Why I'm now leaning toward Copeland//IRV as the best method to promote
-
Copeland//IRV is a ranked ballot system that chooses the Condorcet winner if there is one. If there isn't, it will eliminate all candidates that are not in the Copeland set (those that have more pairwise wins than other candidates), and then do an IRV tabulation to determine the winner.
Remember, it only uses IRV if there is no Condorcet winner. Every single Ranked Choice election tracked by FairVote had a Condorcet winner. So IRV would almost never actually come into play.
This should be as good as any method at ensuring that people don't have an incentive to vote strategically (no need to study the polls or guess how others will vote), and as good as any at avoiding incentives for parties to be strategic by eliminating candidates before the general election.
Why ranked ballots rather than score ballots?
Because most voting machines don’t support score ballots. Score ballots would be a huge barrier to getting an alternative system adopted, as it would be a big expense and complication for any district. [1]
Also, people already understand ranked ballots, as they have been used in IRV elections for a long time now. Even in districts where they don’t yet use IRV, a very large number of people are familiar with the concept.
The vast majority of the education used for IRV still applies. If people have gotten confused, we've learned how to educate them and don't have to start from scratch.
Finally, the term Ranked Choice is now widely known. This can be considered a Ranked Choice method. People like Andrew Yang, who has been very vocal about RCV, don't need to backtrack.
Why IRV and not something else?
Because IRV is as resistant to strategy as most any non-Condorcet method. Because IRV is a known quantity, and has been used successfully in thousands of real world elections. Because almost all voting machines and voting infrastructure already allow for it. On the latter point, this should dramatically reduce the costs associated with switching over.
Keep in mind, the vast majority of complaints about IRV are not an issue here, because it is only used here as a rare last resort. Most of the time, the Condorcet winner will be chosen and the IRV step is completely unneeded.
Why Copeland and not Smith?
Because Copeland is extremely easy to explain and to show results. You just show a list of candidates, sorted by the number of pairwise wins with that number listed next to the candidate. Those that tie for first place are the ones considered in the final round.
You can of course show a pairwise matrix. If you use some color, it makes it all the better. (count the green squares in each row to determine the number of pairwise wins)
Smith is much harder to explain, and any graphical way of presenting the results doesn’t instantly communicate what the Smith set is.
There is a concern that Copeland is susceptible to “teaming”, the opposite of vote splitting. It is a very mild effect, though (given that elections without a Condorcet winner are rare), and can be considered a counterbalance to one of the worst things about FPTP, which is vote splitting and the resulting strategic nominations (i.e. primaries etc).
Is it precinct summable?
Like IRV, it isn't....but....
If the pairwise matrix is submitted (as with STAR), that is good enough, most of the time. It can be determined from if there is a Condorcet winner, and if there is, the winner can be shown immediately. Only in the rare case of there being no Condorcet winner (remember, this has never happened in a Ranked Choice election so far) does anyone need to wait for results.So…..
The main idea here is to get the benefits of Condorcet elections, while making it understandable and therefore marketable to the masses.
Footnote:
[1] here is the the thread which convinced me that score ballots, as much as I personally like them, are a no-go:
https://www.reddit.com/r/EndFPTP/comments/w1r6f4/comment/ih6wab0/?utm_source=reddit&utm_medium=web2x&context=3And here is a bit from that thread (it is much longer than this, and choco_pi clearly knows what he/she is talking about):
But it's said that current machines can handle score voting. Do you have proof on your statement?
.It's hard to respond to claims like this because they're just, super wrong. Like it's so disconnected with the facts on the ground it's difficult to know where to begin.
...
Conversely, 0 of these 11 vendors support scored ballots. Not the ballots, not the voting machine UI, not the tabulation, not the reporting, not the auditing; none of it. You could try do an externally tabulated workaround, but in this case I think you'd find the reality of that compromise to be unacceptably ugly, 10x worse than the RCVTab stopgap. (A ballot with a question asking which candidates you would like to score 0, a seperate question asking which candidates you would score 1, etc.) -
Why do you prefer Copeland/IRV to Ranked Robin?
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 said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
Why do you prefer Copeland/IRV to Ranked Robin?
If Ranked Robin had momentum, I'm all in.
How about this: I'll add Ranked Robin to my signature as my second choice if you'll put Copeland//Score as your second choice.
But really, my main issue is marketing. I think the fact that IRV has so much more momentum than any other alternative system says to me that this would be easier to sell. Borda count is widely seen as unacceptable on its own. I have been convinced that IRV is incredibly resistant to strategy. (again, user choco_pi over in EndFPTP has convinced me of that).
I understand that strategy will play out differently if a method chooses the Condorcet winner first. In practice, Ranked Robin and Copeland//IRV are so similar that they would probably produce different results in one in a billion elections. I don't think anything would be different.... parties, nominations, strategic voting, etc, between Ranked Robin and Copeland//IRV.
I'll admit I am not a fan of the name as it could apply to any ranked Condorcet method, and the method already had a name (Copeland//Borda). But whatever.
The page for Ranked Robin goes into ridiculous detail about breaking ties, which are unnecessary. (these sort of ties are incredibly unlikely) That's less a complaint about the method, but still. Not how I'd present it, it seems more complicated than it is.
But at the end of the day, sure, Ranked Robin is groovy. I think we could sell this one to Ranked Choice people a bit easier, though, given that it so many people already have placed trust in IRV and written legislation for IRV, and this is more likely to be seen as a minor change from regular IRV. (even though, from the point of view of people who like Condorcet and hate IRV, including me and you, it is a huge change)
-
Ranked Robin (RR) has nothing whatever to do with Borda. Borda extracts scores from ranks. RR says first, if one candidate beats strictly more count of other candidates than any other candidate beats, elect that candidate. If there is a tie, break it by counting the voters who prefer the candidate to each other candidate minus the count who prefer the other candidate to the candidate in question.
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].
-
Your signature rates Cardinal Baldwin pretty high. Sara Wolk said, if I remember correctly, that it fails monotonicity.
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 said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
Ranked Robin (RR) has nothing whatever to do with Borda.
Sorry, but it's indeed Copeland as the first step, Borda as the second (if the second step is necessary).
According to Ranked Robin's Electowiki page (in describing the pairwise tiebreaker):
"This is mathematically equivalent to the tournament-style of the Borda count"
Later on it says: "Ranked Robin can fail clone independence in one of two ways: either by its Copeland component or by its Borda component."Note that you said: "RR says first, if one candidate beats strictly more count of other
candidates than any other candidate beats, elect that candidate."Which of course is Copeland.
So Ranked Robin is Copeland//Borda.
I got it first from user cathulu in EndFPTP, where he said: "Ranked Robin is a renaming of Copeland//Borda" in this thread: https://www.reddit.com/r/EndFPTP/comments/w45cf5/what_is_the_name_for_this_simple_cardinalballot/
(regarding Cardinal Baldwin, feel free to make a separate thread if you want a discussion. I don't really like it when threads get derailed by completely unrelated subjects)
-
This post is deleted! -
Would you recommend the method for a State to determine whom to send to the Electoral College for POTUS? I think that if the system were used for a State-wide election, if the IRV tiebreaker were ever needed, it would take too long to tally, causing a crisis.
Also, anything that promotes IRV in any capacity is scary, because IRV advocates would point to it as something that has drawn favor and they would say that justifies using a pure IRV election.
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 said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
Would you recommend the method for a State to determine whom to send to the Electoral College for POTUS? I think that if used State-wide, if the IRV tiebreaker were ever needed, it would take too long to tally, causing a crisis.
Maine already uses IRV for the presidential general election, so I don't see how this one would be an issue.
How long do you think it takes to resolve an IRV election? There's plenty of time.
@jack-waugh said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
Also, anything that promotes IRV in any capacity is scary, because IRV advocates would point to it as something that has drawn favor and they would say that justifies using a pure IRV election.
One, that's an extreme stretch, given that it doesn't use IRV 99% of the time, and IRV is already in use in a large number of elections.
Two, this could actually get people to use something better than IRV.
Three, I think your fear of IRV is WAY overstated. IRV has been working fine here in San Francisco. (and has always elected the Condorcet winner). The one case where IRV didn't elect the Condorcet winner, Burlington, would be solved by this method (or for that matter, Ranked Robin).
I gotta say I'm confused by the way you approach this. I went back and saw you argue against Ranked Robin, saying this:
W'pedia lists for Borda: "When a voter utilizes compromising, they insincerely raise the position of a second or third choice candidate over their first choice candidate, in order to help the second choice candidate to beat a candidate they like even less. When a voter utilizes burying, voters can help a more-preferred candidate by insincerely lowering the position of a less-preferred candidate on their ballot. "
If the system incents such strategies, it's no better than FPtP.
Which shows a complete misunderstanding of the fact that it only uses Borda to break pairwise ties, which are very rare. Any strategy would be far far more difficult if the method chooses the condorcet winner and only if there is none, falls back on a different method.
And now you are saying that the method has nothing to do with Borda, which is doubly confusing.
-
My writing is confusing because I have allowed myself to become confused.
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 said:
My writing is confusing because I have allowed myself to become confused.
Ok. So can you..... work on that?
Or, could I at least request that you don't post so many separate small comments? If you feel you must add another thought, just edit your last comment. It feels like whack-a-mole, and every time I contribute to this forum it just feels like things get derailed. Especially if a comment isn't on topic, it really shouldn't be its own, separate comment. (admittedly, this comment is off topic, but this thread has already gone off the rails)
FYI I've been having MUCH better luck at the reddit EndFPTP forum than this one recently, in terms of having people read what I write and having a discussion that feels like it goes somewhere rather than just being frustrating. Which is a shame. It's not just you, but still, you are the guy running it and I'd suggest you should take some responsibility for the direction of the forum. There is so much potential this board has that, in my opinion, is being wasted.
I also like that that forum has a policy of not slamming on IRV.
Your stridency against IRV --- even in cases where it has hardly any impact on the outcome, such as the above --- simply reinforces the status quo.
-
@rob there are many methods that make strategic voting unnecessary, such as selecting a citizen uniformly at random.
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]
-
@rob The complexity of IRV has slowed its spread, even though there are far more complex electoral rules. At public hearings to implement it somewhere new, it gets criticized for being confusing. I think people in the voting reform space tend to underestimate the complexity of hybrid voting methods like this one because they are composed of building blocks we already are familiar with, but for people not already initiated, having to explain all the building blocks means more of a chance to lose them before you get to make an argument. If I were to try to promote some Condorcet method, I'd probably go with something like Minimax, since the cycle-breaker is fairly simple but also in practice very similar to the cycle-breakers of some of the more highly regarded Condorcet methods.
(I will note that in my opinion, we will almost never see a Smith set with more than 3 candidates, and furthermore, if there is a cycle, all of the pairwise matchups will be close; 51-49 all 3-ways will be far more likely than 60-40 all 3 ways.)
Score Sorted Margins[100]; STAR[90]; Score[81]; Approval[59]; IRV[18]; FPTP[0]
-
@cfrank said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
there are many methods that make strategic voting unnecessary, such as selecting a citizen uniformly at random.
I'm not sure what you are getting at there...I don't think I implied that making strategic voting unnecessary was the only criteria. Was that your interpretation?
I listed some other criteria, but if I didn't state it outright, a primary one was that IRV, while far from perfect, is already considered good enough all by itself by huge numbers of people who are advocating for it, and actually using it in real world elections, including the entire state of Maine, many cities here in the Bay Area, etc.
That is obviously not the case with "choose a random citizen".
Here I am suggesting IRV simply be used in the rare case of pairwise ties. (again I should point out: in all 440 ranked choice elections monitored by FairVote, every single one had a Condorcet winner. One of them, Burlington, did not elect the Condorcet winner with IRV. But this method would elect that Condorcet winner, since its first step is to choose the pairwise winner if it exists) So even if many of us here don't consider IRV "good enough all by itself", I would argue that it is good enough for an extremely rare tiebreaker case.
Notice that a similar method, "Ranked Robin", works exactly the same as this, except for using Borda count for resolving the pairwise tie. It's really a renaming of Copeland//Borda, and while I have no problem with that method, using IRV may have an advantage in terms of being able to market it to all those who are behind "normal" Ranked Choice (IRV).
If you look at this FairVote page, https://www.fairvote.org/research_rcvwinners , they state as a positive about IRV that it tends to elect Condorcet winners if they exist. This method simply says "ok, fine, lets just go ahead and pick that Condorcet winner if they exist, and if not, use your method." FairVote itself may argue against this, but I can't imagine how they'd justify it if they just said that picking Condorcet winners is a good thing.
Back to "choose a random citizen." I'm not sure if you are suggesting it as a way of breaking pairwise ties, or have some other reason to bring it up. Assuming the former, I think it would be a bad choice, not so much because it would choose a bad candidate (after all, the candidates that are tied for pairwise wins should all be considered perfectly viable candidates), but because it is non-deterministic. The same set of ballots could produce different results. That's messy (adding an extra step to the process) and unnecessary.
-
@marylander said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
At public hearings to implement it somewhere new, it gets criticized for being confusing.
I would totally agree if not for the fact that IRV already has momentum, most people who don't live under a rock have some familiarity with it, it has been used in real elections for decades now and.... key point... in all 440 election(*) where ranked ballots have been used, there was a Condorcet winner, so this tiebreaker stage wouldn't even be necessary.
Tldr: it just seems weird to think people would get hung up on the IRV part of this method, since it would be so rare to be needed, and IRV is already such a known quantity.
Let me make an analogy. Say someone is considering an off the grid, solar and battery powered house. If for some reason the system goes down for a short while, it is proposed that they have a gasoline powered generator as a backup. How much time would you expect people to spend debating the complexities of gasoline engines? (despite internal combustion indeed being complicated and having downsides, it is also a well known quantity)
* the 440 elections monitored by FairVote that they had enough information to be able to determine whether there was a Condorcet winner.
-
@rob said in Why I'm now leaning toward Copeland//IRV as the best method to promote:
- the 440 elections monitored by FairVote that they had enough information to be able to determine whether there was a Condorcet winner.
How many of those elections reported the complete rankings of the ballots? In an election where the complete rankings are not reported, it will still be possible to determine the Condorcet winner if one candidate gets a majority of the first choices, and IRV will elect that candidate. However, if no candidate gets a majority of the first choices, then it will not be possible to determine the Condorcet winner. This could lead to selection bias in their dataset.
Score Sorted Margins[100]; STAR[90]; Score[81]; Approval[59]; IRV[18]; FPTP[0]
-
@marylander They say:
"Of the 440 single-winner RCV elections in the United States since 2004 in which we have sufficient ballot data to assess whether the Condorcet winner won the election, all but one — the 2009 mayoral election in Burlington, Vermont — were won by the Condorcet winner."
https://www.fairvote.org/research_rcvwinners
So that says to me, all of them had complete ballots. There presumably were some elections without complete ballots, but they were simply ignored in this sample.
In this article, they go into a bit more detail. mentioning how in the Bay Area, all ballot image is revealed after being scrubbed for personally identifiable information. (in the bay area there have been 138 elections, all had a condorcet winner and IRV elected them, 40 of them did not elect the majority candidate)
https://www.fairvote.org/every_rcv_election_in_the_bay_area_so_far_has_produced_condorcet_winners
-
@Sass, is there any reason to doubt what "Fairvote" reports as facts?
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 The ballot data is publicly available and is free for you to check their claims yourself.
Personally speaking, I spent (way too many) hours scraping publicly available STV ballot data from about 1400 elections in Scotland and in New South Wales, AU. Each one except for around 14 (so 1%) had a Condorcet winner which was elected.
And that's STV, so we might expect an even higher incidence of cycles than with IRV.
-
@jack-waugh Depends on the context. I haven't spent a ton of time scouring their site because it infuriates me, but in my limited experience, their hard data on what voters actually put down on ballots tends to be pretty reliable. Their claims about what different voting methods do and don't do are usually pretty bunk all around, though.
