CHAPTER 6
INSTANT RUNOFF VOTING: A PERSONAL DISSENT
The following items are part of an exchange I had on instant runoff voting with Peter Schermerhorn, a wonderful interlocutor and Green Parry activist much concerned with improving this country. In addition to outlining my views on IRV, they also give some indication of what my efforts to promote voting reform have involved.
An Initial Reaction
March 23, 2004
Although I appreciate being notified about the effort to initiate an Instant Runoff Voting campaign for Ann Arbor, I CANNOT ENDORSE INSTANT RUNOFF VOTING FOR ANN ARBOR. I cannot endorse such a system because I favor a simpler, but elegant system which also eliminates the spoiler effect and wasted vote syndrome--APPROVAL VOTING. Under such a system, each voter in a multi-candidate (three of more candidates) election can give one vote each to the candidate or candidates he/she approves of with the candidate having the most votes winning. The advantage of approval voting over IRV, or what used to be know as preferential voting, is set out in the attached excerpt from the book "Approval Voting" by Professor Steven J. Brams of New York University and Peter C. Fishburn (Birkhauser, Boston 1982, pp. 6-8). In support of the points which Brams and Fishburn make, I would emphasize that approval voting has the advantage of being simpler, being more transparent and being less subject to voting manipulation that IRV. I would urge those interested in improving our voting system to consider initiating a campaign to introduce approval voting into our elections and would be interested in an exchange on this with anyone interested.
John Howard Wilhelm
A Follow-up Response
March 26, 2004
Pete Schermerhorn, I greatly appreciate your response and comments on my note to you, with copies to others, on approval voting. And I would like to take advantage of this to clarify some of my thinking as well as to see if it might be possible to engage you and some others further on this very important issue.
It surely is clear as you point out that the current system of plurality voting is highly unsatisfactory. The issue is what is a better replacement? Based on my reading and thinking over the years, I am convinced that approval voting is a significantly more desirable alternative than preferential voting or what is now called IRV, instant runoff voting.
When I copied the excerpt from Brams and Fishburn's book on approval voting, I realized that with advances in computers and programs since
1982 that running an IRV election would probably not be too difficult or even very expensive. But such a thing does entail more complexity which requires programming and, given Murphy's law, even well designed programs do fail from time to time. Such a failure it seems to me, in light of what I know of the 1976 Ann Arbor experience with getting results in an IRV election, could have bad consequences in terms of maintaining, if you will, a stable, desirable alternative to plurality voting. I do not believe that approval voting entails any such problem since it only involves additional counting.
Another problem I have with IRV is the issue of transparency. It seems to me that a desirable system of election must be, in terms of how its results are arrived at, easily understandable to voters. From my reading over the years, especially in the Economist, and from personal conversations, I am convinced that the results of preferential voting are frequently not understood by voters. Donald Saari has argued that a weighted system of election, the Borda count or a modified Borda count, is the most desirable alternative. But the problem I have had with this is that clearly the issue of transparency becomes even more difficult under such a system.
I do not believe that the reservations Brahms and Fishburn expressed in their discussion comparing approval voting with preferential voting that I excerpted are not particularly probable problems. For some reason, I do remember the 1970 New York senate race. From what I recall, I believe that they are right in saying that that election under a preferential voting system would have eliminated Goodell who was quite probably the candidate most acceptable to voters despite his third place under the system of plurality voting that decided the election. In a close election like what I understood occurred in Ann Arbor in 1976, this could be a problem and could contribute, as it seems to have in Ann Arbor, to such a system being abandoned.
It is my belief that approval voting does not suffer from such a problem. From what Brams and Fishburn have written, it is my impression that in most real world cases that approval voting results in the Condorcet candidate, the candidate who would win in all bilateral elections with all the other candidates, wining.
In the situation where that is not the case, I do believe that approval voting does result in a very strong second best outcome in that the candidate approved by most voters wins. Clearly from the 1970 New York example, IRV does not guarantee such an outcome.
It seems to me that when one considers, as Brams and Fishburn showed in another section of their book, that a candidate under IRV who would otherwise have won can wind up losing should voter sentiment shift in his or her favor, that one encounters an additional serious problem with IRV. Under either IRV or approval voting a candidate could clearly try to improve his or her position by encouraging voters not to vote for a specific alternative. But under IRV it seems to me that there is an additional problem and that is that candidates could improve their position by encouraging their voters to switch their preferences for other candidates to affect the order in which other candidates are dropped and thus, particularly in close elections, manipulate the results. I find myself uncomfortable with this since I know fully well that with modern polling techniques and computer programs such manipulation will occur. But I cannot see how it could occur under approval voting which leads me to believe that this systems will allow voters to better express their preferences as well as avoid the wasted vote syndrome and spoiler role.
From my perspective, approval voting has an addition advantage that I would not emphasize in arguing for its adoption but would like to see used later on. I came across the idea from Brams and Fishburn's 1982 book at a time when I was thinking about the issue of reform in the former Soviet Union. In the summer of 1991 in Moscow I gave a talk on political and economic reform in Russian to a group within the Soviet Academy of Sciences. In that talk, I advocated electing a constituent assembly to sort out the political system of the country and advocated using approval voting as a means of limiting the participation of members of the CPSU in such an assembly. I advocated limiting the number of such members to the proportion that the Bolsheviks had in the 1918 Constituent Assembly that Lenin forcibly closed down. Under approval voting one could easily rank candidates according to the percent their vote total represent to the number of voters voting in each district and use this ranking as a basis for a cutoff that would disqualify the winning candidates who would exceed such a limit on CPSU member participation in such an assembly. I thought then, and believe even more so in retrospect given what happened in Russia since, that such a limitation was a necessary ingredient of successful reform in that country.
It is my belief that the American legal profession is deeply corrupt and corrupting in our society. You just cannot in my opinion have legislatures exercise the oversight of that profession that is needed with the proportion of lawyers that we have in them. I think that eventually a constitution amendment ought to be proposed to mandate approval voting in a similar system designed to limit the number of lawyers in our law making bodies.
There are a number of other issues which I believe ought to be looked at in examining how we can reform our voting system so that it better serves our people. I have tried to get a group together to examine these issues with the idea of coming up with a sensible program of reform. I would very much like to engage you and others in trying to push such an effort and would very much like to get together to discuss this in more detail.
A Useful Exchange
Pete, I want to thank you for your last message, truncated below, to which I would like to respond with some additional comments following the excerpts I have set out.
John Howard Wilhelm
Date: Mon, 29 Mar 2004 12:26:48 -0500
From: Pete <treasurer.a2irv.org@loucks.org>
To: John Wilhelm
Subject: Re: IRV and Approval Voting.
>John, thanks for your reply. I admit that I was confused the first
>couple of times I heard about Approval Voting (from you), and the
>supporting documentation you had for it was long, so in this busy
>activist's life, it did not get read thoroughly.
> Some thought now come to me that I'd like to share.
>Another thing that occurred to me is that Approval Voting asks voters
>to vote only for the candidates that they would approve holding the
>office. If this is honestly done, then great, we have a true sense
>of what the people want. But with Approval Voting, aren't you really
>just eliminating the people that you -don't- want to see in office?
>Doesn't that lead, always, to middle of the road, non-controversial,
>probably bland winners? i.e., the ones that no one actually doesn't
>approve of? I think that's what happened with our vote (and I should
>say I was one of the 'winners'). . .
Pete, On bland winners, there are two comments I want to make. First,
had we had approval voting during the 1992 presidential election, I
would maintain that Ross Perot, not a very bland candidate, would have
won. Second, in most cases Brams and Fishburn in their book and
articles have shown that under approval voting the Condorcet
candidate, the candidate who would win on each one to one contest
among a plethora of candidates, is most likely to win. When that does
not occur under approval voting you would still get a winner approved
by most voters, which surely is not a bad second best. As we well
know this does not occur under pluralistic voting and the candidate
least representative of the voters, a Senator Buckley, can wind up
winning. But as Brams and Fisburn's 1970 New York example
illustrates, while a Buckley may not have won, a Goodell, most likely
the Condorcet candidate, would have lost under IRV in that race
because he would have been eliminated after the first indecisive
results. This plus the fact, as has been well illustrated by Brams,
Fishburn and others, that under IRV a candidate who might have
initially won can wind up losing if more voters move that candidate
into first place in terms of their preferences, makes to my mind IRV
decidedly inferior to approval voting as an alternative to the present
system.
>Other comments below:
>Since you mentioned plurality, I should point out that both the
>current voting system and Approval Voting rely on pluralities, not
>majorities. While it is confusing to some folks how single
>transferable voting works, the one thing that they can agree with is
>that candidates ought not to serve unless elected by a majority of
>the people. . .
Pete, I think this is a very good point and one that I have been
concerned about as well. There is, it seems to me, a very simple
solution and that is for an approval voting candidate to be elected
he/she should have at least 50% of the approval votes cast by voters.
If in a race like the 1970 New York senatorial election you had an
approval vote split of something like 36%, 32%, 32% to take an
example, no candidate should be elected and the law should require a
new election with a different set of candidates. Whether one should
make this part of an initial approval voting election reform or not,
is an issue that I am undecided on which is one reason why I have
tried to get together a group of people to work on thinking about a
suitable election reform.
>. . The mess in Ann Arbor was actually due (according to our
>research and the memories of those of us here at the time) to
>incomplete and improper preparation, a lack of understanding on the
>part of the clerk of what needed to be done, open rebellion of the
>elections employees, and possible malfeasance on the part of staffers
>who did not want to see such a system.
Pete, while you are probably right about your description here I think
you forget another factor. It is my impression that IRV was put in
place in Ann Arbor to avoid the spoiler role of the Human Rights
Party. When it was abandoned to my knowledge, the Human Rights Party
had faded as a spoiler party for the Democrats and there was no longer
a need for them to support IRV.
>. . This time around we still expect such resistance, but we mean
>to do a better job of preparing the public not only before the
>election, but breaking down the results of the election afterwards to
>show how votes were transferred in different rounds, so that folks
>can have maximum transparency. If the AA News won't print it, we'll
>work it up and get it to the citizens anyway. . .
Pete, While such maximum transparency may be desirable, I do not see,
particularly in light of the example of the 1970 New York senatorial
election, how even intelligent voters can really discern what the
outcome really represents in terms of reflection voter preferences.
In the ranking, for instance, are people only ranking the order of
candidates they really want or are they also ranking candidates
according to whom they find least undesirable? If so, in terms of the
results you have a very mixed bag and, other than public opinion
polls, no way of knowing what the results signify. I would submit
that my modified approval voting system suggested above does not
suffer from this problem and on this score, an understandable
transparency of the results, is logically superior to IRV.
>. . I used to grouse at the number of lawyers in a host of
>professions, including sports agents and landlords as well as
>lawmakers. But then I realized - in lawmaking, one needs to know the
>law. So I find it natural that lawyers are our lawmakers - given how
>complicated the legal structures are. . .
Pete, on my idea of using approval voting to limit the number of
lawyers in legislative bodies. I think it is important to keep in
mind that legislative bodies are the instruments that provide
oversight over the legal profession in our country. Given the
statement by Clinton, a lawyer by training, about what is is, I think
that we have a clear example of how corrupt and corrupting that
profession is. While we do need to have professional legal input into
the process of legislation, we are simply not going to have the needed
oversight of the legal profession if our legislative bodies are
dominated by lawyers.
>. . There are already many groups working for different kinds of
>reform, some of which I sit in on. There's Michigan Campaign Finance
>Reform, Election Reform MI, Center for Voting and Democracy (which
>you might know is very pro-IRV), Richard Winger's Ballot Access News,
>and many others. To my knowledge there is no single group trying to
>pull all these reform groups together into a cohesive program of
>change. I know the Greens advocate many reforms such as same-day
>voter registration, abolition of the motor-voter laws which
>disfranchise students from voting in the communities in which they
>live, election day holidays, full public funding of campaigns and
>limits to campaign duration, ballot access reform, abolition of laws
>which treat parties other than the Democrats or Republicans
>differently (equal protection clause and all), and a host of others.
>I'd be more than interested in seeing such a group form and have an
>aggressive, doable agenda.
Pete, On this last comment, I would like to say amen. We very much
need to have a cohesive program of change. I have had some ideas, but
I have no confidence of coming up with something that is both doable
and workable without having a group of people discussing and looking
over ideas. For a number of years now I have tried to engage members
of the former Reform Party and the Greens in a dialog about putting a
good program out in the public forum which was one reason I tried to
engage you and others at the Art Fair on the issue of approval voting.
Some random thoughts. I think we do need to have a national identity
card for everyone over 18 years. Such a card could easily permit
every citizen free access to federal elections by simply turning up on
election day at voting places or arranging for absentee ballots. On
campaign finance reform. I have been very appalled by the
legislation, ideas and discussions of this issue. I think that our
campaign finance reform efforts have gotten bogged down into what I
call the "socialist syndrome". That is, if something is thought to be
desirable and we try it and it does not work well, we just need to try
harder to implement it. I have a lot of impatiences with those who
say we should get money out of politics. I think that our campaign
finance reforms to date have simply made matters much worse. Nobody,
other than a Mitch McConnell, has spoken up about this, and I do not
see anybody, even among those pushing for campaign finance reform,
really dealing with the intellectual discussion needed to understand
this issue and come up with a good system of campaign finance that
will serve us well. On an irreverent thought. You cannot really
develop viable third parties without soft money. So if you ban soft
money, in addition to undermining party discipline, you also lock in
the election hold of the major parties.
Pete, I would like to follow up on this with you and may try to give
you a call about doing so. Thanks for all your thoughts and attention.
Regards.
John Howard Wilhelm