The way we vote: why it's not good, and how to fix it

Over on his blog, “The Least of All Evils,” Dale Sheldon-Hess just put up a posting called “The Tyranny of the Majority Weak Preferences,” which really goes to the heart of the problems we have with our plurality voting system. Our voting system tends to favor extremists, as we are seeing now. (And some people who advocate “instant runoff voting” don't seem to realize that it favors extremists, if anything, more strongly than plurality does.)

Please read that post. And if you want to know what kind of a system might eliminate this bias toward the extremes, read William Poundstone's book, “Gaming the Vote.” It describes even more of the flaws of our plurality voting systems, and points the way to fix them.

You have to understand the rules!

As I mentioned yesterday, the centrist blogger Solomon Kleinsmith is starting a new website called “Uniters.org,”. One of the things that I saw when I looked at that site Tuesday is a plug for the hopeless organization called Americans Elect, a group of centrists who are following in the footsteps of the Unity '08 group who tried to produce a centrist ticket for the last Presidential election. And the fact is, people who want to advance the centrist cause and try to do it in that way do not understand how our system works.

Over two hundred years ago, in 1800, Anthony Lispenard was a Presidential elector from New York State. In those days, the electors did not, as they do today, cast a differentiated vote for President and Vice-President. Instead they simply cast two votes, ostensibly for the Presidency; the candidate with the largest number, if it was at least a majority of the number of electors, got the Presidency, while the candidate with the second-largest number got the Vice-Presidency. So supporters of the Democratic-Republican ticket of Thomas Jefferson and Aaron Burr simply voted for both. In previous elections, and among the rival Federalist Party even in 1800, a small number of electors (even just one!) would vote for the party's Presidential candidate, but vote for someone else, who was not a real candidate, with their second vote, so as to assure that their Presidential candidate got more electoral votes than their Vice-Presidential candidate. (In 1796, however, too many Federalist electors did this, so that Thomas Jefferson, the Democratic-Republican candidate for the Presidency, outpolled the Federalists' Vice-Presidential candidate and became the Vice-President!)

Lispenard, in fact, preferred Aaron Burr, the Democratic-Republican candidate for the Vice-Presidency, to Thomas Jefferson, who was again the Democratic-Republican candidate for the Presidency, and tried to accomplish this by voting for Burr but not for Jefferson. But he didn't understand the system! He cast two votes for Burr. Now if he had voted for Burr and someone else as long as that someone else was not a New York State resident (the rules stated that an elector's two votes had to include one who was not a resident of his own State), Lispenard could have elected Burr President, with 73 votes to Jefferson's 72. Anthony Lispenard was one of the few people in history who could have, by one act, changed the course of history! But he didn't understand what he needed to do, and voted twice for Burr. Now this would not work. An elector could not vote for one person twice. And certainly if he was a resident of your own State, as Burr was in Lispenard's case, even if an elector could vote for one person twice, he could not vote for that person twice because one of the votes had to be for a candidate from a different State! So Lispenard's vote was changed to a vote for Jefferson and Burr, and he did not get his wish simply because he didn't realize how simply he could have assured Burr the Presidency!

What does this have to do with Americans Elect? Simply, in the system we have, which is Plurality Voting, any candidate other than the top two simply hurts the candidate who is closer to him. If a “centrist” candidate appeals to more Republicans than Democrats, he is likely to cause the Democrat to be elected. And if he appeals to more Democrats than Republicans, he is likely to cause the Republican to be elected. So unless he is truly dead center, which is pretty unlikely, he's going to have the effect that the people who voted for him will feel they've shot themselves in the foot. (Probably the reason that Bill Clinton was elected President in 1992 was that more Republicans were impressed with Ross Perot than Democrats were.)

As I have previously said, the way to get centrists and moderates elected, which is what Uniters.org ostensibly wishes to accomplish, is to change the voting system, not start a quixotic third candidacy.

Can we do something to advance the moderate cause? Yes!

Regular readers of this blog know that I generally take positions that can basically be described as right-of-center, and many of my posts have been in advocacy of moderate Republican candidates. But I think that the cause I am about to put forth in this post is one that moderates of both parties and independent centrists should embrace, however we might differ on individual candidates. The thing is: we need to change the way we elect people to office.


The system we use for nearly all elections is called plurality. You can vote for only one candidate, and the one with the most votes wins. If there are three acceptable candidates, you need to pick one, and if some of your allies pick a different one, the vote for the good candidates is split, so that a candidate that you really don’t want gets elected. Because of the results of such a split, we need primary elections to pick one candidate of each party, and it is in these primaries that moderates are getting crushed by extremists of both the left (in the Democratic party) and the right (in the Republican party).


We have seen this over and over. Arlen Specter, who had served with distinction in the Senate as a moderate Republican representing the state of Pennsylvania for many years, figured he could not win a Republican primary – the last time he had run, he came perilously close to defeat at the hands of Pat Toomey, who was going to run in the primary again. So he switched to the Democratic party, where he thought his chances might be better. That did not help; he was beaten by a left-wing Democrat, Joseph Sestak, in that party’s primary. Polls showed that Specter might have won an election in which all Pennsylvanians could vote, but at the primary level, Toomey would beat him among Republicans and Sestak among Democrats. So the Senate was deprived of a voice for moderation.


Last year, at least two Democrats, Chris Coons in Delaware and Harry Reid in Nevada, were elected although the polls showed that both could be defeated if the Republicans had put forth moderate candidates. And there were available candidates: Mike Castle in Delaware, a very popular member of the House of Representatives, who had had to run statewide since Delaware had only one Representative, and Sue Lowden in Nevada, a State Senator and former state Republican chairwoman. But Tea Party extremists in the Republican primaries forced the nomination of Christine O’Donnell in Delaware and Sharron Angle in Nevada, both of whom were guilty of such political gaffes as to make them objects of ridicule by the time of the November election.


Some have said that the solution to this is opening up primaries to voters who are not members of the party in question. But that would be a case where the cure might be worse than the disease. Democrats might vote in a Republican primary to pick the candidate who would most easily lose to a Democrat. So a better solution is needed. And that is to have a better voting system. There are two broad types of solution. One is to have proportional representation, where a party that gets 25% of the votes in an election gets 25% of the seats in the body being elected. This is, of course, not possible in an election for an office like a mayor, governor, or president, so for such offices it is necessary to have a system where vote splitting cannot hurt a candidate as it does in plurality voting.


Some people say that proportional representation leads to instability, because it leads the way to the representation of lots of third parties that cannot form majorities; it has in such cases as France and Italy shortly after World War II and Israel more recently. But there are ways of setting up proportional representation that do not lead to this instability; in Germany and, in recent years, New Zealand, a system has been adopted that works and allows third parties to exist without the unstable governments that were seen in those other countries.


But as stated before, there are offices like mayor, governor, or president, which cannot be divided proportionally. So single winner methods need to be devised where a voter can vote his conscience without worrying that splitting the vote would elect his worst choice. There are systems that will do this. A site, which I created, that compares voting systems is available at http://important-information.net78.net/vote/. There is an excellent book by William Poundstone called Gaming the Vote, which also discusses this problem. And Poundstone, as well as I, has concluded that the best system to use is one called “range” or “score” voting, in which you rate all the candidates, so you can (in one version) give a 10 to the candidate(s) you like best, a 1 to the one(s) you like least, and intermediate values to others. So conservatives might give 10s to the most conservative candidates, 1s to the most liberal, and 4s, 5s, or 6s to moderates. Liberals would do the reverse, and moderates might give 1s to the extremists and 9s and 10s to moderate candidates. You would not need primaries, because if there were five Republicans in an election, they would not split the Republican vote, but Republicans could give high ratings to all of them. And the winning candidates would be the ones who got the most support all along the spectrum, leading to many more moderate victories.


An alternative system, known as Instant Runoff Voting, has been proposed by a group known as FairVote. It is essentially the system that has been used in Australia for many years, and proponents of this system have actually gotten it adopted in places like San Francisco; Aspen, Colorado; and Burlington, Vermont. In a few of those places, it was adopted, and has since been rejected, because it can lead to such bizarre results as in the Burlington, Vt. mayoralty election of 2009, where the candidate who might have been a good compromise, receiving support from nearly all the voters, was actually eliminated in the course of the counting, because he was the first choice of comparatively few. In such an election, some candidates who might be the most acceptable to the voters at large get eliminated, and usually the result is as much polarization between extreme candidates as the worst cases we have seen in plurality voting, so moderates, particularly, should resist the propaganda for instant-runoff voting.


You can read more about range/score voting at http://rangevoting.org – a very useful site.

Stealing domain names and cheating

The Center for Range Voting is a group that advocates what they (and I) consider a better way of conducting elections: a method that has been called both "range voting" and "score voting." (The latter describes it better, so I prefer it.) They have their own website, with the URL rangevoting.org .

A rival group, officially named the Center for Voting and Democracy but now mostly using the name of FairVote, is similarly engaged in advocating changes in the voting system, but mostly favors a system called "instant runoff voting" or "the alternative vote," which has mainly been used in Australia, but in recent years has been tried here. (A couple of years ago it was used to elect the mayor of Burlington, Vt., with disastrous results, showing the weakness of this method, though it has worked fairly well in Australia.) These groups, both advocating different election method changes, find themselves in conflict, with CRV usually having the more logical arguments, but FairVote having the most convincing propaganda machine to many people. But now FairVote seems to have crossed the line and started playing dirty. They set up a blog, with the URL rangevoting.com . It has been set up very recently, apparently to hijack people who intend to get to rangevoting.org but get the URL wrong. (This was reported on Dale Sheldon-Hess' voting methods blog, "The Least of All Evil.")

Has FairVote no sense of ethics?

Voting methods

Recently I made a post on another blog in which I've expressed some of the same thoughts I'd earlier expressed about open primaries. A subsequent commenter made a posting in which he advocated a system that was originally called "the alternative vote" and is now often called "instant runoff voting." I once liked IRV, but unfortunately it can often lead to a siituation in which centrist candidates are rejected in favor of extremists. So it's particularly unfortunate that someone trying to form a centrist party is pushing IRV.

Readers are encouraged to read the links I've made in this post, also an article on a better system, Bucklin voting, and the best system in my opinion, score (or range) voting. Don't make up your minds untill you are familiar with all the options.

Open primaries, "top two" primaries, and other ballot reforms

Last month I made a post in which I indicated my objections to "top two" primaries, as well as to the more traditional type of "open" primary used in states like Virginia. In this, I differ from Solomon Kleinsmith, who likes open primaries, though agreeing with me about "top two" primaries. Now again I see a posting — by Dale Sheldon-Hess — which points out the problems with "top two" primaries, a point with which I concur, but makes other suggestions I cannot endorse.

Dale Sheldon-Hess makes one point of terminology that needs making: open primaries are not necessarily top two, though some people seem to conflate them. And one of his suggested reforms in his post is one I like as well: score voting. But in the same sentence in which he proposes score voting as a way to improve our voting methods, he groups with it approval voting — a system that can be viewed as a simplified version of score voting, but without its good points and with some serious bad points. (I'll get on to my objections to approval voting in another post.)

Open primaries? No thanks! (But there is a good alternative!)

Solomon Kleinsmith is the owner of a centrist blog: this means, of course, a little bit to the left of my politics, but not so far that I don't agree with a lot of his positions. But one thing he favors that I cannot is open primaries.

You might think that allowing Democrats and Republicans to vote in each other's primaries might moderate their choices, but look what this could cause: Consider next year, when Barack Obama is certain to be the Democrats' nominee. Since a Democrat has no likelihood of affecting the primary process in his own party, he might decide to cross over and vote in the Republican primary, and vote for the Republican that Obama has the best chance to defeat. As a result, the Republicans would have someone like a Sarah Palin nominated, just to make it easier for Obama. (Think of Christine O'Donnell in Delaware last year. She wasn't nominated in an open primary, but open primaries might cause a lot more of this type of nomination.)

Of course, one might argue that this sort of thing hasn't much happened in those States that do have open primaries. Well, let's see what has happened in those States. Virginia is an open-primary state; it has to be, because Virginia law does not allow voters to register in a paty, the way Maryland and New York (to mention the two States where I've been registered to vote in my lifetime) do. Wisconsin is very similar to Virginia in this respect. Now can anyone argue that Virginia candidates have been more moderate, more centrist than, say Maryland candidates? I don't think they have been. And certainly the recent impasse in Wisconsin has shown that in that State, exteremism has not been driven out of local politics.

Now Kleinsmith does agree with me on another proposed election rule change: he strongly opposes "top two primaries." These were adopted statewide in Louisiana and subsequently in a couple of other states; the same system is used in my county for the nonpartisan Board of Education elections. In a "top two primary," all candidates (in a partisan election, all the candidates of all parties) run against each other, and two (who might be both from the same party!) are chosen to run in the final election. In Louisiana, this has led to the bizarre "lizard/wizard" election of 1991. It is clear that it causes all sorts of problems, and a whole chapter was written in William Poundstone's book, "Gaming the Vote" about the 1991 Louisiana Gubernatorial election. Suffice it to say that that's not the answer either.

Is there a solution that will help moderate candidates get elected? Possibly. Poundstone's book suggests that another system would be better: score voting, also known as range voting. In this system, the voter rates all the candidates instead of picking just one. And a moderate candidate might, by getting high, but not maximal, ratings from both left-of-center and right-of-center voters, actually win in a score voting election. Check out the Center for Range Voting site, or Poundstone's book. You might be impressed.

Perils of plurality voting

One of the perils of plurality voting, the system we use here in the US for most elections, is that the candidates most similar to each other are the ones who will attack each other the most viciously. There are two candidates for Governor of New York in yesterday's election who could be described as "libertarian": Warren Redlich, who actually had the Libertarian Party's nomination, and Kristin Davis, who ran under the banner of the Anti-Prohibition Party (apparently favoring legalizing prostitution). Davis had tried to gain the Libertarian nomination, and before Redlich had secured it, he called her a "whore." (She had actually run an "escort service," though apparently she was not a prostitute herself.) Later on, just before the election, campaign flyers were distributed, it would appear by members of Davis' campaign staff, though with the actual connection obscured by using a fake sponsor's name, calling Redlich a child molester. Obviously, these two candidates, who were seeking to get the same voters' support, felt it necessary to attack each other rather than their more remote opponents.

And this is only one such example. I remember one election, many years ago, also in New York when I still lived there, when the Socialist Workers' Party and the Socialist Labor Party candidates found themselves on adjacent lines of the ballot. Now it is granted that these two parties differed more than their names would imply, but they were the most left-wing of the parties on the ballot and were obviously competing for the votes of those who favored some sort of socialism. So again, they felt it necessary to attack each other rather than their more remote opponents.

It's one of the plurality system's biggest flaws. Candidates who should be boosting each other as second-best end up fighting tooth and nail. If we had score voting, or approval voting, or even instant runoff voting (though supporters of SV and IRV mostly oppose each other's ideas almost as strongly as supporters of rival, but similar, candidates in plurality do!) you would not see this. Candidates who were rather similar would be more likely to boost each other since it would not hurt themselves to do so.

That alone is a good reason to change our voting system.

A viable third party? I'd bet against it

Over on the "Big Tent Revue" blog I'm reading yet more about the prospects of a centrist third party. I can't see it happening. The rules of the game just make it imposssible.

First of all, we have "single-member plurality" elections for all offices. It is really well known that this leads to the demise of minor parties. This fact even has a name: Duverger's law. There are countries with SMP that have some viable minor parties, but they are either regional, or they have very few seats despite the number of votes they receive; look at Britain for examples of both. [Regional parties like Plaid Cymru and the Scottish Nationalist Party do reasonably well. But the Liberal Democrats, while getting nearly as many votes as Labour in the last election (23% vs. Labour's 29%), got just enough seats to make it into a coalition, less than 9% of the parliamentary seats.]

But even those 9% of the seats were enough to make a difference, because Britain is a parliamentary country and it required a coalition to form a government. In a presidential system like the US, you don't ever need to form coalitions. So third parties just die. Instead we get coalitions within the parties. Neither the Republican nor the Democratic caucus in either house of Congress is monolithic, and even a Joe Lieberman, defeated in his party's primary, joins the Democratic caucus when he's elected. Similarly, James Buckley, who ran as a Conservative against the Republican in New York in 1970, joined the Republican caucus. That's the way it has to be; the rules of both houses of Congress make them operate as two-party systems as well.

So I just don't see any third party in the near future.

Board of Education elections

As I said in yesterday's post, I do not vote in elections for the Board of Education in my county. There are three reasons.

First, I think boards of education should not be elected in the first place. These are technical positions, not political ones. They should be reserved for people who know something about education. In my native city of New York, the Board was appointed by the Mayor, though there were local boards elected by the people, clearly below the citywide board in rank. And I think that is the way they should be. (In the case of my county, where the school system is countywide, not tied to a city, this appointment would be by the County Executive.)

Secondly, if one feels that the public should have input into the way the schools are run, and so there needs to be an elected board, certainly only people with children in the school system should vote. I can't vote in elections that don't concern me because I'm geographically out of their jurisdictions, like the Governor of California or the Senate race in Florida, though I actually care who wins in those elections. (And the Senate race actually does concern me, because the winner will vote on organizing the Senate!) So why should I be eligible to vote on the membership of the Board, when I do not have now, nor ever have had, a child or children in the schools?

The third reason I do not vote in elections for the Board of Education is simply that they are non-partisan, and I have little use for non-partisan elections. While in a party primary I do not have the guidance of a party label to help me choose who gets my vote, just as in a nonpartisan election, the stakes are higher in an election with no subsequent general election. In a primary I can at least assure myself that regardless of who wins, the winner will be a Republican. I do not want to vote without at least the clue of party label to help me make my decision. (In fact, I did vote once for a Board candidate. He had, years earlier, run as a Republican for County Executive, and my vote was essentially a thank you for running in an election with little chance of winning, in the past.)

We need Proportional Representation!

The 2010 election is getting closer — less than three months away. But the chances that I will find a person to vote for to represent me in any legislative body are really remote. I live in a district that is represented in the House of Representatives by Chris Van Hollen, one of the most liberal members of the House. The VoteView rankings show Van Hollen as significantly more liberal than Nancy Pelosi! And since last year I have lived in the State legislative district that Van Hollen represented as a State Senator. Lots of luck finding anyone, in Annapolis or Washington, to represent me!


The problem is that our electoral system, known technically as Single-Member Plurality (SMP) generally allows for only one member of a legislative body to represent a whole geographical area. (In Maryland, the lower house of the State legislature, the House of Delegates, is mostly elected from three-member districts. But the voting method there (called block voting) guarantees that the three will be chosen by the same pluralities, so they are going to be ideological clones.) It doesn't have to be done that way. In foreign countries, Proportional Representation methods (PR) are used. A district elects several members, divided proportionally. If a district elects, say, five members, and 60% vote for one party and 40% for the other, then the first party will get three members and the second two. Both sides have representatives they can contact who will speak for them. In particular, it would be very easy for the Maryland House of Delegates to adopt a system called the Single Transferable Vote (STV), since three-member districts are already used, just like the system in Ireland. (Not all the districts in Ireland elect three members; some elect four or five. But for Maryland, the easiest way to go about it would be to use all three-member districts.) Actually, there is a better way than STV that has been proposed by voting methods expert Warren D. Smith, but it is untried, so it would be hard to get adopted: Reweighted Score Voting.


But whatever form of PR is used would give a better result than the way we do it now. The only problem is to convince the powers-that-be in the various legislatures. They got elected by the present SMP method, and many might lose their jobs if a new voting system were adopted.


(You might notice that I've been posting a lot more frequently recently. This is deliberate. Sometimes it's seemed that nobody has been reading this blog, so it's been hard to get the motivation up to post here. But lately, Google has provided stats, so I'm able to see how often people have come here. And it's been a lot more than I thought. Also, it's been said that you get more visits from frequent postings. Looking at July, I got about three hits per day; in August so far, it's been more than 8 — and it's only been the last few days I've really increased my posting frequency, so this is not the full story! That's why you'll be seeing me post a lot more from now on.)

Modified approval voting

benham said:

What do you think of Range with a 0,1,2 ballot (or Approval with a voter option
to "half-approve", or Approval-minus-Disapproval with a voter option to do
neither)?

This is fine in my estimation. My only objection to Approval Voting is that one cannot separate disapproval from neutral, and this solves the problem. Actually, I have in the past favored just such an option as benham's "Approval-minus-Disapproval with a voter option to do neither."

Calculations of Bayesian regret

Broken Ladder referred to the "millions" of simulated elections in Smith's calculations. But any type of calculation of this kind is going to depend on the algorithm that generates the data. And a different algorithm might generate a different set of data.

I do not mean to denigrate Smith's experiment. It performed a worthwhile function. But it is still an estimate of the validity of these different electoral systems, not a mathematical demonstration, and has to be considered in that category. Even so, I think the experiment really does show how much better range voting is than any of the alternative systems. But Smith himself would not (I imagine) claim those numbers would have perfect validity in all cases.

"Leaving me cold"

I originally wrote:

I've never been very happy with the way we vote here, using plurality or "first-past-the-post" voting, but most of the proposals I've seen for reform leave me cold.

Apparently "benham" thought I was referring to IRV ("instant runoff voting," usually referred to in older literature as "alternative vote") and asked what about that system leaves me cold. Actually, in a system such as Australia's, where there are really only two parties that count (actually three, but the Liberal and National parties are so closely aligned that they can be considered one party), I think it works rather well. If you like a minor party, you can give it your first preference, then save your second preference for a major-party candidate and actually make a difference to the voting result. I'm not so sure it works as well if you have a number of different parties which agree on some issues and disagree on others, but I've never seen a place other than Australia that uses it (it's been adopted in some cities in the USA, but so recently that the results of the experiment haven't been seen).

Actually, the systems that "leave me cold" meant approval voting, the Borda Count, and Condorcet voting, and there will probably be further posts of mine addressing these specifically.

What a surprise!

This blog has been around for a little over 2 years — my first postings were dated February 10, 2006. And up until this week, it's looked as if almost nobody ever looked at it — there were so few comments that I sometimes wondered whether anyone saw my words. I posted on a lot of controversial topics, from Iraq to gun control and intelligent design, yet got almost no response.

Then, almost as a throwaway, I made a post about a topic that seems almost too technical for most people to care: range voting. It was a digression from the usual topics of discussion here. And all of a sudden, to that one post I have gotten more responses than to the whole two years of previous posts, several times over! What a surprise!

Because there are so many comments that address a number of different aspects of this topic, I'm going to split up my discussion into several posts today, each responding to something different. But thank you all for coming in here to discuss it.

And now for a digression...

Taking a bit of a break from the 2008 election, I've just discovered a fascinating site. I've never been very happy with the way we vote here, using plurality or "first-past-the-post" voting, but most of the proposals I've seen for reform leave me cold. In particular, I'm rather hostile to the "approval voting" system, where you vote for all the candidates you can support, but with no way to distinguish between someone you really don't like and someone you're unfamiliar with. Now I've found just the system that fills the bill; it's called "range voting." You can rate each candidate, something like the way we rate books or CDs on Amazon, and the highest total rating wins. See the site here.