there would be doubt that any particular function was the “most realistic” one. We wouldn’t know which result was most true to reality. But the reason Warren Smith (the Princeton math Ph.D. who performed the simulations Dale mentioned) was so impressed with Score Voting (aka Range Voting), is that it came out first place in every permutation of environmental variables he tried. And over most of the alternatives, it was by such a large amount that it left lots of room for error.

    So the simulations are actually quite reliable. And in fact, I believe they are far more scientifically valid than any alternative measure that has currently been devised or employed.

    I’m glad he recognizes that people are inherently poor judges of their own utilities, but amused that he believes someone who is not those people can actually better judge their subjective utilities (again, check the public choice lit).

    No one is attempting to judge any voter’s subjective utilities. That is not how Bayesian regret calculations work. We use simulated humans because we cannot read real voters’ minds.
    ScoreVoting.net/WhyNoHumans.html

    And the “public choice lit” is just way behind the state of the art in social choice theory. In just the last decade, mathematicians like Warren Smith and Forest Simmons have essentially revolutionized voting theory with advanced computer modeling and groundbreaking mathematical theorems that aren’t even discussed in academia for the most part. Examples:
    ScoreVoting.net/SimmonsSmithPf.html
    ScoreVoting.net/WtPositnl.html
    ScoreVoting.net/CondPF.html
    ScoreVoting.net/BestVrange.html

    That is a common pattern, by the way. For instance, I am currently working as a software engineer at Pivotal Labs, where Pivotal’s engineers are helping my team to learn better workflows and best practices, from pair programming to test-driven development. These processes, tied to the “Agile” and “Extreme Programming” schools of thought, were not even mentioned in my university classes. It can sometimes take decades for academia to catch up with discoveries made by independent researchers.

    Beyond that, political science is largely considered to be one of the liberal arts, and institutionally governed as such. And so the extremely mathematical and game theoretical aspects of social choice theory are often misunderstood and/or under-appreciated by the very people with the potential to insert them into the curricula of their respective institutions.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s