**INTRODUCING** 2-Choice Voting (2CV) - An Improved Iteration on RCV and STAR

cfrank

@psp_andrew-s unfortunately I am not available for a video chat at the moment, but I can elaborate later about the example. The key point is that the coins that get distributed over larger groups of similar candidates will cause those platforms to weaken, and that this enables minority rule.

psp_andrew.s

@cfrank Ok, I think I kind of understand where you're going with that, I'll think hard on it and let you know what I find. I'm also updating the model to support 8 candidates to run it through.

I will say, however, that I think the hypothetical is flawed, as I can't think of a scenario whereby 5 candidates from 1 single party could, or ever have, competed in the same general election as 3 candidates from ANOTHER party. Typically the POINT of parties is to align around 1 candidate, so those parties will have their own Primary before selecting whom to choose as their Party's candidate in a given general election.

Please advise and let me know if my criticisms of your hypothetical are valid or not.

Edit: I think your hypothetical requires the running of MIXED Primaries, which I neither support nor think are a good idea. But we can get into that separately if you genuinely believe that mixed Primaries are preferred to our current unmixed systems.

cfrank

@psp_andrew-s the reason we don’t see multiple candidates running on similar platforms is precisely because the “vote for one” system we currently employ suffers severely from vote-splitting. So platforms that run with more than one candidate are essentially doomed from the start, which is exactly what happened in the infamous Bush-Nader-Gore election. The primaries really just function to strategically navigate the failure of independence of clones. In a system that satisfies independence of clones, there is no intrinsic reason for many candidates with similar (read as: appealing largely to the same group of voters) platforms not to run for office. What I’m saying is that 2CV will probably converge to 2 large parties nominating 3 clones each, which is only nominally different from what we already have.

psp_andrew.s

EDIT: I may have misunderstood, I think your hypothetical assumes that there are 50/50 weighted voters for liberal vs conservative, right? I'll readjust and look at it.

psp_andrew.s

@cfrank I actually STRONGLY disagree with your premise that having clones in a general election is desirable. I think it's actually highly UNDESIRABLE and that we should want our general elections to offer us REAL CHOICE, meaning true, tangible, distintinct differentiation of candidates and ideas. No healthy election should only offer a choice between 3 shades of Red and 5 shades of Blue.

The party system is good and works, so long as it allows for alternative parties / moderates to win by getting a large amount of general approval (even if they're not 1st choice of the highest number of voters) - I think we can all agree with this premise.

That's the problem with our system - not that it doesn't promote the running of candidate clones (that's actually a GOOD thing, in my opinion), but rather that it doesn't allow for sufficient voter expression whereby they can choose a 2nd or backup candidate to gauge candidate acceptance / support more broadly than Single-Vote can indicate, and allowing for the victory of more populist candidates, which 2CV does elegantly.

I've thought very hard about your hypothetical, and I've come to the conclusion that it's an impossible situation, since it would likely NEVER be the case in an 8-candidate Mixed Primary Election (which is undesirable in and of itself, in my opinion) that no candidate would attempt a strategy of populist appeal and be more "moderate," to gain 2nd Choice Votes, since doing so is a huge competitive advantage. So I don't personally see that as a valid criticism of 2CV, though I'm open to more discussion if you have counter-arguments

cfrank

@psp_andrew-s I don’t get the impression that you are evaluating your own stance objectively, and I again do not mean to condescend, but it seems like you will need to do more research into this matter.

You are right that a healthy election should offer more than 3 shades of red against 5 shades of blue. What I am telling you is that I am very confident your system, if implemented, would converge almost always to 3 shades of red and 3 shades of blue.

cfrank

@psp_andrew-s no, my hypothetical assumes that, for example, there are 38% in one party that nominates three clones, and 62% in the other that happens to nominate 5 clones. Which is, actually, roughly realistic, assuming a small degree of naivety among candidates and that a larger and more diversified platform encourages more candidates to run for office. Which it should.

Jack Waugh

@psp_andrew-s said in **INTRODUCING** 2-Choice Voting (2CV) - An Improved Iteration on RCV and STAR:

@cfrank I actually STRONGLY disagree with your premise that having clones in a general election is desirable.

I don't think his point was that it was in general desirable or undesirable, but rather, that it should not be available as a strategy whereby a strong-arm party could prevent democracy.

Jack Waugh

@psp_andrew-s said in **INTRODUCING** 2-Choice Voting (2CV) - An Improved Iteration on RCV and STAR:

On Frohmayer Balance
If there are candidates ABCDE and 1 voter selects A/C as 1st/2nd choice, another voter can cast E/C as 1st/2nd choice to create balance between A/C and E/C preferences.

These two votes do not balance, because in the context of the other votes (the ones in the election but not stated in the example) the first one could deny candidate B the win, and the second would not restore B's win.

Jack Waugh

@psp_andrew-s said in **INTRODUCING** 2-Choice Voting (2CV) - An Improved Iteration on RCV and STAR:

Typically the POINT of parties is to align around 1 candidate, so those parties will have their own Primary before selecting whom to choose as their Party's candidate in a given general election.

Assuming that, first off, the voters, in order to have power, are required to organize themselves into parties, which I object to, and assuming that a given party conducts a primary election to determine its one nominee for the general, which I read you above as recommending, are you suggesting that the party is to use 2CV for the primary? That just pushes the problem from the general to the primary. Are the different factions of voters within the party supposed to organize themselves into subparties, and have a primary in each of those to determine their nomination for the party primary?

cfrank

@jack-waugh @psp_andrew-s this is precisely correct. Whether having clones in an election is unconditionally desirable or undesirable is immaterial and not a statement I make. What I will argue to be undesirable is minority rule by strategic nomination and tactical voting.

psp_andrew.s

@cfrank Please provide an example, because I cannot come up with one where I've seen multiple candidates from Party A running against multiple candidates from Party B in a general election. This should never happen and it's a bad system if it allows as such (does not promote sufficient candidate diversity).

Jack Waugh

@psp_andrew-s You shouldn't need an example from to-date real-world practice. The question is whether your proposal would encourage such a tactic.

I have seen a cartoon that shows Republicans rejoicing at the announcement that Nader is running. Is the suggestion in the cartoon plausible, even if not proven? Did Republican partisans have plausible reason to be glad at Mr. Nader's candidacy, on grounds of it improving the chances of their candidate Mr. Bush?

Jack Waugh

I suggest that for evaluating any multi-round voting system, a helpful (although not comprehensive) step is to ask whether, when it eliminates candidates in a non-final round, it eliminates the correct candidates based on equality of political power from one voter to another. And your first round shares with choose-one voting the fact that voters do not get to weigh in on every pair of candidates, but have a limited count of coins to spend. Accordingly, your first round inherits the problems of choose-one plurality voting, with slight mitigation in some cases, since your voters have two coins and voters in a choose-one single-winner election only get one coin. The improvement could show up when there are few enough candidates. With seven candidates, it's not a real improvement. Someone having money support for advertising, or distinction by fame or infamy, will move forward into the final round, because the voters face the prisoner's dilemma.

psp_andrew.s

@cfrank I have done an immense amount of research and have yet to find significant flaws that would make 2CV a worse system than the currently proposed ones such as RCV/STAR/YNA etc. That said, I consider myself HIGHLY objective and I'm willing to admit where I'm mistaken. I'm open to discussion on it, preferably by voice / video chat, since I feel that would be more productive. I will however say that, in EVERY scenario I've tested the model, it DOES NOT allow minority rule by strategic voting. I can show you if you want to look at the model together. Add me on Discord @lamppost

@Jack-Waugh If 2CV were implemented in our own General Elections, it wouldn't matter what method Parties used in their party primaries. Parties are necessary evil, like it or not, since people WILL form groups/coalitions to collectively enact legislative change. Therefore, implementing 2CV would allow multiple Parties to thrive and be competitive SIMPLY by getting 2nd Choice Votes. This is a huge change and significant upgrade on our current voting system while being easy to understand and implement, and works within our existing government and electoral frameworks.

To address your concern - using 2CV in a Ralph Nader-esque scenario would have still allowed Gore, who was more BROADLY approved of, to win, which I believe is the desired result given the polling data. If you want to have a video/screenshare discussion, we can run through a 7-candidate model and you can tell me yourself if it produces the desired result. Add me on Discord @lamppost

And in your scenario of ABCDE candidates, if a A/C 1st and 2nd choice vote would "deny" candidate B the win, then a balanced countervote would be B/C 1st and 2nd choice, which would equivocally restore their win.

Jack Waugh

@psp_andrew-s I'm not suggesting there should be laws to try to prevent people from associating with each other in groups and endorsing, as a group, a candidate. I believe in freedom of association. However, a good voting system will make parties for purposes of nomination so redundant that eventually, even the thickest-headed will see that they are no longer needed for that purpose.

In the Frohnmayer balance test, it's not allowed to assume the voters can know who the frontrunners are so they can tailor their vote. It's necessary that for every vote that can be cast, a single antivote will suffice, regardless of the other votes, to cancel that one.

Jack Waugh

@psp_andrew-s said in **INTRODUCING** 2-Choice Voting (2CV) - An Improved Iteration on RCV and STAR:

it wouldn't matter what method Parties used in their party primaries

If that's the case, why would anyone want to join? If the method used is inadequate, joining would not add to the member's political power.

What would prevent just as many candidates running in the general as would have, had there been no parties? How is it that parties are part of the solution to your system's vulnerabilities? Why do you oppose Approval Voting, which is so much fairer and also simpler (fewer rounds)?

cfrank

@psp_andrew-s I did already explain why this does not occur in the present system. Also I explained why the hypothetical example I gave is fairly reasonable, even though it is idealized.

I think your statement about “promoting sufficient candidate diversity” is also confused: what I mean is, how could, in effect, restricting the number of candidates who can run promote diversity? Even three shades of red has categorically more diversity than just one shade of red. I understand that certain candidates may be “crowded out” by a more homogeneous population of candidates, but what are the odds that a unique candidate gets supported by a large political party as opposed to one of the swath of other potential candidates in a system that suffers from vote splitting? It happens, but it’s rare. Why not let that unique candidate run on his own and argue his merit rather than screening him out from the beginning for the sake of a tactical response to an unnecessary conflict of interest?

Something to understand is Duverger’s law, and the forces that lead to the formation of a small number of large, competitive parties. One major factor, arguably the most important factor, is vote-splitting. Majoritarianism is another. “Patterns of Democracy” by Arend Lijphart lays out the distinction between majoritarian as opposed to consensualistic democracies, in some respects the problem always goes deeper but it’s a start.

I don’t believe that instantiating my example as described will fail to yield the outcome I mention without additional tacit and potentially inaccurate assumptions. One of these tacit assumptions I think I gleaned briefly in a table you had produced attempting to illustrate the computation of my example, which is that virtually everybody in each of the two parties has exactly the same order of preference. If you allow for more diversified preference profiles, I do think you will see the problem.

For the record, STAR voting also fails independence of clones. And if by “RCV” you are indicating “Ranked Choice Voting,” you may actually mean “Instant Runoff Voting” (IRV), which is a very specific and in some respects a particularly pathological example of the more general body of “ranked choice” systems.

Also I think this assumption illustrates correctly that your system is conceived of as operating within a one-dimensional political paradigm. This is the paradigm we currently face, but it is arguably a suboptimal consequence of the current two-party system in the U.S., and of the party-coalitions system in the U.K. (Duverger’s law in action). In other words, with an alternative system of representation in place, the political landscape may very well change to become multi-dimensional and issues-focused rather than one-dimensional and values-focused, if my meaning is clear.

SaraWolk

@psp_andrew-s said in **INTRODUCING** 2-Choice Voting (2CV) - An Improved Iteration on RCV and STAR:

Simple and easy to explain to prospective voters (THIS IS IMPORTANT)

Thank you all for taking the time to comment. I agree with most of the concerns about this proposal that people have already raised. I raised a few of them myself in the Twitter Space where I first heard about it, on Twitter, and also in Open Democracy Discussion.

To summarize a few.

I don't think the proposal is simple to explain, implement, or use, nor are the implications or relative benefits transparent.
A better voting method needs to be able to prevent vote-splitting and the spoiler effect and this does not.
A better method should work in a variety of political landscapes and this does not. That should include crowded fields, primaries, nonpartisan or partisan elections, single and multi-winner races, and many more variables. The stress test elections that are truly competitive are the ones where the system matters the most and failure in that scenario is not acceptable.
A better method should not magnify "electability bias" and by extension magnify the influence of money in politics. This one does that because with the limited choice, there's a strong incentive to only support frontrunners, not underdogs.
Since first hearing about this method last week I've seen many people spend many hours sharing constructive feedback and patiently explaining some of the central issues with it. Rather than hearing that feedback, Andrew, this conversation seems to be going in circles, so my central concern is that this is derailing in nature rather than contributing something novel and beneficial to the conversation.

psp_andrew.s

@sarawolk

We can address these.

Our research has shown that a 2-choice voting system is the 2nd easiest for voters to understand when compared with STAR and RCV. We can do some more research and even conduct live, representatively sampled polls with your assistance if you're interested in finding out which option lay-people feel is easiest to understand.
A better voting method does not need to completely eliminate vote splitting. In fact, any system that forces voters to make any kind of explicit preference decision will incur vote splitting. That's not an issue as long as the systems sufficiently reduces the IMPACT of vote splitting, which 2CV does to negligible levels.
You've given no evidence to support this. Every scenario that we've tested 2CV in, which includes a variety of political landscapes and candidate quantities has produced "desirable*" election outcomes. Please give some examples where this might not be the case.
2CV does not magnify electability bias compared to plurality voting. It may have slightly more bias than unlimited-choice systems, but unlimited choice methods introduce either gamesmanship (in the case of unlimited-choice Borda) or significant complexity (normalization of scores in Score voting). Most voters only have 1 or 2 preferences anyway, as every election results study has shown, and eliminating all data from any system on rankings of 4th choice or lower almost never has a material impact on the election outcome. 2CV asks voters for enough data to create a statistically sufficiently informed single-winner choice (assuming 250k+ voters, representatively sampled).
I've heard all the feedback. I've listened and taken it into account, which led to modifying the runoff election so that no candidate can win without 50% support (either a 1st or 2nd choice from the majority, if none of the top3 get 51%, we drop the 3rd place candidate by points and upgrade all the their voter's 2nd choice votes to 1st choice votes). 2CV now ensures, unlike any other proposed system, that the winner will, under all circumstances, be one who received at least 51% of 1st or 2nd choice votes. Which leads to majority-consensus election outcomes, always.

* We can have a discussion as to what "desirable" means, but for our purposes it means that the winning candidate had a significant amount of 1st choice preference among the voting population, as well as acceptance/preference from among 51% of the entire voting population.