New Voting System

I think that if the voting system for elections were done differently, we wouldn't need parties and primary elections in the US, and we'd end up with better results. Suppose we ask people to rank the candidates rather than just mark their top choice. This provides a lot more information, but what would you do with it. It's still not a simple process to find the best choice, even if everyone supplied their rankings, but I think pretty much any reasonable system that could be devised would be better than the current system.

Let's consider what is currently done. The votes are just added up, which is like taking all of the ranking that everyone (hypothetically) filled out on their ballots and just adding up the first place rankings, ignoring all the rest. One known problem is that if two candidates are similar they might split the supporting votes between them causing them both to lose. This is part of the justification for primary elections among parties. But primaries have their own problems. For one thing, they have the same problem, meaning two candidates of the same party can split votes in the primaries causing a third to win the nomination. But even more importantly, the moderate candidates that capture a lot of the popularity in the middle (and probably should be the ones elected) might have a hard time winning their party, so they don't even get a ticket to the dance.

OK. Back to the question of "what would you do?" You have all of these rankings from each voter. Now what? You could do something like giving more points for 1st, less for 2nd, etc. on each ballot.

I'll tell you the simple idea I came up with in a minute, but first consider this example. There are four candidates: A, B, C, and D. A gets 40% 1st place votes, 30% 3rd, and 30% 4th place. B gets 30% 1st, 35% 3rd, and 35% 4th. C gets the same as B. D gets all 2nd place votes. So who should win? It seems clear that A beats B and C, although I think that's a little too hasty. You might have to get into the individual votes rather than the aggregates to actually make that judgment. Anyway, the important question is, what about D? In our current voting system, he doesn't get any votes. But I would argue that D should win. He beats any other candidate in a head-to-head competition. He is the candidate that makes the most people happy. 60% of the people like D better than A. 70% like D better than B, and 70% like D better than C.

So here is what I would propose. Essentially you would do head-to-head competitions because those are easy and unambiguous when everybody reports their rankings, and you would start eliminating people based on that. Now, in the case above, no matter how you do it D wins, but that's not always the case. We've all seen round-robin tournament where it's hard to pick a winner because you get a loop of people beating each other. I think that could happen here, so the order of comparisons becomes important. You'd have to set up a bracket like a tournament. Your seat in the bracket would be based on some other simple measurement of the votes. It could even simply be based on the current way we count votes (counting 1st place votes only), or something slightly better. I don't think the seating is the most crucial part because hopefully the outcome won't be so sensitive to the order. If the outcome is so close that the particular choice of seating algorithm makes a difference then it sounds like you've got several good candidates to choose from anyway, so your going to end up with one that a lot of people like either way you go.

From a practical point of view, you wouldn't have to have everybody rank all the candidates since they probably don't know most of them. They could just choose to rank the ones they care about, and the rest all fall in the lowest rank by default.

Notice that you're not splitting any votes if two similar good choices are both running against each other and others. If you list them as 1st and 2nd then it's still counted that you favor both of them over the other candidates. This way you don't need primary elections with arbitrarily drawn party lines and bickering back and forth between them. The other justification I hear about for primary elections is to pool all the campaign money together, but that doesn't seem to be the main reason. That's a challenge that someone else will have to work out.

P.S. You could actually gather more information during the voting. You could have each person give a score to each candidate, that way you know how strong their preference is for each of them. From this you could of course derive the rankings that I was suggesting, and from that you could of course count 1st place votes only the way we currently do. But I don't think this level of detail is necessary, and it would certainly make voting a lot more involved. Then again, maybe people would like to feel more involved.