A tweak to IRV to make it a Condorcet method
-
@wolftune this is a well-known Condorcet method due in spirit to Tideman and called “Bottom N Runoff” where N=2 (hence “Bottom Two Runoff,” I.e. BTR or B2R). Generally speaking, these methods use some kind of absolute criterion (like least number of first place votes, lowest score, lowest approval, etc.) to decide which “bottom” candidates to subject to an elimination round, eliminates a Condorcet loser among them, and iterates until the desired number of winners remain. You are right, they’re pretty good methods. I like them.
score-stratified-condorcet [10] cardinal-condorcet [9] ranked-condorcet [8] score [7] approval [6] ranked-bucklin [5] star [4] ranked-irv [3] ranked-borda [2] for-against [1] distribute [0] choose-one [0]
-
@cfrank yeah, after posting I went further into the electowiki link etc
But I think my key view isn't just the decency of the method but the marketing as "IRV tweak". I imagine so much more success talking to RCV advocates and saying, "hey, there's a tweak that really improves RCV, it's just doing a bottom-two runoff when eliminating candidates…" and it sounds so much more "yes and" than other ways of bringing up issues with methods in practice…
-
If equal-ranking is permitted at the bottom (e. g., by leaving candidates unmarked) but not permitted at the top, then the Frohnmayer equality constraint is violated.
-
BTR-Score is easier to explain than IRV. No need to talk about transferring votes.
Rated methods are generally simple, here is may take on a few:
Score Voting Methods – Score, STAR, and BTR-Score -
@gregw for some reason, I can't click the link to see that other post you are referencing, but I was able to look into BTR-score mentioned in various places. I agree it is fine, but it is much bigger step (less of just a tweak) to go from people assuming RCV/IRV is all there is. Given a world of naive people just thinking IRV is the only alternative, the smallest tweak from there is easier to promote. With BTR-IRV, the ballot style can stay the same.
I'm not saying BTR-IRV is ideal, I'm saying it's more practical in the context of political reality.
-
@wolftune, any number of times, I have responded to PropagandaName's Tweets in favor of not-further-specified RCV by asking "Why do you oppose bottom-two runoff?". There has not been much response. Maybe more voices repeating the same question would help draw attention to it.
Approval-ordered Llull (letter grades) [10], Score // Llull [9], Score, STAR, Approval, other rated Condorcet [8]; equal-ranked Condorcet [4]; strictly-ranked Condorcet [3]; everything else [0].
-
@jack-waugh if you are describing RR from FV, I once had an argument about the bullet-vote claim and score voting, and it devolved to be obvious that some people have no interest in even softening their confidence about something in light of being questioned. I don't use Twitter/X at all anymore. I think best to bring up BTR in whatever random conversations when RCV comes up. More people will bring it up when more people know it is a thing etc.
-
-
Link fixed, I got a little sloppy with the paste.
-
@wolftune I do like this idea, it would also improve IRV-PR. Have you consider proposing this to RCV advocates?
-
@brian-lackey I think the main point is to use this to prompt IRV advocates to look at this issue, yes. I do not have plans to spend lots of time making this happen myself though
-
@wolftune My understanding is there are two variations of BTR-IRV. One simpler variant in which the bottom two candidates are only ever determined based on first choice votes, and a second variant in which the eliminated candidates have votes transferred to remaining candidates for determining the next round's bottom two.
In your write-up, you describe the simpler variant that is only ever based on first-choice votes. I like that variant better because it is simpler. Was that an intentional decision on your part? Is it important to you that votes not be transferred, or would you be okay with either method?
As an aside, in your write-up on the website the description of IRV does not mention the step where votes are transferred from eliminated candidates.
-
@kodos IRV my write-up on the website? I don't know what write-up or website you mean, I don't think I did a write-up. I shared a link to someone else's write-up though.
I would think that like with IRV in general, when a candidate is eliminated, the ballots that had that candidate first move to the next choice among the remaining candidates if the ballots marked any of them.
The simpler or not-simpler thing you mentioned makes no sense to me. No round is different from any other. Anything that uses different first round or something is less simple.
Think of it like paper ballots. You just put a ballot in the pile of whatever candidate is highest on that ballot among the remaining candidates. When you check for bottom-two, it just means the two smallest piles.
-
@wolftune Oh I see, the link is just from someone's website not something you wrote up. Got it, sorry.
I guess where I had seen the description of a "simplified variant" was on the ElectoWiki site, at the bottom of the page. https://electowiki.org/wiki/Bottom-Two-Runoff_IRV
Simplified Variant
If you remove the redistribution step, leaving the candidates in the initial 1st choice sort order for the entire process, BTR-IRV becomes precinct summable. Vote counting only requires the 1st choice vote counts and the pairwise preference matrix from each precinct, not the complete ranking counts.I think it's an interesting difference. Using your paper ballot example, this would mean putting the ballots into piles based on first choices only, and never redistributing them for the purposes of selecting the bottom two candidates in any round. This is what I understood from that write-up that is linked, as it does not mention any redistribution. It's "simpler" in the sense that you can skip the redistribution process at each step. It is also "simpler" in the compilation complexity sense, that it is precinct summable.
It raises an interesting question- if someone were to go to an IRV advocate and pitch them on a tweak, which variant to propose? BTR-IRV with distribution is closer to traditional IRV, so perhaps that is a reason to propose it, as it is more similar to traditional IRV. But on the other hand, BTR-IRV without distribution is simpler in some ways, which can be viewed as a virtue as well.
It's also worth thinking about how much the distinction matters in practice. Curious if anyone has thoughts on that.
