Yet another wording of IRV fundamental issues

This is yet another wording to try (worked for me recently) in trying to get it to click for people that IRV counts unfairly and unequally:

In RCV, if my 2nd-choice is eliminated in the first round, I don’t get that preference ever counted, even if my 1st choice gets eliminated later.

Some preferences get counted, others ignored, and that simple fact makes it far less democrat and fair. If I really care about my 2nd choice getting counted, I need to strategically put that 1st (favorite betrayal), and then we’re back to lesser-evil voting!

That should make it clear, and again, every single pathology of RCV can be derived from this fact.

A spelled out example case (click to expand)

For example (just to show the concept), say A wins over C. The 2nd-choice of most voters, B, was eliminated in the first round. Only the B-1st voters get their 2nd choice to decide between A vs C. The craziest, craziest fact (non-monotonicity) is that if C voters want to stop A, they can vote for A first!!! By switching their votes to A, that means less 1st-choice votes for C. If C drops enough, then C will be eliminated in the first round instead of B. That means the C voters are the only ones who get their 2nd choice counted. And since their 2nd choice is B, B will win. Of course, it makes more sense for the C voters to just strategically vote for B rather than vote for A. The point is: the C candidate is a spoiler. By existing, they remove enough votes from B to get B eliminated, but some B voters have A second, which makes A win. So, the C voters need to either directly help B or just otherwise sabotage C (their actual favorite) so that only the C voters get their 2nd choice counted. The mathematical fact is that one way to sabotage C is to vote for A, and that means more votes for A can make A lose! Really, this is just a bunch of nonsense around worrying about who gets their 2nd choice to count. It all happens only because RCV fails to just count everyone equally.

1 Like

Wouldn’t this immediately backfire? Their votes would count for A in the runoff between A vs. B, guaranteeing even further that A wins. They’d have to vote for A just enough, but not too much.
The other thing, if someone tries to say that this can be gotten around by polling and coordination: it can’t. Polls only show 1st-preference support, which people naturally equate to total support, so C-top voters would have a hard time convincing fellow C voters that B has any better of a chance of winning than C when he got less votes in the first round and was eliminated first. Maybe wealthy candidates can assist by publishing head-to-head polling, though this is less likely.
I do want to point out one thing about Burlington, which naturally comes to mind after looking at the example: a lot of Republicans would have to had be willing to compromise to make the Democrat win instead of the Progressive. The final runoff between the Progressive and the Republican ended up with the Republican losing by only 3%, a margin so narrow that I’m not too sure Republican voters would’ve traded it in exchange for a sureshot win for the compromise candidate that would completely eliminate the chances of their favorite. Even if Burlington Republican voters had access to all information and were fully aware of IRV strategy, it would’ve required a full quarter of them switching their 1st preference to the Democrat to make them win, and it’s not certain they would’ve embraced that choice, at least not before the election had taken place.
Anyways, I think the point I’m trying to add to yours is that while IRV does actually allow for better elections, its polling issues and complexity make it functionally almost impossible for voters to work out how to make their choice. In a cardinal system, the choice of whether to compromise or take a risk on your favorite is crystal clear, but in a ranked system, there’s a lot more computation and information required to make that decision.

Sure, it’s not the actual smart strategy to vote for your least-favorite. It’s just a proof of concept of non-monotonicity. You can move a small number of votes to an otherwise winning candidate and cause them to lose. The very concept is intuitively grotesque. Whatever method takes away votes from the spoiler candidate (C in my example) will give the win to the spoiler’s 2nd choice (B).

This doesn’t happen to work in FPTP though. C moving to A there won’t stop the spoiling because FPTP never combines the C and B votes for B.

Right, but that doesn’t matter. The point is that if just a small number (precisely 753 in the Burlington case) votes were gone from the Republican, that would make the Republican get eliminated instead of the Democrat. That tiny difference then moves 100% of the Republican vote to their 2nd choice (the Democrat), thus giving the Democrat the win.

Overall, IRV probably will have improved outcomes over FPTP, but it will have problem situations often and the expense of a screwy unfair system isn’t nothing. I wish I could come around to the peaceful tactic of embracing all positive reforms. However, I believe that IRV having its serious flaws is a threat to the credibility of voting reform overall. I think Burlington (particular the Republicans there) are likely to oppose all reforms after having been burned by IRV. I think bad or mediocre reforms can at least sometimes spend limited political capital and close doors for better reforms rather than being a step in the right direction. It depends case by case. I’m not convinced IRV is safe. I think moving from IRV to STAR is less likely than just FPTP to STAR etc.

It seems to me Burlington was a very special case of having 3 strong candidates. The number I remember hearing is something like 15% of all close 3-way contests will have non-monotonicity failures, which I interpret to mean not Condorcet-efficient. However, you do have strong evidence in your favor, since almost half of the towns that have implemented IRV (ignoring from decades ago) have repealed it in the USA.
Certainly, I’m not recommending IRV with strength, but it is a good idea to be at least mildly encouraging towards it at all times, since IRV advocates can and have done damage to voting reform movements (i.e. cardinal) that were overly critical of them. Also, at the local level in particular, IRV has the greatest benefit, since you often have several weak candidates splitting votes with the top couple of candidates, so there you might find it easier to endorse there as well.

Well, the whole premise of preferential voting is that we shouldn’t be stuck with situations where there’s only two viable candidates!!!

So, yes, IRV works fine at letting weak-candidates show that they have real support, as long as they stay weak. IRV solves the problem where the wrong one of the strictly two front-runners is chosen because of vote splitting with a weak third candidate.

The whole point is that once you use IRV to show that more than 2 candidates may have substantial support, you inherently bring in the prospect of 3+ strong candidates, and that’s exactly where IRV fails most badly!

So, yeah, two strong candidates is the norm with FPTP, but not necessarily with IRV at least initially. Except it degrades to two strong candidates once voters realize that they need to do favorite-betrayal to stay safe given IRV’s flaws.

Basically, you’ve got the cause-effect direction backwards. 3 strong candidates isn’t very special because of something fundamental, it’s special because FPTP can’t handle it, and that then causes people to focus on picking two front-runners to focus on…

Considering that Burlington is the only city in the USA to have a known Condorcet failure, and since they repealed it, we have no evidence of voters learning to Favorite Betray in IRV, what are you basing this off? My theory (well, not mine, but I agree with it) as to why it devolves into two candidates is not so much that people have experienced Condorcet failures everywhere, but rather that they assume candidates with less 1st preference support have a lower chance of winning, naturally tending to picking only from 2 candidates over time.
I agree IRV is terrible for 3rd parties, though I think it likely makes the two parties more compromising, since they don’t have to worry about trying to pick up protest votes from 3rd parties (so long as they stay small.) Australia is an example where the duopoly tends to be more centrist from what I’ve heard.

My personal strategy has been to absolutely oppose ‘ranked choice voting’ (‘RCV’/‘IRV’) and advocate the simplest form of score voting – “1” to “10” plus ‘abstention’. I don’t want a “0” vote because I think it’s barbaric. Further, voting with “0” does absolutely nothing except express dislike; it does not affect any (simple) score election at all. People need to focus on making elections really DO something, not on letting them become expressions of irrelevant opinion. I have started picketing in front of city hall.

My ‘criteria’ seem mostly unique around here. I absolutely reject computer voting and demand hand counted paper ballots. I believe elections are primarily contests between aristocracy and commonalty – not primarily between the candidates, whom I see as essentially ‘pawns’. I believe that, in order to defeat the aristocrats, voters must vote strategically. (How can dishonesty even exist in a private voting booth, where no force or fraud is possible?)

The ‘non-monotonicity’ argument has performed good service, although it certainly seems too abstract. So here’s what I think is a better one: ‘The Sooner No Harm Criterion’. There is a concept of so-called ‘clones’, and the candidacy of these so-called ‘clones’ could cause otherwise winning other ‘clones’ to lose to some largely disfavored candidate. I don’t ever speak of ‘clones’, since it’s unrealistic to think they could be completely alike, so I call them ‘kindreds’ – they are perhaps very similar but of course not identical. So here’s a very real problem for RCV: You, the voter, step into the voting booth, and there are, say, “6” kindreds (from your perspective), that is, they are all pretty much equally desirable to you. But you are forced to rank them, so you pretty much at random pick one for first place-in-rank, and then you rank the other five more-or-less at random also.

There happens to be one candidate that you regard as a heel, so you rank that one in the lowest possible place (if any). And lots of other nice people pretty much like you do this very same thing. Except their random choices happen to be ranked in various different orders, naturally.

However, there is a very small, evil minority that favors the heel. And they all bullet vote for this heel, and of course they rank your favorites in the lowest possible places (if at all). But these evil ones all do this same thing – they ‘bullet vote’ for this one heel. So after all the counting and eliminating, the heel with the tiny evil minority comes out the winner. The overwhelming majority loses, and the tiny evil minority carries the day. This is how bullet voting can trigger RCV’s ‘sooner no harm’ violation. I did give a very slap-dash, but well fleshed-out example of this phenomenon at:

[HERE]

And score voting is not susceptible to this sooner no harm effect, nor to ‘bullet’ or ‘min-max’ voting. The only real winning strategy would be to give your desireds “10”, and lesser evils “9”, “8”, “7”, etc. (I call this ‘hedge voting’.) Bullet voting would be useless.

It’s more than a bit odd that I’ve seen ‘RCV’/‘IRV’ proponents complain endlessly about the ‘hazards’ of bullet voting for score voting elections.

Nor do we have evidence to the contrary. We have little insight into the thoughts behind the ballots overall.

I don’t know enough about Australia or Ireland situations. But I do have the clear impression that people find the IRV tallying relatively opaque. There’s probably lots of cases of Condorcet failure. Burlington was most unique in that they released all the ballot data so we can see the failure clearly. Most cases we can only speculate.

Hmm, I’m not sure. Maybe. But the whole premise of IRV is that you don’t vote based on who has the most chance of winning. If chance-of-winning is still being used for people to decide their 1st-place votes, then the whole concept is failing at a more deep level. I suppose I have to accept that some people like winning and vote for the probable winner just to be on the winning side, though that’s ridiculous.

If you do average score and ignore abstentions, then labeling 0 vs 1 is just a label difference. If you include abstentions in the average or do plain sum of scores, then abstention is a 0 (i.e. less than a 1).

Dishonesty is a much bigger topic than merely strict fraud. Force seems irrelevant. Anyway, it’s perhaps an overly strong word, but the idea is that if we consider a ballot to be an expression of one’s political support for a candidate, one can have a “dishonest” expression (one that is different from their sincere political views). If you don’t think of voting that way but only as a tactical motion, then there’s no honesty or dishonesty. Those who use that language want voting to function more as a real expression of political views.

Actually, IRV is pretty robust in the face of this. If all the kindred voters are aligned, IRV will inherently work out to move them all to one of the group of kindred candidates. The only way the heel wins is if some portion (still a minority perhaps) of the voters who support various kindred candidates actually like the heel enough to put the heel as a 2nd or 3rd choice.

While the IRV folks overplay the bullet-voting issue in score, it’s not entirely wrong. And your claim about the only winning score strategy is wrong.

Say there’s a candidate with majority support. They will lose to a consensus candidate in your strategy. Say candidate A has 55% favorite but 45% hated (polarizing candidate). B is accepted by 100% of voters. B will will if 100% of voters give be a decent score. But the A voters could force A to win by simply bullet-voting for A. That’s the later-no-harm complaint about score. It’s not entirely nonsense. It says that a majority voting block can always get their way by bullet-voting in score, but since you can’t be certain you have a majority, you can hedge toward hoping to get your way by bullet-voting anyway. A plurality block could win with bullet-voting too. So, if your opponent’s block is bullet voting, you may want to bullet vote for their opponent (or min-max, vote 10s for all the possible opponents).

Later-no-harm is oversold and the risk of bullet-voting is not what IRV proponents claim, but it’s non-zero.

That’s why STAR voting is the best system. It has nearly all the benefits of Score but is more robust at reducing any chance of bullet voting. STAR best strategy is more like what you describe.

People want an option for 0. If it convinces more volunteers and voters to support scored methods, it’s worth a little mishap. It’s hardly worth losing that much support and simplicity in exchange for a slightly better method. It’s also a lot simpler to do the math when the minimum is a 0. On top of that, it sounds like part of your dislike of 0 is because you support averaged Score. If that’s the case, keep Majority Denominator averaging in mind: if 5% of voters score a candidate 10/10, we pretend that 51% of voters actually scored the candidate in the average. With 100 voters, the average of that candidate would drop from 10 to 0.9, meaning you’d need closer to a majority highly scoring your candidate, but not necessarily a majority, to win. Also, it would not affect the denominator and averaging of a candidate who was scored by a majority or more in the first place. Under this system, any candidate scored less than a majority receiving a 0 from a new voter would be treated exactly the same as if that new voter hadn’t scored that candidate, because this system essentially “fills in” the nonexistent majority with 0s anyways. This way, the impact of write-in candidates supported by less than a majority receiving 0 largely is… 0.

The problem is that the IRV polls are only showing you 1st place support, and as people see that some candidates are apparently low on support, not only are they giving up on them, they don’t even look enough to realize they like the candidate anymore. If you keep putting someone 1st and they lose, and they don’t grow in polls (which is hard when the support they need is split with other candidates also vying for 1st place support, just like FPTP vote-splitting), then you might not even bother putting them 2nd next time. It’s so difficult to figure out who has the best chances of winning, and the pollsters only have the illuminating space to show you the final round… which conveniently only shows two final contenders. And as people learn that IRV tends to only elect two parties, they will see 3rd parties in the same light FPTP voters do: as curiosities.
To put it in a way that connects to your original point, people will realize any candidate not in the Top 2 has a high chance of elimination, and so might want to shore up another more viable candidate’s chances instead to be in that Top 2, since your 1st place vote is valuable.
With Approval or Score, 1st place and 2nd place support are counted exactly the same. The worst case scenario for these methods is people only voting for their 1st place, making their polling look like IRV’s. But because these methods create strategic incentives to vote for your secondary preferences, they actually make the polling more accurately reflect support, creating a much better cycle of growth in support for good candidates. That’s how I hope it can be advertised, anyways.
Another way of putting it: a candidate with 10% 1st place support and 80% 2nd place support will be seen as only receiving 10% of the vote. While Approval might not show this candidate with 90% of the vote, they would likely have something significantly higher like 30%.

=/ On top of that, it sounds like part of your dislike of 0 is because you support averaged Score. /= said AssetVotingAdvocacy just above.

Well NO! I am not in favor of averaged score. I advocate ‘simple score voting’, which is plain vanilla score voting with the votes simply totaled up, and the candidate who receives the most votes, wins. That’s it. I’m pretty sure that ‘averaged score’ was developed by Warren D. Smith in an effort to allow candidates with limited publicity to win by granting them extra votes via a simple method based on the averaging. It’s a great idea, but it soon led to some odd pathologies, so I rejected it. (I never said I supported averaged score, but it seems Smith started something that people persistently feel compelled to dwell upon.) I regard the STAR voting method as yet another attempt to ‘gild the lily’ of score voting, and while it has some nice features, it just is not simple enough. Worst of all, it makes the hedge strategy more difficult for the voter by causing him or her to have more considerations to think about.

The presence of the zero (“0”) vote is another idea that has long been assumed to be a good feature by Warren D. Smith (he seems to require it to make the ‘averaging’ method function properly). To me, it seems naturally inclined to lead to consideration of ‘negative votes’ (say “-3”). Negative and zero votes seem strange to my mind, but everyone who has read Smith’s works just seems to accept zero votes. I suspect that folks who have not read Smith will, like myself, find the zero vote disturbing. (Smith is great, but unlikely to be right about everything.) Really I would not mind if there were an ‘abstention’ vote, which is just the same as the ‘zero’ vote, except that it sounds more civil. I have argued with NoIRV about this trivial detail far too much!

Approval voting is often viewed from the mathematical perspective as being merely a constrained (or reduced) form of score voting. But it is really a very different animal. The ballot design for an approval system is just inadequately expressive. You cannot possibly use the hedge strategy with such a ballot. This would make it much harder for the common voters to disrupt party lock-in.

It seems clear that employment of the bullet voting strategy with score voting would be just about impossible. Voters in a minority ideological ‘camp’ would only grant votes at all for candidates who support that camp. On the other hand, with 'RCV/‘IRV’, it definitely appears that a small minority could elect a small-minority supported candidate if a non-organized large minority ranked a few favored ‘kindred’ candidates in a random fashion. And any form of organization inevitably leads to the hazards of co-optation. I think this is a very good argument that the 'RCV/‘IRV’ supporters will not be pleased to hear about.

I suppose if simple score voting gains a favorable consensus there will perhaps be less material for the election methods analysts to study. But I believe the common people are more sane than the aristocrats, and am thus driven to follow whatever path will take me to the achievement of real democracy. I am strongly committed to activism.

It almost passed in a county of 300,000 people only a few years after inception, so I wouldn’t write off its simplicity. Further, it seems more likely to pick up support from “majority rule” fans while losing little support from the utilitarian crowd. It may be useful for your activism, if people oppose the radical change of Score.

I suppose it seems odd to me not to include 0 because the whole point of Score is to express your support/liking for each candidate, and not having a “no support” option reduces expressiveness. On top of that, since 0 would be the equivalent of abstention with summed Score, it makes no difference whether everyone scores you 0 or doesn’t score you, so it seems really like you object to the psychology of voters expressing dislike, which I think might actually be a selling point. I remember seeing a Score Voting advocate sell it as (crudely): “You wanna say ‘F*** that guy?’ You can say ‘f*** that guy.’” Your points do make me curious whether voters would prefer a 0 to 10 scale, 0 to 5, 1 to 10, or 1 to 5 in polling though. I do oppose negative votes, however, since there’s simply too many ways it could fail, and it makes reading election results confusing (the winner would likely have <10% of possible points, or might even be in the negatives!)

If everyone votes for the “white hat” (good) candidates, as you call them, and also votes “grey hat” (lesser evil), it only takes a few voters to give crossover support to the other side’s white hats to ensure they win. Failing that, we know there are always passionate bullet voters; a few of those can put the “white hats” over the “grey hats.” Also, polling allows these shifts to take place much before an election.

Do you have real-world examples of this happening? It seems unlikely.

1 Like

At first I thought this site was going to be about election methods analysis. Turns out it’s also about election method advocacy. And it now appears the advocacy is about to become another subject for the analyses. So we can now add to our theorizing the question of ‘what will sell’.

Well, I have suggested before that the voters ought to be able to simply vote for the election method to be employed in the following election cycle. This vote could actually be done using the ‘venerable’ single-selection method. Let the voters decide how they want to vote!

Oh, and here is a fascinating little new election method that will greatly amuse your ‘RCV’/‘IRV’ promoting friends: instant runoff score voting (‘IRSV’). It is extremely simple: you just run a ‘plain vanilla’ simple score election, except that instead of merely declaring the candidate with the greatest number of votes to be the winner, you eliminate the weakest candidates one-by-one until some candidate obtains a 50% + 1 absolute majority. Here is an example:

Let’s say we have candidates “A”, “B”, “C”, “D”, “E”, “F”, and “G”. And listed below are the sums of their scores. (Figures that must be rounded off are treated with ‘banker’s rounding’ – the values are simply rounded off to the closest even integer.)

A <- 11
B <- 6
C <- 20
D <- 18
E <- 12
F <- 5
G <- 2

All in total: 11 + 6 + 20 + 18 + 12 + 5 + 2 = 74

We now eliminate the weakest candidates one-by-one until one of them has an absolute 50% + 1 majority.

All above - G: (74 - 2)/2 + 1 = 37 (Needed for a majority)

All above - F: (74 - 2 - 5)/2 + 1 = 34.5 ~= 34 (Needed for a majority)

All above - B: (74 - 2 - 5 - 6)/2 + 1 = 31.5 ~ = 32 (Needed for a majority)

All above - A: (74 - 2 - 5 - 6 - 11)/2 + 1 = 26 (Needed for a majority)

All above - E: (74 - 2 - 5 - 6 - 11 - 12)/2 + 1 = 20 (Needed for a majority – We have a majority winner now: C has 50% + 1 with 20 votes)

This is obviously the very easiest method!

Seeing as most voters are unaware of other voting methods, they will likely see this as unnecessary, confusing, or risky (such as if some bad/confusing/unknown voting method wins with a mere plurality.) I’d have to say that this is a tougher sell than just focusing people’s attention on one method. It also causes issues for election auditors and clerks, who must be ready at a 2 years’ notice to program the town’s next election for whatever method may win. That is okay with hand-counted ballots perhaps, but as of now, it seems machine counting dominates and will continue to do so.

I can’t see how this wouldn’t just elect the Score winner in every scenario, or even elect someone with less overall support than the Score winner, with no attention paid to whether that results in more support from a majority or just overall less support. Part of the terminology issue here is that you say that a voter gives 0-10 “votes” per candidate; most of us think of it more as giving 0% to 100% “of your vote” to each candidate. From that point of view, whether a candidate has “60% of a vote” from every voter says nothing about whether “60% of the voters like them.” Anyways, if you’re seriously advocating this to “trick” IRV supporters, FairVote would quickly untrick them on this system.

=/ I can’t see how this wouldn’t just elect the Score winner in every scenario, or even elect someone with less overall support than the Score winner […] /= – said above by AssetVotingAdvocacy

It’s not meant to be a practical system in any way, it is intended to demonstrate the intrinsic falseness of the artificial majority concept that is at the heart of ‘RCV’/‘IRV’. It would (clumsily) give exactly the same result as would simply declaring the candidate with the greatest number of votes to be the winner. It would easily be enhanced with a giant SNARK warning, but I wanted to get folks to confront the amazing oddity of how easily the artificial majority concept can be applied to a method that goes absolutely nowhere.

It raises some intriguing mathematical questions, which I would try to solve if I had the time today. The candidates run from “A” to “G”, and “C” has the most votes (“C” has 20 votes). Suppose there is an additional candidate “H” who also has 20 votes? Well that’s a tie, and there’s virtually never a neat solution for that. But what if “H” has 21 votes? Will the system automatically self-correct, or will we have two people with 50% + 1 votes? I will attempt to solve this when I get some time and energy.

I am convinced that the voters should get to choose their voting method. I think it is downright undemocratic to assume that the voters are stupid. At this time, yes, they are profoundly uninformed, but they would be foolish, at this time, to invest in becoming informed, since the existing system gives them no say in any matter. If they find themselves possessing effective input, I think they will make that investment. At some point, having a ‘class’ of esoteric election method theorists who steer the process of reform from atop a sort of ivory tower becomes… just downright undemocratic.

The act of voting has many of the qualities of a casual (not theoretical) game, and people generally enjoy casual games. In fact I would have them vote with two different ballots and two corresponding different systems, and then consecutively (at a later date) go to a runoff to vote on which of the two outcomes should prevail. I’m ordinarily quite loath to promote consecutive runoffs because that asks for a bit too much fuss for voters, but it would most probably be worth the effort for a few election cycles (maybe about six of them) to obtain a method that voters would be satisfied with.

Perhaps a quicker way to illustrate this is to point out that if you eliminate all candidates except the winner in IRV, they end up with 100% of the vote.

That’s a fair point, but maybe we ought to focus when we’re so small to begin with? After all, with Approval, STAR, and maybe even Score entering the mix in the coming years, and PR initiatives hopefully taking off soon, voters themselves can start ballot measures for what they like where they live. It’s not quite the democratic dream you’re proposing, but it can be arrived at gradually. On top of that, I don’t believe, but am not sure, that any current state constitutions provide for multi-choice ballot measures that are binding.

If you look at the homepage of the forum, you’ll see there are separate categories for each. I posted in the Advocacy category. There are dedicated categories for theory. I wanted to focus on how to pitch and explain an issue rather than discuss the reality of the issue, though I guess it all blurs together sometimes.

This is really tangential now. This is where you should use the reply-as-new-topic or just start a new topic. Please don’t hijack topics by just replying with whatever tangential thoughts.

That goes for everyone in general: please try to stay on topic more. My point about a good way to help people understand the basic IRV flaws is totally lost in this exchange here. Please make more effort to split new topics off as needed and not derail things with so much unrelated discussion.

Polls show 40% Red, 25% Orange, 35% Yellow, with Orange supporters backing Red more than Yellow. As a Yellow supporter, I know my candidate is doomed. So, I vote Orange > Yellow > Red and hope for the best.

In fact, it is the Red supporters who should be considering voting Yellow > Red > Orange, hoping to knock out Orange first and hope enough voters transfer.

If the margin of defeat for Yellow vs. Red is low enough, you might take the risk and not compromise, but overall fair point.

My point here is that if too many Red voters put Yellow first, they will not only help Yellow survive, but their vote will help Yellow defeat Red in the final runoff. They’d have to really thread the needle here, which seems unlikely. I think in the end, Red voters would stick to their honest ordering, because any strategy when they’re so close to victory could backfire badly. Also, it is a pity we don’t have any examples of voters trying these kinds of tactics in response to polling and learning strategy.

Hey, I just got through saying:
=/ It’s [instant runoff score voting (‘IRSV’)] not meant to be a practical system in any way, it is intended to demonstrate the intrinsic falseness of the artificial majority concept that is at the heart of ‘RCV’/‘IRV’. /=

You just said:
=/ This is really tangential now. /=

You began with:
=/ Some preferences get counted, others ignored, and that simple fact makes it far less democrat and fair. If I really care about my 2nd choice getting counted, I need to strategically put that 1st (favorite betrayal), and then we’re back to lesser-evil voting!

That should make it clear, and again, every single pathology of RCV can be derived from this fact. /=

Well, I agree with your first paragraph there. I simply do not know if the second paragraph is accurate or not.

There is something astoundingly strange about ‘RCV’/‘IRV’ that I don’t really understand. It has pathologies that pop up in astonishingly varied ways! Are they really deeply connected (as you assert)? That is quite mysterious indeed.

This is why I have been advocating the most simple possible systems. The issues here seem to involve the phenomenon of ‘complex ramifications’ – things that start with simple concepts, and nonetheless go on to endless complexity.

Maybe your assertion about how the fact that some significant voter input gets ignored is central to this issue. But I have many questions about the many mysteries of it.

I certainly had no intention of derailing your thread. I just believe the questions about the strangeness of ‘RCV’/‘IRV’ pathologies are just astonishingly profound.

It seems to me the significant pathologies of IRV are at least somewhat connected to its desire to optimize for “the majority”, rather than everyone. This doesn’t make sense when individual issues may have very different majorities in favor of them, and naturally leads to bipolarization and the search for the “true majority” which doesn’t really exist. Condorcet’s improvement is trying to find the majority-supported candidate most supported by the minority, which is why it doesn’t have as many problems, if I understand it correctly.
Maybe this can be part of the explanation to IRV supporters, that IRV’s majority fails to acknowledge individual issues and their importance?

1 Like