'Ranked Pairs' can be explained to a person of average intelligence in a minute or two.
By
kybernetikos
on April 27, 2007
I shall back up this claim by attempting to explain it. I ask that you pretend to be a person of average intelligence for the purposes of this discussion...
The simplest kind of election is where only two people stand against each other, and the one who wins a majority of the votes wins. When choosing between more than two people, it's not practical to have a separate election for each possible pairing of candidates. However, we can work out what the result would be in each case by asking the voter to rank their preferences rather than vote for a single person. Once we have every voters preferences we can work out easily what the result would be if any one candidate would be compared directly with any other candidate, so if Bob, Alf and Jimminy were standing, it's easy to work out from the preferences who would win if Bob were to just stand against Alf, if Alf were to just stand against Jimminy, or if Bob were to stand against Jimminy.
If you imagine that we actually conducted all these mini elections, we now have to go from the results of those mini elections to an overall result. We consider as most binding the mini election that was won with the biggest majority. So if Alf trounced Bob, and Jimminy squeaked a win against Bob, the result we consider as most important is Alf trouncing Bob. Now we have a list of overall preferences, in order of importance. If there are any contradictions between two results, we throw away the less important result.
From the overall list of preferences, we can work out the winner. So if our list of preferences (in order of importance) goes
1. Alf is preferred over Bob
2. Jimminy is preferred over Alf
3. Jimminy is preferred over Bob
Then the winner is Jimminy. (He was the candidate of conscience anyway).
This claim was inspired by
Similar claims
- 4-4 Approval voting leads to massive tactical voting.
- 22-5 I would vote for a 3rd party governor if I agreed with their platform, regardless of their chance of winning.
- 25-0 Sometimes elections are not a matter of picking who is the best for the job, but the choosing the lesser of many evils.
- 4-11 Voting age should be lowered to 17.
- 20-7 The French electoral system sounds nifty.
Discussion (7)
I'm not sure I like the system, though. It's sufficiently complex that I worry that some counterintuitive results could arise.
I can't really deny that, although it does have some nice characteristics that many more popular forms of social choice method don't.
Kenneth Arrow proved that no voting system with 3 or more alternatives can satisfy all the desirable criterion, so we have to choose one that we think does the best on the most important ones.
Since I'm particularly offended by vote splitting and strategic voting, I like systems that reduce those effects.
Here's the ranked pairs scorecard:
Passes
Passes a weakened form of
Fails
The only time Condorcet (Ranked Pairs) is counterintuitive is when the Condorcet paradox arises (i.e. A beats B, B beats C, but C beats A). The Condorcet paradox is an actual conflict in the preferences of the voting public; the people honestly can't agree on who's the best candidate. Other voting methods just ignore the paradox, whereas Condorcet actually acknowledges that it sometimes happens.
When the Beatpath variant is in use, the resolution to the Condorcet paradox is just about the most intuitive resolution you can hope for.
This is much more concise than wikipedia, but I think it would still take 3-5 minutes to explain on average. There might be the rare person who will understand immediately in a minute or two, but I don't think that's the average case.
I do think ranked pairs is a huge improvement over plurality, though. If I had choose between plurality, approval voting, and ranked pairs, using each voting system in turn, my votes would be thus:
Plurality:
approval voting
Approval voting:
plurality - no
ranked pairs - yes
approval voting - yes
Ranked pairs (1 is best, 3 is worst):
plurality - 3
ranked pairs - 2
approval voting - 1
I'm going to disagree with the idea that it is understandable by enough people. Besides, there are better, simpler to understand systems. Range voting for example.
I've converted to the Range Voting religion since my last comment on this claim, which completely sidesteps Arrow's Impossibility Theorem by not being a "voting system" per Arrow's overly-narrow definition, but I continue to agree that Ranked Pairs is straightforward enough to meet the terms of the claim.
Two problems.
1) Our current voting system disenfranchises people under some circumstances, and it is pretty simple.
2) Ranked Pairs is more complicated than 'one vote' and range voting in terms of the voting mechanics IMHO.
Thus, even if you can explain it to the person of 'average' intelligence, there are many people who won't get it. Too many for it to be a good plan versus a better, more understandable method.