Michael Goldman posted this YouTube video Why Is Ranked-Choice Voting Bad for Third Parties?
This simulation looks at ranked-choice voting (a.k.a. instant runoff voting, hare method, alternative vote). In particular, it looks at how voters still get punished for ranking their honest favorite as first.
The video is interesting. It would take me too much time at the moment to study it closely enough to find the fly in the ointment, but I suspect there is one. In the example of the problem with rank choice voting, I think he made a couple of questionable assumptions, First, on a bell shaped curve it is not the width of a segment, but its area that counts. If you take a wide segment that includes parts of the curve with a low height, then the area will not be comparable to the width. Second is the assumption that the blue segment being squeezed out from below would all go red. In the current election I found some voters who were truly torn between Donald Trump and Bernie Sanders. So there may be people on the far right who would also vote for some one who is pretty far out on the bell curve to the left. In the world of infinities, sometimes the far left merges with the far right.
These are my suspicions, but I would really have to write my own computer model to test my suspicions. I have the skills to do it, but not the time or inclination right now.
I want to keep track of the link to the video because it does point out that it is complicated to figure out what voting method is “best” in all cases if there even is one that is “best” in all cases. The other reason for keeping track of it is that when I have time to write my own model, I might find out that this video is correct.
January 15, 2017
Found the Fly in the Ointment
First, the video plots people in a bell shaped curve along a political spectrum. That in itself is a fantasy with no real-life meaning. There is no tight correlation between a place in a political spectrum if you could even define what that means and how a person would vote for one of three candidates. If you could find a valid way to plot people along a political spectrum, you could not bunch together all the people who voted for one of three candidates along the whatever that measure was that you came up with to judge their place in the political spectrum.
Second, the example where a shrinking middle of blue people who are on the left side of the middle put a red candidate as their second choice, you cannot say that they failed to get their preference when the red candidate won. If they voted the red candidate as their second choice, then that has to be who they deserved to get as their second choice. If they did not understand how the voting rules worked and they voted a red candidate as their second choice when the green candidate was really their second choice, then that is the equivalent situation of the butterfly ballot in the 2000 election in Florida. By a fluke or a trick many people did not vote for the candidate they wanted. That is an issue that is entirely separate from ranked choice voting versus the current system.
You have to be suspicious when pseudo-science enters the realm of human behavior. In this case, making a choice between three candidates just does not fit on a smooth, continuous bell shaped curve. To plot it this way is to put mathematical precision on something that is not even measurable. To then have to misuse the meaningless plot to prove your point, you must really not have a valid point to make.