Topic: Ranked-Pair Voting (Discussion)
Okay so as a need for discussion has arisen, here is a thread to discuss Ranked-Pair voting.
So I'd like to illustrate what's actually happening with ranked pair voting, or more specifically what's happening with the current vote for TC.
So for the sake of illustration, we'll assume ranked-pair voting for an election with 4 candidates: A, B, C, and D.
What everyone's seeing is this:
Candidate A
Candidate B
Candidate C
Candidate D
...and you are asked to order them by preference, highest being the one they most prefer voted into office, lowest being the one they least prefer voted into office. (By the way, the voting booth has the 4 candidates set to scramble the order, so odds are the ordering on your ballot will be different from roughly 96% of all other ballots issued - just mentioning this because this is strictly an example and the letters are not meant to be attached to any individual's names)
So you might set yours to look like this:
Candidate C
Candidate A
Candidate B
Candidate D
...and what then happens is the voting booth assigns the following votes:
Candidate C - 3 votes
Candidate A - 2 votes
Candidate B - 1 vote
Candidate D - 0 votes
So I think we can agree that without looking any deeper into this, things LOOK a little arbitrary. Okay, fair enough, so I'm hoping to elaborate a little further and maybe this will make sense.
What's ACTUALLY happening is there are 6 basic plurality votes going on in this single ranked-pair vote. Ranked-pair votes are actually several basic plurality votes happening simultaneously and it's done in such a fashion that it's easy for the voter to do in a short frame of time and it's easy for those counting the votes to do quickly and efficiently also.
What do I mean by 6 basic plurality votes? And what are these 6?
Well here are the 6 votes:
Candidate A vs. Candidate B
Candidate A vs. Candidate C
Candidate A vs. Candidate D
Candidate B vs. Candidate C
Candidate B vs. Candidate D
Candidate C vs. Candidate D
You'll notice that each candidate appears in 3 of each of these plurality votes and that they are put against each other candidate exactly once.
So let's assume you voted this way in ranked-pairs again:
Candidate C - 3 votes
Candidate A - 2 votes
Candidate B - 1 vote
Candidate D - 0 votes
So in the ranked-pairs, this voter said they prefer C to all other candidates, they prefer A only to B and D, and they prefer B only to D. I'll translate this to the 6 basic pluralities so you can see what's happening.
Candidate A vs. Candidate B - Candidate A wins (+1 vote for A)
Candidate A vs. Candidate C - Candidate C wins (+1 vote for C)
Candidate A vs. Candidate D - Candidate A wins (+1 vote for A)
Candidate B vs. Candidate C - Candidate C wins (+1 vote for C)
Candidate B vs. Candidate D - Candidate B wins (+1 vote for B)
Candidate C vs. Candidate D - Candidate C wins (+1 vote for C)
Let's count it up! C got 3 votes, A received 2 votes, B received 1 vote, and D received no votes. That's exactly what just happened in ranked pairs! (Notice: now that you've seen it in this light it's apparent that ranked-pairs captures 6 of the voter's preferences which is the full list of all possible preferences)
So IF we stuck to basic plurality, this voter would have said "C" and if their choice for a candidate wasn't the one everyone else went with then they have no say in who of the other 3 gets to win the election. This doesn't exactly seem fair. We could do elimination, but then we'd have to do multiple voting rounds to determine who gets the position which will take longer which also doesn't seem right. So ranked-pair voting does all of this.
Now realize that the voting power of the individual under basic plurality is 1 vote and every person gets 1 vote. The voting power under ranked-pairs (in this instance) is 6 votes and everyone has 6 votes. Where this is different though is that if their first choice doesn't win, they still have some say over who gets into office after that.
-pause-
So ranked-pairs can be viewed in a different light, namely that it is 6 basic pluralities happening at the same time. This is convenient because it gives us the full spectrum of the voter's opinions, pitting each candidate against each candidate once and over 1 vote. You can accomplish the SAME results a few different ways:
Elimination: you hold a number of votes equal to one less than the number of candidates, each time the voters choose who they DON'T want to win (with the option in the last round of voting for who you DO want to win). Elimination in this example would take 3 rounds of voting but would still accomplish an ordered set of preferences since this voter would vote out D first, then B, and then A (except if their choice doesn't match the majority and then we will not see all their preferences). In this example, the voter can declare no more than 3 of their maximum 6 preferences for candidates.
Ladder: as in sports, they tend to put only one team versus another team and then the victor goes on to play another victor. You could do the same with elections, choosing pairs of candidates to go head-to-head. So for instance, if the vote was A vs. B and C vs. D, this voter would say A and C, and then in the second voting round vote C which still establishes a ranked-pair (except we never learn whether they preferred B versus D and to what extent they preferred A to D and C to B, so it doesn't actually supply us with complete information of their preferences). In this example, the voter declares exactly 3 of their maximum 6 preferences for candidates.
Primaries: this is mostly similar to a Ladder except that it generally provides even less information because candidates are grouped based on certain identifiable characteristics (i.e. by political party) and then it's votes based on a pool of at least 2 candidates (were it only 2 in each pool then it would be a ladder) and then the final tete-a-tete. In this example, the voter can declare no more than 3 of their maximum 6 preferences for candidates.
Each of these requires multiple voting rounds, and they don't always give us complete information about an individual's preferences. It's of note that basic plurality identifies only 3 of the maximum 6 preferences (in the given example, that they prefer C to all other candidates).
Ranked-pair voting consistently gives all of the preferences of each person on the full spectrum of candidates. It does so fairly and quickly and it gives the person some say in who gets office if their first choice isn't the majority choice (that's a stab aimed at basic plurality).
Here's an example in opposition to basic plurality. Let's say there are 3 candidates, X, Y, and Z, and there are 2 candidates that are highly popular, say X and Y. People in favor of X and Y are so fixated on those two candidates, they don't remotely suspect Z would even win. Think something along the lines of elections in the US: we have Republicans and Democrats one of whom always wins (maybe one day the US will break the vicious cycle), and then a few other people that most voters think won't win.
It's entirely possible in basic plurality that the voters in favor of Republicans and Democrats squabble among each other so much that they ignore the possibility of a candidate winning that is neither Republican nor Democrat who both consider to be a REALLY REALLY BAD choice for President (I won't try to come up with a party that the Republicans and Democrats would both hate more than each other, but I'm sure your imagination can come up with something), and that they ignore the possibility so much so that their votes wind up the minority and this really really bad President takes the White House.
In the above example, ranked-pairs would give the Republicans and Democrats the ability to say "I want my party to win, but I'd rather have a Democrat in office than this REALLY REALLY BAD OTHER GUY!" If the Republicans and Democrats agree that they'd rather have their rival party win than this third candidate, then they may rally together and hold majority to dodge a really nasty bullet that is this third party candidate they really strongly dislike.
So I guess ranked-pairs could serve to aid in compromise by allowing us to say "I'll settle for ______ if the person I really want to win doesn't win", and it also makes it easier to determine tie-breakers (examining who has the higher concentration of votes in-favor).
While this thread is supposed to be for "discussion", I feel my above explanation might put a damper on such a notion, so I'd like to make an inquiry as to whether we'd prefer another system, i.e. elimination, ladder, or primaries (assuming we can figure out some sort of system for grouping people)? And maybe with what reasons you feel another system would be more beneficial? Or contrariwise, if you'd like to make mention against a system please feel free to do that as well.
After my above explanation, I think it's apparent that basic plurality is lacking when there are more than 2 candidates, so I think it goes without saying that I'm against basic plurality when there are more than 2 candidates but you're welcome to your opinion of course.
I will say though that a major problem with doing basic plurality in our existing system is that we allow for a candidate to decide, having seen the election results, to decline their office. In basic plurality there are two ways to handle this and they're both bad:
Take the runner-up: that's giving one person (the winner of the election) the power to say "thank you for the votes, but I'm going to throw away the majority of the votes which means only the people that didn't vote for me have say". Currently our legislature says the runner-up is next which works well for ranked pairs but not basic plurality.
Re-vote: obviously this is the most democratic way of handling things but it's quite time-consuming.
Effectively with ranked-pairs, we already have enough information on the ballots to recalculate the winner if one person withdraws so a re-vote isn't necessary and the runner-up is already a close match. We don't have it in our legislature that a recalculation is needed if someone steps down partly because that takes a fair bit of record keeping and calculating on the part of the EVC but statistically speaking we aren't taking a very large hit to our numbers, very significantly less of a hit anyways than what we would be taking under basic plurality (if anyone wants to know precisely what % of a hit it is, I can calculate that and probably will do that over lunch now since I'm intrigued).
~Robert
