In San Francisco — which is one of America's most, well, experimental cities — they're trying a new twist on how voters choose their municipal officials. At a minimum, it's worth watching. It's called "ranked-choice voting," a less-than-useful descriptor for an interesting concept that has been kicking around in think tanks for some time.

It works like this: San Franciscans will vote by listing their first, second and third choices for the various offices (assuming there are three or more candidates in a given race). If any candidate gets a majority of first-place votes, game over. That person is elected.

But if he or she has only a plurality, then the vote counters will toss out the first-choice votes that went to the bottom-ranking candidates. They'll retabulate the whole race, this time counting those voters' second- and third-place choices.

Sound complicated? It is, somewhat. The procedure could require a few days to arrive at final outcomes. But it eliminates the hassle and cost of separate runoffs. And, if the idea were to catch on and expand to major races — perhaps, in the distant future, even to presidential contests — it would also eliminate the "spoiler" effect. In 2000, that would have meant that backers of Ralph Nader could send their message of dissatisfaction at the ballot box, but without any prospect of skewing the ultimate outcome.

To carry the Nader example further, in 2000, he didn't come in first or second or even close in any state. So, under ranked-choice voting, in any state (read, Florida) where neither George Bush nor Al Gore had tallied an initial majority, the votes for Nader would have been set aside. Those voters' preference would then be reallocated based on second choices. It is beyond reasonable doubt that such a procedure would have made Al Gore president of the United States. You may think that would have been a desirable outcome or a bad one — be my guest — but it certainly would have been different.

This wouldn't always change the outcome. In 1968, for instance, George Wallace captured 46 electoral votes on an independent ticket. But reliable after-the-fact polls indicated that four-fifths of Wallace voters would have gone for Richard Nixon if Wallace hadn't been on the ballot. Either way, advantage, Nixon. On the other hand, in 1948, a year with a powerful third-party showing, ranked-choice voting might well have ousted Harry Truman from the White House and installed Thomas Dewey. (Thomas who? See what I mean? Today, he's barely a footnote.)

For readers with long memories, this may call to mind the case of Lani Guinier, a law professor nominated in 1993 to head the Justice Department's Civil Rights Division. But it quickly emerged that she was an advocate of something called "cumulative voting." In the ensuing controversy, President Clinton withdrew her nomination quicker than you can say "hanging chad."

What was the issue? Under cumulative voting, let's say there are five county commissioner positions up for grabs. Each voter gets five votes and can distribute them as he likes — including piling all five on just one candidate. One outcome could be that if a minority community hung together in solidarity and massed their cumulative vote behind one candidate, they could all but guarantee election of a minority — a phenomenon that, in the past, perhaps had occurred rarely if ever.

Is that social justice? Again, choose up sides however you like, but it is considerably out of the mainstream — much further than the ranked-choice arrangement. Cumulative voting will strike many as doing violence to America's one-person-one-vote tradition. Some will call it a quota scheme. (But it's worth noting that it has been used for local elections in communities as diverse as Peoria, Ill., and Alamogordo, N.M.)

Ranked choice, by contrast, strikes me as a common-sense, relatively uncomplicated way of smoothing out a wrinkle in the fabric of democracy. It's the voting-booth equivalent of saying, "Sorry, we're out of strawberry. Would you like tangerine?" It eliminates "wasted" votes for no-hope candidates. It allows citizens to symbolically register unhappiness without damage to the final outcome. It eliminates plurality victories, the electoral equivalent of kissing your sibling.

Not all left-coast experiments strike me as worth paying attention to. This one does.