An election result like this ought to convince you that first-past-the-post is a lousy way to run an election. Never mind for the moment who won -- none of the candidates came anywhere near a majority.

Ideally, you want a system where voting for your favorite candidate can't help your least favorite candidate win. You want a way to handle races with more than two candidates so that the winner is the candidate who would have beaten each of the other candidates in a head-to-head election. You want a system where no candidate -- not a Ross Perot, nor a Gary Richardson -- can be a spoiler.

In this election, there is no way to know for sure if Martinson would have beaten Phillips or Harer or even Nichols in a head-to-head race.

Adding a two-candidate runoff gets you closer to the ideal system I described above, but with the top three so close, there is still the possibility of the order of finish varying had the minor candidates not been in the race. Between them Nichols, Harjo, Weaver, and Jackson received 601 votes, and there were only 63 votes separating 2nd and 3rd place, 74 separating 1st and 3rd. If only the top three had been in the race, where would those votes have gone? We can't know, but any two of the three might have wound up as the top two.

Louisiana has a system where all candidates run against each other, regardless of party, and if no one gets 50% of the vote, the top two candidates face off in a runoff. In 1991, with twelve candidates in the race, disgraced ex-Governor Edwin Edwards received 34%, former Ku Klux Klan leader David Duke received 32%, incumbent Governor Buddy Roemer received 27%, and the remaining nine candidates received 7%. The runoff was "between the crook and the Klansman," but we can't know what the initial result would have been if the nine minor candidates had not been in the race. Roemer may well have received enough of that 7% to put him in second place instead of Duke. There's little doubt that in a head-to-head race with either Duke or Edwards, Roemer would have won. There was a similar result in the 1990 Democrat primary for Governor in Oklahoma (David Walters, Mike Turpin, Steve Lewis) -- the top three clustered together, and enough minor candidate votes that any two of the top three might have made the runoff if the minor candidates had been eliminated.

What voting system eliminates those sorts of anomalies? Instant runoff voting (IRV) does. Under IRV, voting is simple. Voters rank the candidates in order: I mark a 1 next to my favorite, then mark a 2 next to the name of the candidate who would be the my choice if my favorite weren't in the race, and so on down the list.

It's called instant runoff voting because it's equivalent to having a series of runoff elections, eliminating the low vote-getter each pass and choosing among the remaining candidates. The advantage of IRV over a series of runoff elections is that you only have to open the polls once. IRV is used to elect the President of Ireland, members of Parliament in Australia, and here in Tulsa it was used at the 1st District Republican Conventions of 2000 and 2004 to elect delegates and alternates to the Republican National Convention. I first experienced IRV in college -- we used it in our fraternity to elect officers.

At the very least, Tulsa needs a runoff in special elections, but it would be better still to use IRV in all elections. As a charter city, Tulsa could choose to do that.