Possible editorial, title = "Voting systems - advantages and disadvantages" author = "Warren D. Smith" Note to editor: _italics_ *bold*. This is 7002 characters & 1177 words which is a little shorter than the original Richie editorial at 7335 characters and 1206 words. Note: this was published in the Takoma (MD) Voice, Monday 29 Oct 2005. For the graphical sidebar add-on material: http://math.temple.edu/~wds/crv/takomagraphic.html ===================================================================== Voting systems - advantages and disadvantages Rob Richie recently wrote an editorial in this newspaper advocating replacing the usual "plurality" voting system in Takoma Park, by "Instant Runoff Voting" (IRV). Richie is the founder of the Center for Voting and Democracy (CVD), a group advocating IRV. Although we believe IRV has advantages over plurality, considerably greater advantages can be achieved with greater simplicity and less cost by using a different system, _range voting_. We would both like to see Range implemented in a city such as Takoma Park, and for Richie and his organization to recognize Range's superiority and instead advocate it. *Plurality*: your vote is the name of a single candidate. The most-named candidate wins. *IRV:* your vote is a preference-ordering of the candidates, such as "Amy>Bob>Cal>Dee." The candidate top-ranked by the fewest voters is eliminated from the election and from all votes. (I.e. if Bob were eliminated, then this vote would become "Amy>Cal>Dee"; if Amy were eliminated it would become "Bob>Cal>Dee.") Then we do another round using the modified votes and reduced candidate set - and so on round by round, until only a single candidate remains - the winner. Richie also apparently is advocating permitting ballot "truncation" where voters only rank some but not all of the candidates, e.g. "Amy>Bob" or just "Amy" would also be permissible votes. With truncation, if your ranked candidates all lose, then your vote gets discarded and is unused in the remainder of the election process. *Range:* your vote is a score for each candidate in some fixed numerical range, for example 0-9. For example, you could vote Amy=9, Bob=9, Cal=4, Dee=0. We also recommend permitting voters to express "no opinion" about candidates by instead writing an X (intentional blank), e.g. Cal=X. The candidate with the highest average score wins. There are many reasons why range voting is superior. *Expressivity:* You can express not only who you prefer to whom, but also by _how much_. You can also rank candidates as equal. You can also express ignorance about a candidate ("X") to leave the scoring about them to other, hopefully better-informed voters. Or, you can give them 0. With plurality, you are forced to give the minimum possible amount of information in your vote, making it the worst possible voting system in this respect - while Range is more like the best possible. With IRV, if you want to say Hitler is the worst, Gandhi is the best, and you are ignorant about Snickel and Boodle, you simply are not permitted to say it. *Voting Machines:* Range voting can be handled by any voting machine in the USA, right now, with no modification required, because it can actually be treated as several artificial plurality-style elections. IRV cannot. In fact, although Richie claimed Takoma's Diebold voting machines could handle IRV after modifications which cost "extra," the only way that could be done is if either all the machines were connected together in a network, or if all the votes were _individually_ sent to a central tabulator: not just the vote _totals_ - the votes _themselves_. This can make election administration a lot more difficult and dangerous/hackable. For example, after San Francisco adopted IRV in the 2004 elections it immediately experienced massive problems with their ES&S machines and had to delay reporting all the non-obvious election results for weeks. The Utah Republican Party stopped using IRV at its party conventions in January 2004 because "after three years of doing it with delegates, there was still a lot of confusion" and counting votes took them hours longer than usual. (The quote is from Spencer Jenkins, executive director of the Utah GOP.) Although many US cities adopted IRV during the 1900s, almost all abandoned it. *Spoiled ballots:* With Plurality, an accidental "overvote" invalidates your ballot and it is discarded. With IRV, there are many more ways to spoil your ballot, for example by accidentally ranking two candidates the same. But with Range, _every_ way to award candidates a single-digit score is legal, and there is no such thing as an overvote. And even if you do make an illegible score, it can be treated as an "X" in which case most of your vote still remains operational and undiscarded. *Tie-crises:* With IRV, every single round gives an opportunity for a tie or near-tie Florida-style crisis, increasing the likelihood of crises and major delays. With Range, in contrast, the chance of a tie is _reduced_ versus either Plurality or (especially) IRV. *Better results:* Our computer simulation studies indicate range voting yields higher-quality winners, on average, than any other common voting system proposal, across a very wide variety of scenarios. *Simplicity & Familiarity:* Who has not been asked to rate things on a scale of 1 to 10? Everybody is familiar with the system and has seen it on TV in the Olympics. *Nader:* Richie went on at great length about how IRV solves the "Nader stigma" problem where voters who prefer Nader over Gore over Bush, by voting Nader, sadly cause _both_ Nader and Gore to lose. (Similarly, Buchanan votes caused Bush to lose some states to Gore.) Voters therefore are motivated to "dishonestly" vote Gore, thus hurting Nader far more than he deserves and artificially perpetuating the USA's 2-party domination stranglehold (of the top 600 US federal positions, not a single one is a member of a third party). But Richie's claim is only true sometimes: it is _still_ possible in IRV that your vote Nader>Gore>Bush will cause both Nader & Gore to lose, whereas voting Gore>Nader>Bush would cause Gore to win. (Two example elections of that kind are on the CRV web site.) So you can still be motivated to vote dishonestly. That's recognized by IRV voters, and hence the 2-party stranglehold and Nader effect can continue with IRV. This presumably is exactly why every IRV country, such as Australia, is 2-party dominated. Indeed, Richie _admitted_ that "[With IRV] minor-party candidates aren't likely to win office much more than under plurality rules." But with range voting, if you like someone the best, you _really can_ honestly give them a full 9. This in no way hurts Gore or Bush in their battle versus the other, because your decision about what scores to give Gore & Bush, are totally _independent_ decisions: you can give them 9 and 0 if you want. *Clones:* Finally, both IRV and Range totally solve the "cloning" problem where introducing a "clone" of some candidate into the race, causes them both to lose, even though either running alone would have won. With IRV or Range, candidates neither gain advantage nor lose it when clones enter (aside from the victory moving between the clones). To find out more about range voting, check the Center for Range Voting web site http://math.temple.edu/~wds/crv . The CRV was founded by PhD mathematician Warren D. Smith.