Vicki Walker, Democratic Candidate for Secretary of State sent a response to a member of FairVote Oregon clarifying her research on advanced election methods such as Instant Runoff Voting. It appears that she's just leapfrogged the intelligence of Brown and Metsger on the IRV issue, and she brings up a number of important points to consider about the issue. Here, I will directly address her response: I support many election reforms -- campaign finance reform, fairness in initiative reform, greater accountability in auditing the vote counts and final election results, fusion voting and some other system of voting that makes Oregonians feel as if their votes count -- as evidenced by my endorsement today from the Voter Rights Coalition and the Independent Party of Oregon. Vicki's certainly leading the pack in reform organization endorsements. I spent a lot of time reviewing IRV because that seems to be the favorite option among many Oregonians. My first question about IRV is whether it would interfere with fusion. That question has been answered satisfactorily that both methods can work in tandem. Correct. One is a method of labeling candidates, the other is an actual counting method. Wholly distinct. The next concern was with regard to the Oregon statutory authority. It is clear there are at least four provisions of the Oregon constitution that would apply to the use of IRV in local elections, and a home rule county or city would only need to adopt a change to their charter to facilitate IRV. However, questions have arisen as to whether state statute is necessary to implement IRV. There was a bill in the 2005 legislative session, HB 2638, that would have provided the enabling legislation but it did not even get a hearing and died in committee. There may have been a similar bill in 2007; I do not recall. I suspect the failure to move forward on this issue is that the current Election's Administrator, John Lindback, does not support preference voting, believing it gives some voters more votes than others. His other objection that our election equipment would not facilitate IRV is no longer valid because voting machinery and software is coming on line now that would allow for such voting methods. I agree that John Lindback appears to be the main anti-preference advocate. In my opinion, he's abusing his position to put forward incoherent objections to preference voting in the name of protecting his party from legitimate and fair competition, but I'm going to reserve calling that a fact until I can talk to him personally about the mathematical mistakes he's making. First, it makes no sense that some voters get "more votes" under an IRV election. Mathematically, every "republican form of government" (as opposed to, for example, direct democracy) voting system cannot accurately represent the electorate because the number of representatives is always fewer than the number of people it's intended to represent. Technically, the closest way to approach accurate representation is to select representatives out of a completely random sample, but historically, that's difficult to audit because one man's random is another man's conspiracy. Even though we have mathematical methods to analyze randomness, such a determination wouldn't be able to build confidence among the represented body because the processes are most likely above people's heads, and if jury selection is an example of how we currently attempt random selection of a jury of peers, I'm not sure it'd be a whole lot better than what we have now. The other problem with random selection is that making coherent laws is a specialty that requires some expertise. The entire system would require changes where new representatives are educated as they come in within a framework. We'd have to essentially have something akin to judges that oversee the process, and that also might introduce issues. In theory if these hurdles can be overcome, it's my preferred method of selecting representatives, though. Arrow's theorem shows us that given some arbitrary metrics of fairness, no election system can provably satisfy all of them due to inherent contradictions in the ideas of fairness that he chose to include. Many election reformers and researchers have been attempting to resolve the fairness questions by changing his ideas of fairness to meet what they consider important or not and then picking the "best" system from their criteria. I think the way around the issues of Arrow's theorem and related theorems is to understand that government is composed of checks and balances. Since you can't satisfy everybody all the time, you should at least try to satisfy everybody at least some of the time. What I mean by that is we should use different election methods for different purposes and ideally set them off as checks and balances against the biases of each other. With that philosophical foundation, selecting election methods becomes a matter of understanding the point of each branch of government or elected office within those branches and designing a system that works to maximize the value of each type of election method. But now, back to IRV and why it doesn't do what John Lindback thinks, according to Walker, "It gives some voters more votes than others." Fillard Rhyne gave a mathematically sound response to this complaint: after each round of voting and the elimination of candidates as per its criteria (which is the source of its "bias", which I will get into later), each vote is counted. The only way a voter can lose their vote in a new round is to not fully specify their preferences. That's a decision the voter makes akin to not even voting in the first place, so I don't see that as an issue. The important characteristic of IRV is that, if voters fully-specify their preferences, the candidate selected by IRV is ensured to have majority support as the highest ranked candidate among those not already eliminated due to not having enough votes in the previous rounds. The major complaint against IRV that I'm aware of is that by using the highest ranked candidates in previous rounds, it's ignoring "general-but-not-highest" support of candidates who may be more favorable in average, but not the most favoriate of a substantial number of people. This introduces what's called "non-monoticty" (one of Arrow's criteria) -- that ranking a candidate higher may actually hurt the candidate due to a secondary effect that it may cause votes that would have gone to them to go to another candidate because they still aren't high enough to prevent the candidate from being eliminated and the downranking of the higher up candidate may lead to the election (or near election) of a candidate higher up on their ballot. This is where the bias of IRV comes from: how candidates are eliminated. It eliminates those who show the the least amount of "strong" support after each round, so, naturally, it elects those who have the strongest support of the majority (or near majority in the case of not-fully-specified ballots) of voters. In the extreme, between a candidate with 100% approval but not strong approval versus a candidate with 51% approval but in all cases stronger than the the other candidate, the 51% approved candidate will win. It's important to note that compared to normal plurality voting (where only one round is used), IRV is in all cases superior. Plurality voting in fact has a very similar bias: in the situation above, the 51% approved candidate would win anyways! IRV's weaknesses compared to single-round (or technicaly "fewer-round") plurality are thus from being too similar to plurality voting in the first place! That's why such objections without a coherent replacement are merely advocates for keeping the even worse plurality system. One valid reason for liking plurality voting over IRV voting, other than the simplicity arguments, is that when given multiple candidates, the two highest candidates end up facing off against each other with a stress on cooperating and eliminating the other candidates for the risk of spoiling. They thus create a duopoly, when the process is repeated in subsequent elections (iterated, in math terms). If you're a fan of the two-party system, then, plurality voting seems awefully favorable. That's exactly why we need to get rid of it: because the two-party system is not an accurate representation of the electorate. Two-party systems when themselves iterated (remember, repeated in successive elections), create positions where each candidate of the two major parties only wants to differentiate themselves on a few single issues and otherwise be as non-descriptive of their platform as possible. If a candidates can successfully frame the debate betwen a set of gerrymandered litmus tests (for example, abortion rights, as happened recently, easily splits the country in two along rural/urban lines), the two parties can effectively gerrymander the electoral process to eliminate third parties. Ironically, opposition to the occupation of Iraq, long a staple of third-party politics, is becoming a litmus between Democrats and Republicans in many parts of the country. That's simply not an acceptable way to run government. While I can personally disagree with occupation and war in the middle east and be pro-choice, it's frustrating that we can only address one issue at a time, and those issues are merely one part of a myriad list of issues that need to be addressed that run the gamut from Universal Health Care to Voter-owned Elections. Finally, IRV is just one form of preferential voting. As I have learned, there are numerous variations: Borda, Bucklin, Condorcet, Range Voting, Runoff Voting, Exhaustive Ballot, to name a few. In addition, the mathematical theories as to why one system is better than the other are so complex they make my head spin. In my research, I've determined that the Preference systems with the most mathematically positive properties are IRV (and related STV), Condercet (and the related and much easier to count Range Voting). One of the other methods can usually be reduced to having similar properties of those classes. I tend to group them in two ways: those methods ideal for Proportional Representation (non-monotonic) and those methods ideal for General Representation (which usually happen to be monotonic). In my platform, using my philosophy that we could simply use both categories of preference voting depending on the desired type of candidates we seek, I outline how we can use Condercet, IRV, and STV together in a very well-balanced electoral system, where we use IRV for issue-based seats like statewide offices such as Attorney General or Secretary of State (and local commissioner seats where they oversee specific parts of government), Condercet for Governor (and executive-like seats like mayor), and STV for representative bodies, where we use multiple seats to ensure some form of proportional representation. State-wide, I also proposed keeping some form of regionalism by breaking up the multi-seat House of Representative elections into large super-regions to ensure bioregional interests are still intact). The bottom line is that Oregon should explore using any one of these methods or a combination thereof to make its election system work in a way that every Oregonian feels their vote is counted. IRV seems to be the most popular in the U.S. -- San Francisco has been using it in nonpartisan elections since 2003, and many other cities have joined in, and several states have enacted legislation that will eventually allow IRV. As you know, Australia started using IRV in 1893 and continues using it today. Yes, Oregon should explore and begin implementing preference voting in honor of the centennial of voters wanting it in the first place. My record in the legislature has been strong, consistent and unwavering on giving every Oregonian a voice and a seat at the table. I am open to supporting any efforts to improve upon the plurality voting system we have now. If the legislature is not willing to have public hearings on this issue, then as your next Secretary of State I will commit to appointing a citizen commission to begin the public discourse. That's a good first step, but as Secretary of State, you'd have the ability to not only begin the public discourse, but to allow it to be implemented at the local level without even enabling legislation. The entire "enabling legislation" idea is a ruse created by Lindback, probably at the behest of Bradbury. The fact is that the state statute that requires a plurality to be elected is entirely consistent with IRV. IRV just adds an additional requirement that rounds be done to make sure that the plurality becomes a majority. It's not mutually-exclusive. It's merely a failure to understand the math behind IRV that could lead one to think that plurality excludes IRV as an election counting method. Here's some links I discovered in my research that you might find useful. I particularly appreciated the analysis by FairVote as to why the 2001 Eugene ballot measure failed. http://www.fairvote.org/?page=1492 http://rangevoting.org/ I've helped publicize range voting in the past, and consider it a viable system for selecting for selecting a governor or executive among multiple candidates, although it's not without its strategic gaming possibilities (most people will just choose all or nothing) and it's incompatible with none-of-the-above directly (although one could say candidates that couldn't get greater than half the range could be indirectly interpreted as "none of the above". Bayesian regret, though, appears to be low. I'm leaning toward Condercet-style elections, personally. http://www.votefair.org/compare.html I'll note that the votefair person (a local of Portland, OR) makes some fundamental arguments that mistake IRV. IRV can support equally-ranked candidates. The votes are just split in half and counted thusly. In reality though candidates are truly never equal, so this is mostly a non-issue if a slight improvement is made to typical IRV counting methods. Secondly, the votefair person's arguments against proportional representation are specious and illogical. PR works in countries without prime ministers and with presidents. It can also be used in single-seat elections when you have many seats in the form of IRV, but it requires specific "tasks" for each office to be distinct and representative of issues in government (environmental officer, elections officer, attorney general, etc.). It also criticizes party-ballot-line PR and blasts STV (the form of IRV they mention). It seems they fail to understand (as published in a recent paper on the issue of non-monotonicity and PR), that STV's non-monoticity "unfairness" is precicesly what makes it create proportionality. If monotonicity were required, the majority could take all the seats, leaving no minority representation, regardless of the strength of the minority's passion. the "votefair" method is unable to do PR by design. Furthermore votefair appears to be just a slightly improved variation of Condercet (vulnerabile to Arrow's Theorem as well) with equal ranking allowed and with a tie-breaking method. All my references to Condercet voting would thus apply to votefair, except for the above modifications. It might be usable for a Governor race or other executive, but because it is terribly bad at allocating proportional representation, it is entirely unfair for multi-seat elections and wouldn't be desirable for issue-based and task-based single-seats. Be wary of persons promoting a silver bullet for all elections: they don't know what they are talking about. He also doesn't acknowledge his debt to Condercet voting adequately, and his comparison table leaves out a number of additional criteria one may consider more important than his. http://www.blueoregon.com/2005/03/instant_runoff_.html (this was a very comprehensive discussion) http://www.instantrunoff.com/ Again, thank you for your inquiry. And please understand that while I support a wide variety of election reforms, my top priority upon my election will be directed at campaign finance reform, because I believe all roads lead back to the fundamental problem of big money pervading the functioning of our democracy. Here, I definitely agree with Vicki Walker. THE major stumbling block to implementing key election reforms and bringing government auditability back into the public discourse is big money and a lack of campaign finance reform. I think Metsger's position on opening up the airwaves back to the public is the most key federal action we can do (FCC being a federal body), but locally, putting in place already-voted-upon-and-passed campaign finance regulation is the first step of any truly progressive Secretary of State.