Another three-way governor's race that will end with a sub-majority winner has fueled calls for Minnesota to adopt ranked-choice voting.

In 2009, Minneapolis became the first Minnesota city to use ranked-choice voting in about 50 years, and its experience offers some cautions.

The allure of the idea is clear: It has been 16 years and four elections since a Minnesota governor gained office with more than 50 percent of the vote. Ranked-choice voting is a way to get a winner with majority backing, even in a three-way race. Proponents say it allows people to vote for their favorite candidate without fear of helping to elect their least favorite.

Under the system, a voter lists first, second and third choices, with lower choices coming into play only if that voter's first pick is counted out.

But that comes at a cost. Ranking candidates statewide would mean counting votes by hand, as Minneapolis did last year, or spending millions of dollars for machines capable of doing the count automatically.

More would be spent to educate voters. Minneapolis spent 30 percent more than normal in an election year to launch ranked-choice voting, chiefly to count votes.

Ranking candidates in Minneapolis failed to accomplish some of the things advocates promised. It didn't raise participation in the absence of a close mayoral race. Indeed, fewer voters turned out. It didn't swell the number of candidates -- 94 filed for municipal offices in 2005, the last year those elections were held, and 95 did so last year. The same number of racial minority candidates ran for office as in 2005.

Also, ranked-choice voting did not change the order of candidates in any single-seat race. Those who got the most first-choice votes wound up finishing first in the three single-seat races in which second choices were tallied to get a winner with majority support.

Four of every 10 voters said in a post-election survey that they didn't bother to vote for more than one candidate, the whole point of the new system. Some didn't find more than one candidate acceptable; others didn't know enough about the field to award a second or third choice, they told pollsters.

But some say it may be too soon to draw conclusions.

"I'm not sure based on one election what we really learned about it," said David Schultz, who teaches election law at Hamline and University of Minnesota law schools and analyzed Minneapolis' use of ranked-choice voting. "There may be a lot more being promised here about what ranked-choice voting can do than what it can deliver."

Some goals realized

Minneapolis is the eighth U.S. city to try ranked-choice voting in recent times. Most are liberal enclaves where ranking candidates isn't likely to reshuffle results appreciably, said Schultz, who served briefly on the board of FairVote Minnesota, which advocates ranked-choice voting.

The group has had some successes. Minneapolitans approved the ranking of candidates in a 2006 referendum, with 65 percent supporting the charter amendment making the switch. According to a St. Cloud State University post-election survey, an equal percentage of those who said they voted in the 2009 election thought the city should continue to use ranked-choice voting, with most respondents finding it simple to vote.

The same day that Minneapolis voters inaugurated the new system, St. Paul voters approved a switch to ranked-choice balloting for city seats, but the charter amendment got a much slimmer margin of support, with 52.5 percent backing. There also have been discussions of switching to ranking candidates in Duluth, Red Wing and Hopkins, but none has a formal charter proposal pending. Hopkins was the last Minnesota city to use ranked-choice before Minneapolis, dropping it in 1959.

Ranked-choice voting did accomplish some goals. Genuinely undecided voters got to support more than one candidate. Voters generally gave the system high marks in the St. Cloud State poll. There were no indications of mass confusion after more than $100,000 was spent to prep voters on the new method. Even the hand-counting of ballots, which delayed some official results for up to 17 days, happened faster than anticipated.

Hand-counting was needed because no equipment was certified nationally and in Minnesota to tabulate rankings of candidates. One manufacturer has since gained federal certification for equipment that it bills as able to handle the voting method.

Significantly more expensive

The cost of paying workers and renting space to sort ballots by first, second and third choices was a major reason that ranked-choice voting cost $345,727 more in Minneapolis than the city's traditional 2005 election. That's despite the city holding both primary and general elections in 2005, compared to 2009's all-in-one election.

The higher costs for ranked-choice voting are likely to continue regardless of whether machines are available to count the vote. The city expects to repeat most of its voter education spending in 2013 after several intervening non-city elections in which traditional voting will be used. Interim election director Virginia Gelms estimated the recurring extra cost for ranked-choice voting at $108,692, even with automated tabulation. That doesn't count the cost of new machines.

Schultz also called attention to a higher number of voter errors in 2009 than in 2005, some of which vote-counting machines caught and gave a voter a chance to correct. He called the rate of errors by voters and spoiled ballots "potentially troublesome" and significant enough to affect future outcomes.

Hennepin County expects to begin discussions with cities and other counties next year on replacing its vote-tabulation system, an expenditure that could top $3 million. Election manager Rachel Smith said the county will research systems that can accommodate ranked-choice voting.

The St. Cloud survey found some differences among voters in their use of the rankings for candidates. The youngest and lowest-income voters and members of racial minority groups were most likely to use only one of their three choices.

How many candidates were ranked also varied widely by precinct, according to a Star Tribune analysis of returns from the 2009 mayoral contest. The range was from 19 percent using all three choices in one precinct to 92 percent in another.

Even in one of the most competitive elections -- a five-way contest for the Nokomis-area park commissioner -- only a third of voters ranked three candidates.

Steve Brandt • 612-673-4438