Hacker News Reader

Comment by crazygringo | Hacker News Reader

I absolutely support ranked-choice voting as an improvement to the current system.

BUT, the proposed system is specifically Instant Run-Off Voting (IRV), which does not make a centrist third-party candidate likely to win. Because it removes whoever gets the least first-place votes, the only improvement it gives is that a third-party candidate doesn't spoil an election, because your second-place vote will still count. (E.g. Nader wouldn't have taken votes from Gore, so Gore would have won instead of Bush.) But you're still stuck with one of the polarizing two-party candidates who will win.

Contrast this to the Borda Count [2], where points are assigned by rank. In this case, suppose about half of everyone votes first-place for an extreme liberal, and the other half vote first-place for an extreme conservative. But everyone votes second-place for a moderate centrist they can live with, and third-place for the opposite-party candidate they detest. With IRV, the centrist is ignored and one of the extremists will win. But with Borda, the centrist candidate will win.

So to really get away from the poisonous political polarization of our times, ranked choice is necessary, but it needs to be Borda or similar, not IRV. Still -- it's a start, and I'm grateful for that alone.

[1] https://en.wikipedia.org/wiki/Instant-runoff_voting

[2] https://en.wikipedia.org/wiki/Borda_count

View on HN

But also it gives voters the option to actually vote for whom they want to vote for instead of who they think will win. Big game changer and will fundamentally affect everything about the vote.

Big vision here is that it also shifts the discussion to how can we make the voting system better, just by implementing _something_ better.

I agree with you hugely that "it's a start, and I'm grateful for that alone". And we need to push this across the finish line, even though it isn't _perfect_. We just need _better_ right now, we need momentum.

> But also it gives voters the option to actually vote for whom they want to vote for instead of who they think will win.

You already have the option to vote for who you want to vote for in "first past the post", you just might get an outcome you don't like very much...

IRV and every other voting system is the same way: you can vote your heart, but you'll get an outcome you prefer more often if you vote tactically. There's even a theorem: https://en.wikipedia.org/wiki/Gibbard–Satterthwaite_theorem

For the specific case of IRV, consider an election like:

* 32% of people: Clinton, Sanders, Trump

* 33% of people: Sanders, Trump, Clinton

* 35% of people: Trump, Clinton, Sanders

Now, this is a terrible election and there's no clear winner because people's preferences point around in a circle. Under traditional voting Trump wins, while under IRV Clinton is eliminated and Sanders wins. But if a few people preferring Trump > Clinton > Sanders had instead voted just Clinton > Trump > Sanders we could have had:

* 32% of people: Clinton, Sanders, Trump

* 33% of people: Sanders, Trump, Clinton

* 32% of people: Trump, Clinton, Sanders

* 3% of people: Clinton, Trump, Sanders

Then Trump would have been eliminated first, with only 32% of the first place votes, and his second place votes would have gone to Clinton, making her win. So by voting for a candidate they liked less, these voters got an outcome they liked more.

I'm not at all a fan of G-S and the Arrow Theorem, since they make widely unrealistic assumptions about voting (e.g. Arrow's IIA is nonsense - intensity matters!).

G-S and Arrow damaged tremendously any attempt to look at electoral systems rationally by imposing a mathematical description which is simply irrelevant to actual concerns.

In your given example, the particular unrealistic assumption is the way it looks at voting. Voting is not an extended opinion poll. It is use of political power where each voter is given an equal amount. Therefore, there's no need to reduce "tactical voting" to a minimum (the useful requirements are far laxer than that).

The voters in your example did not "fail" or "cheat" by voting for a candidate they liked less. They compromised, and therefore got a better result for them. Arguably a voting system that pushes people into consensus and majority is actually desirable and not a flaw.

Aside, I do prefer Borda or Approval over IRV, but the better reasoning is just they are far simpler overall while preserving the actually important advantages.

Indeed. I'd be in favour of people who know Arrow's Theorum designing the voting system; but not because I care about tactical voting.

I rank the voting systems [Anarchy] -> [Dictatorship] -> [First Past the Post] -> [IRV] -> [Approval].

Anarchy is sort-of the base case. Dictatorship is better than anarchy because at least it is large-scale organisation and infrastructure might happen. FPTP is that and the dictatorship can't be truly unpopular (<35% general support). IRV adds in a really stabilising ability for 3rd parties to successfully build support with a slow rolling campaign - makes it really hard for major parties to move away from the centre because people have a powerful signal for what they want. Approval adds in a level of simplicity to IRV.

I like IRV for the same arguable reason you like tactical voting - a good system should really be a bit random in tight situations. 50.1% of the votes isn't really different from 49.9% and voting systems don't need to differentiate. In Australia (IRV) when we had our last really tight election a bloke got elected on a couple of hundred primary votes from the Australian Motoring Enthusiast party. His platform was basically that he liked souped-up fast cars. Nobody had ever heard of him. Dude was an excellent Senator, he read the legislation and thought about things. Big win for Australia, IMO. Better than some political swamp-creature elected on a tiny margin.

Also; my favourite voting link: http://zesty.ca/voting/sim/

EDIT I think the order is obvious, but I upgraded my '>' to '->' for tmporter.

Your ranking list can be interpreted in two very different ways depending on whether you consider ">" to be an arrow or a greater-than sign.

Nitpicking, but Ricky Muir (the car guy) was elected in the Senate, which does not really have instant runoff in the way described here. (For outsiders: our House of Representatives is IRV, our Senate is Proportional Representation which is like IRV but with multiple winners on each ballot)

Agree with you that electing a random citizen gave much better results than party slime.

The Old senate system cannot fairly be called STV in its effects not in the options available to voters. Above the line voting with no (voter controlled) preferences made in more like a list.

In addition to the simplicity issue, IRV can have a practical problem of legitimacy, when the elected official is everyone's 10th choice (I exaggerate for effect).

There are some arguments for allowing that, but I think the argument against is stronger - officials need legitimacy to serve effectively, otherwise we have a period of political instability which will end badly, and 10th choices won't practically have legitimacy. In Approval voting the preference order is a bit hidden so this doesn't happen. One of those cases where treating an election as if it's not an opinion poll at all is the better choice.

If the preferences have run off to 10th place; the first 9 choices are controversial enough they wouldn't have legitimacy either.

Look at America - the president has so gained so much legitimacy from FPTP that the House is trying to eject him from office. Voting system doesn't help with that much.

That is an argument like the Chinese claiming democracy wouldn't work because China is different. In theory it can't be instantly debunked, but in practice it seems to be a non-issue (compared to alternatives) when IRV is tried.

What's happening in America is not due to the election result, not anymore. That has more to do with the President's... temperament, and some very particular circumstances.

If it was just FPTP where he won more electors but less votes, he would have been seen as legitimate by now, just like Bush was.

I disagree. Polling in 2016 clearly showed Trump and Clinton were each individually the two least-liked nominees in their parties in modern presidential history. First-past-the-post and bitter primaries were largely responsible for the result-- over a third of voters remained at home on Election Day 2016. In some electoral systems, they even have an option to reject all nominees in these kinds of situations.

> over a third of voters remained at home on Election Day 2016

Isn't that an unusually high turnout?

I thought about the general elections. As for the primaries:

* Clinton locked out nearly all of the viable opponents by using institutional control, the only one left was one sufficiently outside the Democratic party so she had a strong institutional advantage. No voting system could have fixed that.

* Trump actually had some popularity as a second choice among Republican primary voters, and we see that later on when only Cruz and Kasich were left. The other candidates' failure were more mundane - Not taking Trump seriously, op research failure, Trump being able to coordinate with Christi while the others were less able to coordinate, etc.

* The option to reject all nominees.. is interesting, but can lead to a 2nd/3rd ballot while the government is semi-paralyzed. It may not be a good idea to allow that.

> Clinton locked out nearly all of the viable opponents by using institutional control, the only one left was one sufficiently outside the Democratic party so she had a strong institutional advantage. No voting system could have fixed that.

Uh, yeah, simply not having voting superdelegates in the nominating contest (a reform to the candidate-selection voting system Democrats made in response to 2016) would have likely fixed (or at least mitigated) that, since the nominating contest was close even with the early superdelegate commitments and the effect that had on the perception of inevitability.

A general election direct (no electoral college) voting system like Bucklin or IRV, modified some that the same ballots, skipping votes for the winner, were tallied again by the same method to select the vice president, encouraging a party to bring it's two independently strongest candidates into the general election (and increasing the space for other parties or independent candidates) would absolutely both discourage that and limit the effect it would have on constraining viable general election choices.

(The same system internally to the party for choosing the Presidential and Vice Presidential nominees also would fix it.)

The superdelegates went for Clinton because there wasn't any serious institutional D challenger (Sanders was an independent before), so Clinton had by default a huge advantage there. If there was a serious internal D challenger, the picture would have been far more equal.

IRV can make the contest more fair than the previous voting system, but it can't make candidates run...

> The superdelegates went for Clinton because there wasn't any serious institutional D challenger

No, that's s almost exactly backwards. A key way Clinton locked out other traditional candidates was to secure an unusually large number of superdelegate commitments extremely early, needle other candidates would normally commit to the race.

> IRV can make the contest more fair than the previous voting system, but it can't make candidates run...

Yes, it can. Running in a Presidential primary (and the same is true of many other races) is an investment of time and resources people make—or avoid—in part because of the perceived prospects for success factors like a competitor having a substantial share of the total available vote sewed up before you decide, or a voting system that naturally narrows the field people will actively consider during the campaign do, very much, effect who decides to run.

> No voting system could have fixed that.

Literally the point of IRV, Approval Voting and so forth is allowing viable 3rd parties not endorsed by the official channels of the major parties. They fix precisely this sort of problem. Democrats could have voted for a different left-leaning candidate without advantaging the right wing of politics.

There wasn't a coordination issue between 100 small left-leaning candidates for the Clinton opponents to choose - in that scenario IRV/AV would have helped. In the real scenario, the problem was solved for them by having only 1 option...

The candidate who is everyone's 10th choice can't win. To make it through the first exclusion the candidate has to be more people's first choice than at least one other candidate.

In practice the most outlandish result you get is someone winning from third on the primary (first choice) votes (eg https://results.ecq.qld.gov.au/elections/state/State2017/res... or https://results.aec.gov.au/15508/Website/HouseDivisionFirstP... ), but such a candidate still needs a considerable amount of first choice votes to be in the running (and of course a very strong flow of second/subsequent choice votes).

The legitimacy argument of IRV is that the winning candidate needed to accrue 50%+1 of first-and-subsequent choice votes, so they are at least the preferred winner over the second-place candidate by a majority of the electorate.

I did exaggerate (and said I did), but my point was that IRV makes it very obvious when a candidate loses badly on 1st order preferences but wins based on subsequent choices.

That can be argued as an IRV benefit over other systems (the winner knows s/he's on thin ice! The winner does have 50%+1 overall!), but the other way to see it is that the winner would be hobbled and would have difficulty governing effectively, and a voting system that emphasizes the ranking a bit less might have had the same result but with the winner having more legitimacy following the election.

Ricky Muir was not elected on IRV. He was elected on an exhaustive flexible list. Voting above the line without (voter controlled) preferences in the old system, which is what got him elected, has none of the characteristics of IRV, and was basically like a strange list that required all candidates to be on all lists.

> Arrow's IIA is nonsense - intensity matters!

Independence of irrelevant alternatives (IIA) says that adding another option shouldn't change a voter's (or the election result's) relative preferences among all other options.

"The table's waiter comes by and says their desert options are blueberry pie and vanilla ice cream. You order the pie. Then the waiter comes back and says, we also have chocolate cake. 'In that case, I'll have the ice cream!'"

I don't see what this has to do with intensity?

Arrow's theorem does assume that intensity doesn't matter, but this isn't IIA but is rather the unnamed predicate that voters express their preferences by ranking candidates with no ties allowed, rather than an approval voting scheme, or a "on a scale of 1-10" scheme.

Your example only looks irrational when one looks only at the overall preference list and furthermore thinks that the only important result is the winner. i.e. when looking at elections as an extended opinion poll and applying a purely mathematical logic ignoring politics, just like the Arrow Theorem does.

If a voter went for chocolate cake despite polls showing that this may cause blueberry pie to lose the election, maybe the voter does not care that much for blueberry pie over vanilla ice cream, or the voter really really cares for cake far more than the other choice - both of these are intensity issues and the election method de facto taking them into account is not irrational.

Maybe the cake was the only nut free option, and some voters deciding for the cake despite not having nut allergy themselves and despite losing convince the restaurant to offer more nut free foods later.

99% of the 'irrelevant choice' cases actually have some political logic like the above behind them which Arrow theorem just blithely ignores. The 1% are usually some attempt to mislead the voters (say by running a hopeless candidate with a similar name to another in order to siphon off votes), and should be handled by the electoral commission.

One thing to consider is that because gaming the vote is harder to understand, it may have less of an impact. People can easily understand that voting for 3rd party candidate is not voting for a candidate that actually has a chance. Fewer people will be able to understand the second example you give, meaning it will have less impact on people gaming their votes.

But you also have to consider that most people don't understand it because it's currently not relevant to them. If the system was put in place, I would imagine people who hear a lot more about that strategy.

This is a good point I think. If this system were adopted, all it would take is a small number of activist people to understand how to game the system, then post a bunch of YouTube videos explaining to all their acolytes how they can join in and game the system.

The lived experience of IRV is that this does not happen.

There are probably a few reasons for this. Firstly, the situations where tactical voting can be of benefit tend to arise only rarely. Secondly, when these situations do arise, it is not usually apparent in advance of the election. Thirdly, even when such a situation arises and it is reasonably forseeable in advance of the election, it is easy for the tactical voting to backfire and hinder the candidates chances rather than helping them, if the vote shares/exclusion order aren't as predicted.

>The lived experience of IRV is that this does not happen.

What "lived experience"? IRV has never been used in American elections at least, unless you count some towns.

It's been used for single-member constituencies at the Federal level in Australia for 100 years, and in the various States for at least 50 years. French Presidential elections have been using run-off elections for even longer, and those are in theory open to the same kinds of tactical voting in the early rounds as IRV.

People not understanding how their vote affects the outcome is usually considered a downside.

Every single voting system has some element of fault. Very few voting systems are close to "perfect representation" and they tend to be more math heavy, in a way that would lead to distrust from the average voter.

IRV might have some corner cases, such as the one you listed, where strategic voting is beneficial for some. IRV reduces that element drastically, and encourages positive voting and true representation to a much larger degree.

Essentially, everything the NYT article says.

> There's even a theorem

Calling it the "same" is missing the point. It's a corner case in instant runoff, while in first past the post it's just normal voting.

IRV failing to capture voter preferences is not a "corner case". It is absolutely unclear that IRV is better at all than FPTP. See a set of voting simulations with candidates and distributions here for different voting methods:

http://zesty.ca/voting/sim/

Alarmingly, IRV is the worst voting system of all just by eyeballing these simulations. The charts it produces are bizarre, and it tends to select for extremists and/or candidates that do not capture the voter's preferences!

It's not just of theoretical interest either, IRV famously spoiled the 2009 Burlington mayoral election, resulting in its repeal: https://en.wikipedia.org/wiki/2009_Burlington_mayoral_electi...

The only good takeaway from IRV picking up steam is that maybe it puts us on the path towards something like approval voting, which is much harder to game, and is much easier for voters to understand (I doubt more than 1 out of 10 people could explain how IRV works).

As someone who voted in the 2009 election I disagree that IRV actually spoiled the election, people - when voting in an IRV system, need to comprehend that they don't need to rank all candidates and doing an exhaustive ranking is a tactical decision to vote against an ultimate candidate.

That "spoilage" ended up being less about IRV and more about the fact that Bob Kiss was someone a lot of people ranked without good knowledge of him was more at issue, along with his personal character. Don't forget that he wasn't some 5th place follower who managed to pick up protest votes, he was the 2nd place candidate in the primary round.

I just wanted to add, the fact that IRV was repealed over this removal from office was a travesty and the result of shady politics - please understand this is a terrible case study.

IRV picking up steam might also have negative effects towards people's general opinion of alternative voting systems once they see all the strange outcomes it can have.

But I think that the positives attention around it brings outweigh the negatives. The first (recent) overhaul to a voting system is more difficult than subsequent changes.

Voter Satisfaction Efficiency (http://electionscience.github.io/vse-sim/VSE/) is another way of looking at different options, and IRV compares very poorly by it

I hate the way they've defined their metric:

> VSE is expressed as a percentage. A voting method which could read voters minds and always pick the candidate that would lead to the highest average happiness would have a VSE of 100%. A method which picked a candidate completely at random would have a VSE of 0%. In theory, VSEs down to negative 100% would be possible if a voting method did worse than a random pick

If VSE is going to be a percentage, it should be a percentage of the maximum happiness -- that is, the total happiness if everyone got their first choice, not the total happiness that is the highest any candidate can actually achieve. I say this mostly because it's much easier to calculate.

But more fundamentally, there is no reason to expect the worst-case candidate to be equally as bad as the best-case candidate is good. It's a huge mistake to rank systems on a scale from +100% to -100% where the units on the positive and negative "halves" of the scale are different!

And it's also a mistake to define 0% as whatever sortition achieves. It's trivial to construct a pathological system in which 0% VSE is equal to 100% VSE. (Everyone is equally un/happy with all candidates.) This should be impossible.

Let's consider an example:

You have 100 candidates. For 99 of those candidates, the entire electorate agrees that any one of these interchangeable guys would do a fine job. Nobody prefers any of them to any other.

The last candidate is Mao Zedong. Under his leadership, the country's industrial and agricultural bases will be systematically destroyed, anyone who owns a rental property will be executed, and tens of millions of people will starve to death for no reason.

The electorate agrees that Mao is an undesirable choice.

So: any method that can't pick Mao achieves a VSE of 100%.

Sortition, by definition, achieves a VSE of 0%. But it's still pretty good -- there's a 99% chance of selecting one of the good guys, and a 1% chance of epic, country-shattering disaster.

A method that is guaranteed to select Mao has a VSE of, I assume, -100%. But it is much, much worse than sortition (the notional midpoint, 0%), while the VSE of 100% is only moderately better than sortition.

These numbers don't make sense.

That is very interesting and I hadn't seen it before. IRV still does better than plurality by quite a bit, although it doesn't do well compared to the rest. As the chart shows, one of the biggest outliers is 100% honest voting with plurality.

I hadn't heard of 3-2-1 voting before but that is interesting. Maybe it would be one of the best. I generally prefer ranking to scoring (I'm not sure how much better someone prefers a candidate should really be part of the process) but the simpler tallying and presentation of results is nice. I also like that in the "scenereo type" breakdown the easy case gets the best results, condorcet cycles get the worst results, and the rest are in the middle (all with little effect of strategy). It seems like the right distribution to me, but none of the other methods measured get that distribution. IMO a ranked preference is usually much easier to determine how to vote (although it can still be an issue when some candidates are much closer than others), although I guess the limited effect of strategic voting should hopefully mean that it usualy doesn't make much difference exactly how you vote. OTOH, at least the names of the options should be changed; it is the rare race where I could describe any of the candidates as anything other than "bad".

https://electowiki.org/wiki/3-2-1_voting

IMO, proportional representation would be a much better change of voting system rather than just changing the method. Voting for one person is almost always going to leave lots of people unrepresented.

This particular example is a bit of a corner case, as it has no Condorcet winner. This is rare: https://en.m.wikipedia.org/wiki/Condorcet_paradox#Likelihood...

IRV still has serious failure modes in its typical deployment, but this is not a great example.

The Wikipedia article for this particular example says Andy Montroll, the Democratic candidate, was the Condorcet winner.

Gah, too many examples. I was referring to the synthetic example further upthread.

There should be a rule - no majority = no win.

Let alone - anything is better, than a simple plurality based system.

Very often when there are more than two candidates, none of them have a majority of supporters. In fact, sometimes that's true even with two or even one candidate...

What do you do when votes lead to a plurality?

What election are you thinking of where majority of votes didn’t win?

Because if you say 2016 / electoral college, I’ll have to remind you that the winner had a fairly vast majority. If you are thinking Canada or some ranked system, iirc they still elect by majority.

“No majority no win” is a great recipe for tyranny of the majority and why we have representative democracies, isn’t it?

> What election are you thinking of where majority of votes didn’t win?

https://en.wikipedia.org/wiki/United_States_presidential_ele...

> Because if you say 2016 / electoral college, I’ll have to remind you that the winner had a fairly vast majority

If you mean the winner who lost, sure?

The winner of the 2016 election received the majority of the votes cast in the electoral college by Presidential electors

Those electors voting for the winner were themselves, however, elected by only a minority of those voting for Presidential electors in elections which misleading had the name of Presidential candidates but not electors on the ballots.

This is pointlessly pedantic. The person clearly meant they preferred elections be won by a majority of the popular vote.

In the USA we don't use popular vote. We use state votes. The winner won with a vast majority of those state votes.

You can be upset at the outcome, but you can not argue the facts of the system.

I wasn't arguing the facts of the system. I wasn't even stating a preference (though, yes, I'm upset at the outcome). I was responding to a thread where:

1. Person states preference that majority wins 2. Person states that majority did win 3. I demonstrate that majority did not win 4. Person #2 says majority won EC 5. I point out that it was plainly obvious that person #1 would prefer a majority of the popular vote win

Canada uses FPTP voting and it has highly skewed the last few elections, sadly there was an expectation that the last federal election would result in election reform but that never materialized.

The fundamental reason for this: IRV doesn't care what your preferences are other than your top choice, until your top choice is eliminated.

Until choice A is eliminated, votes for A>B>C and votes for A>C>B are treated identically, and neither one will affect whether B or C gets eliminated first.

Better systems take all preferences into account from the beginning.

> there's no clear winner because people's preferences point around in a circle.

Is there evidence around how often such cycles actually occur in practice? Seems like a convoluted example, intuitively speaking.

When voters are all arranged on a single primary axis it doesn't happen, but when voters care about multiple axes at once (ex: socially liberal / conservative + economically liberal / conservative) it's relatively common.

One well-known and widely discussed case is https://en.wikipedia.org/wiki/2009_Burlington_mayoral_electi...

This particular case was brought up previously, but as a participant I'd like to mention that this election actually went quite well, the candidate with the second most primary round votes ended up winning. I believe it's often cited and discussed because the election was repealed and the candidate thrown out of office, which was hugely frustrating... the voting system was scrapped because the electorate selected a bad candidate.

Let me draw a parallel, let's say that Trump is impeached - when he's impeached should we all say "Oh hey, well that democracy thing... what a dumpster fire - let's go back to a monarchy!". No, we shouldn't, but high amounts of pressure from the dominant parties were deployed and caused a voting system to be repealed just because someone elected by it was removed from office.

Seriously, the BTV election is a terrible case example, it was intensely tainted by lobbied interests and used as an excuse to regress voting rights.

A (Constitutional) Monarchy is better _because_ it has no legitimacy. The purpose of the Monarch is not to Rule but only to provide a Figurehead, a living Symbol of their country.

That's what Liz is for and she's really good at it.

If the US had an elected Figurehead president, Trump would have been a perfectly functional drop-in. Likes gaudy things, talks bullshit, eats too much fast food - symbol of America, works for me.

The problem is that the US combines the Figurehead role with Executive leadership. That's crazy, there aren't going to be many suitable candidates for either role, and now you're asking the electorate to vote for a single individual to do both, no surprises the results aren't good.

For groups of individuals, it is quite possible https://en.wikipedia.org/wiki/Condorcet_paradox

It's obvious that these cycles can exist when the things referred to have labels like "A, B, C" or "carrot, potato, lettuce", but how often do they occur for political candidates like the ones described above?

I'm willing to buy that IRV is objectively not suitable for picking meals, but that in practice it is suitable for picking presidential candidates -- or that it's always inferior. Seems like an empirical question, not a theoretical one.

The wikipedia page includes a cite of such an empirical analysis: https://en.m.wikipedia.org/wiki/Condorcet_paradox#Empirical_...

Thanks! I'm just going to quote here because the footnote was interesting:

> A summary of 37 individual studies, covering a total of 265 real-world elections, large and small, found 25 instances of a Condorcet paradox, for a total likelihood of 9.4%[14]:325 (and this may be a high estimate, since cases of the paradox are more likely to be reported on than cases without).[13]:47

> 13: ""most election results do not correspond to anything like any of DC, IC, IAC or MC ... empirical studies ... indicate that some of the most common paradoxes are relatively unlikely to be observed in actual elections. ... it is easily concluded that Condorcet’s Paradox should very rarely be observed in any real elections on a small number of candidates with large electorates, as long as voters’ preferences reflect any reasonable degree of group mutual coherence"

I think I have to file this one under "it's complicated, must dig more".

Thankfully lots of smart people have written papers about how that's incorrect. They just use generic terms like "A,B,C"

Yep, it is flawed and, conversely, can be gamed with unintuitive choices. Yet, it is an improvement nonetheless just because fewer people get an outcome they liked less.

i know very little about the various methods of voting and wanted to learn something about it here. while i’m not a republican (i’m independent) it’s very hard to take you seriously when your post reads of anti-trump. if people really want voting reform the reason needs to be something other than stopping trump. otherwise it seems like an attempt to game the system and i’m less likely to vote for reforms. granted it looks like you tried to make it neutral but the last paragraph kinda ruined it

What part of that post is anti-Trump?

How is the post anti-Trump?

Voting tactically will give a false impression of endorsement. If all the third party voters voted for Hillary, it would send a hugely wrong message about what the people actually want.

Dude... In the two cases you list the least objectionable candidate won. What is your problem? Do you not understand how you rank your options?

Could you clarify? I'm not sure what you're objecting to.

> We just need _better_ right now, we need momentum.

Agreed. But we shouldn't be relying on local governments to provide that momentum. I think the people in the best position to legitimize improved voting systems are the pollsters. If they allowed the people they poll to rank or rate all the candidates, in the Democratic primary for example, for whom they have an opinion, then I'm pretty sure the political media (always desperate to fill air-time right?) would start reporting how the candidates would be fairing under different voting systems (that more accurately reflect their inherent popularity). Once these voting systems become familiar in their media, I think the public would be much more ready to accept them at the ballot box.

> it gives voters the option to actually vote for whom they want to vote for instead of who they think will win.

They can already do that now. The idea of voting for the “winner” is just ridiculous. Vote for who you want.

Totally with you on supoorting ranked-choice voting. It seems like a clear and obvious upgrade over what we're doing now in ways that can be mathematically verified.

Question though - why prefer the Bourda Count over a Condorcet method[1] like the Schulze method[2]?

I recently went down the voting method rabbit-hole when recreationally overengineering the procedure for selecting a fun offsite activity for my team. I picked the Schulze method basically because it satisfies more[3] of the formal criteria that seemed important, including the Condorcet criterion, which seems especially important.

The Bourda Count method doesn't[4] seem to satisfy the Condorcet criterion, so it's possible that a candidate that would win every head to head match up would lose the election. Also, like most (all?) methods that allocate variable numbers of points to candidates, it's subject to tactical voting[5], i.e. it might be rational not to allocate any points to your second place candidate in order to maximize the margin between your first place and second place candidates, even if you prefer your second place candidate to several others.

1 - https://en.m.wikipedia.org/wiki/Condorcet_method

2 - https://en.m.wikipedia.org/wiki/Schulze_method

3 - https://en.m.wikipedia.org/wiki/Schulze_method#Comparison_ta...

4 - https://en.wikipedia.org/wiki/Borda_count#Evaluation_by_crit...

5 - https://en.m.wikipedia.org/wiki/Borda_count#Potential_for_ta...

Every “Game Form” (which is like a “Game” in the Game Theory sense, except instead of having the results be utilities for each outcome, it just has labels for the outcomes) which is “straightforward” (meaning, for each participant, what utility that participant alone assigns to each outcome label (not depending on the utilities anyone else assigns) is sufficient to determine what choices they should make to maximize their expected utility) must be either a simple vote between exactly two options (as in, everyone votes for one of the two, and an outcome label happens if it gets enough votes), or dictatorial (a single person picks an outcome), or a probability mixture of games of that form (for example, pick a random voter’s ballot, and pick whoever they picked. Or, another example, pick a random pair of candidates, and hold a vote between the two).

That is to say : any deterministic voting system with more than 2 candidates, and where no voter is a dictator, is subject to tactical voting.

This is Gibbard's 1978 theorem.

Though, I wonder, are there game forms which are not “straightforward” but which are a Pareto improvement (in the sense of “regardless of their assignment of utilities to the possible outcomes, every voter’s expected utility is at least as high as it would in the alternate option) over game forms that are straightforward?

I think probably yes? If each voter puts their utilities, and then the system computes each voter’s utility under random ballot voting, and then checks whether any probability distribution over candidates would result in a strict Pareto improvement on the expected utilities of the voters, and if there is, pucks one, and if not, defaults to the distribution from “pick a randomly selected voter’s first choice”,

Well, we know that this wouldn’t be a straightforward game form, because it isn’t (equivalent to) the form required by Gibbard’s theorem, So, strategic misrepresentation of one’s (normalized vNM) utilities must be sometimes useful in it (because if it were not, then one would only need to put the same utilities each time, making it a straightforward game), but perhaps the degree of misrepresentation that could be useful would always be small enough that it would never select an option as being a Pareto improvement of the reported utilities from the random ballot method, unless it was also an actual Pareto improvement over it, using the true utility assignments?

But, I’m really not sure.

Another thing that I think is relevant is the "revelation principle", which says that any "social choice function" that can be implemented by some mechanism, can be implemented by a truthful social choice mechanism.

But it seems clear that the thing that I proposed as being an improvement on random ballot can't count as a mechanism which implements a "social choice function", because there can't be an equivalent mechanism which is truthful (... well, I assume that no probability mix of pairwise votes and random ballot stuff is equivalent to the thing I described. Can't see how it could be.).

And I think the reason why is that the thing that I described (or, any fully specified version of it) would have to have multiple Nash equilibria.

So, for it to make sense to call it a Pareto improvement, I think would need to have that _all_ of the Nash equilibria would be Pareto improvements on the thing.

I am now less confident than I was that such an improvement exists.

One way to think about the Borda is that it gives you the least hated candidate, instead of the most liked.

From Wikipedia: The Borda count is intended to elect broadly-acceptable options or candidates, rather than those preferred by a majority, and so is often described as a consensus-based voting system rather than a majoritarian one.

I don't think it works though.

Unless you are careful, partisan voters will likely vote their candidate maximum and other candidates minimum - this is called bullet voting on Wikipedia.

Non-partisan voters will likely vote genuinely.

Since bullet voting is the dominant strategy, the election essentially reduces to First Past the Post and contains all the bad (majoritarian) incentives that FPTP does.

Most people aren't completely partisan, when given multiple palatable choices. This is less relevant in the US, with its two-party system, but is quite relevant outside of it.

My concern is that if even a small number of people bullet vote, other people will see that their votes have less impact and adjust accordingly, in a self-perpetuating cycle.

Exactly. And (since all voting methods are flawed in some direction), the question is what do you prioritize in elections: the will of the majority, or the "acceptable" choice of even more people.

Personally, I think Borda is terrible.

Let's suppose the 2016 election was: Clinton, Trump, some milquetoast Republican like Jeb Bush, and someone everyone hated, bringing to total to 4.

Suppose 73% of the population were fanatical Clinton fans. However, they didn't want Trump to win, so they voted Jeb #2, even though they'd much rather he didn't win. The 27% of Trump voters also listed Jeb #2.

Jeb would win, in this case, even though he was a distant second choice for 73% of the population.

(100 * 3 points > 73 * 4 points).

Even worse would be if there were 5 or 6 candidates. Now every Clinton #1 vote is worth 6, and every Jeb #2 vote is worth 5. Jeb would win even if 83% of the voters were fanatical Clinton fans.

There's no way to distinguish your second vote as "this person is also great" vs "this person is the least worst alternative." The only way to not let your far-distant second choice win is to not vote for them at all, or recruit extra candidates as shields for your preferred candidate. But then we're bringing in strategic voting and all sorts of other messiness.

This seems perfect for the US, and they could even pitch it as such. Although you do start to open the game theory floodgate once you veer beyond dead simple voting systems, which could turn into a gerrymandering game if not carefully restricted.

> all?

All: https://en.wikipedia.org/wiki/Gibbard–Satterthwaite_theorem

I missed your choice because I can't ever remember how to spell Schulze so I always look it up by Path Vote (my post beneath mentions both names after I found the Wikipeida article again).

Seems good, but perhaps a bit difficult to explain to the average voter?

Condorcet isn't particularly hard to explain. Easier to explain than IRV anyway. Call it "Instant Round Robin" - if one candidate would beat every other candidate head-to-head, then that candidate should win.

The annoying part about Condorcet enthusiasts is that they insist on lumping the uncommon "tie-breaking" methods (i.e. when there is the rare loop) in with the above general definition. That's when people start talking about Schulz, Ranked-Pairs, etc etc. I personally think a loop is valuable and shouldn't be tie-breaked - it's an indication that the voters are not ready to decide and that there should be another round of debate among the finalists before revoting.

In my opinion, the main flaw with Condorcet is that it is possible to have a Condorcet Winner that does not have the "approval" of a majority of voters.

>another round of debate among the finalists before revoting

What do you do if revoting doesn't break the deadlock? Also, just in general, I think a lot of people don't want a system that involves potential revoting.

You get the current BRexit situation, and we'll hopefully soon see some kind of historical test case the refer to...

> Call it "Instant Round Robin" - if one candidate would beat every other candidate head-to-head, then that candidate should win.

And 90% of peoples eyes just glazed over, all to fix some almost entirely theoretical flaws with IRV, that's why condorcet isn't better in the real world.

90% of which people, the people that are enamored by "count up all first place votes, then find the person in last place, then go look at everyone's ballots and remove that person from everyone's ballots, and then count up all first place votes again, and then repeat that process until you're left with one person and then that candidate should win"?

If you're number 1 choice is knocked out you're vote goes to you're number 2 choice, if they're knocked out it goes to number 3... It's not about understanding the intricacies of how the count works but understanding what their individual vote will be doing. When it comes to what you're individual vote is doing Condorcet is much more of a black box.

Isn't quadratic voting both simpler and less prone to tactical voting?

I don't think Borda count is an acceptable alternative. It's vulnerable to the teaming problem; elections would turn into contests of which party can nominate the most candidates.

The simplest alternative you can actually get people behind and that doesn't have serious problems is probably approval voting. (For those unfamiliar: Just vote for any number of candidates, most votes wins.) It's simple, it works well, it doesn't really have glaring flaws like vote-splitting or teaming. There are various ways it's still not ideal, but if you want something for people to rally behind, I think its simplicity makes it the best candidate for that.

> I don't think Borda count is an acceptable alternative. It's vulnerable to the teaming problem; elections would turn into contests of which party can nominate the most candidates.

Isn't the simple solution there to limit the number of candidates a party can put forward? Forcing the party to try and be united about who they want representing an area?

OK, let's say we institute this cap. The obvious thing to do then is to spawn off officially-separate parties that can nominate more candidates for you. You could outlaw that, I suppose, but now we're getting into the area of things you need a judge to adjudicate, and, well... yeah.

Anyway yeah you see where I'm going with this. When I say "party", that's shorthand -- I'm not talking about literal party organizations. The teaming problem is just that having a bunch of candidates that are similar to each other increases the chance that one of them will win; it's the opposite of the vote-splitting problem, where having similar candidates decreases the chance that any of them will win. This problem doesn't require a formal party organization in order to manifest; strategic nomination is still possible.

Ideally one wants a voting system that is "cloneproof", i.e., has neither of these problems. (IRV is cloneproof, actually, but it has other glaring problems -- most notably its lack of monotonicity.)

You can always launch several parties.

Would that force them to be united? Putting some kind of rule in place might just force parties to balkanize. The barrier to candidate proliferation might be the ability to raise funds. Would a party get more out of one candidate, or two candidates with half the money?

Republican and Democratic parties are already conglomerates of ideologies.

Both should split up, for the better gauge of people's opinions.

IRV produces a direct improvement in the politics in that it promotes coalition-building among minor candidates to defeat better-equipped competition: "vote for me and for this guy too".

That means that instead of everyone running alone and mudslinging each other, which gives voters little narrative to work with(which in turn tends to lead towards the incumbents and partisan votes winning) there's more time and space going towards specific issues and platforms that consensus can be built around.

As a voting system there are certainly better ones in terms of the vote itself, but having experienced the change I now think that the impact on consensus-building is an worthy criterion too.

Exactly. Toxic two-party politics is a direct result of FPTP voting, not a reason to dismiss IRV

Preferential voting, including IRV, effects change to toxic politics by encouraging a greater number parties, and encouraging them to work together and form coalitions to build strength through consensus and common ground.

RCV/IRV absolutely reduces negative campaigning. Because all challenger candidates want to be your second choice.

If your goal is to give outsider candidates who can't get as many votes as mainstream candidates the win, then no, IRV isn't going to work. But if your goal is to break the Duverger's Law spoilering problem a bit, to give third parties a chance to build up from small rather than their current direction of starting with President of the United States, IRV is good.

I've been living with IRV in Minneapolis for a couple of elections now, and what it's led to so far is tolerable, non-radical mayors. The most recent race was particularly interesting, as it took like five rounds to get a winner. I didn't vote for him, but he's been a much better mayor than I expected.

Also, some interactive examples of different voting techniques:

https://ncase.me/ballot/

I came here to post this. I don't understand why Approval Voting and Score Voting don't get more love. The UX is easily understandable as we have been trained to thumbs-up and X/Y star ratings on everything from yelp and netflix. A simplified version in the form of the old pick one UX could be provided for voters who don't understand the new mechanism.

They both seem strictly better than what we have and yet I never hear anything about them. They even solve Arrow's Impossibility Paradox and have the confidence of Arrow himself.

FairVote has a lock on the "we need to change our voting system: this is the best alternative" message, and from what I can tell are dogmatically opposed to cardinal systems in general.

There's also a history of bickering and personal attacks between different voting reform activists, including people in leadership positions.

The various ordinal voting advocates (Equal Vote, CES, etc.) really need help on the messaging and movement-building front. A lot of lay people struggle to "get" the abstract framing these issues are often presented in. It comes easy to computer programmers and mathematicians because we're relentlessly trained to think abstractly about algorithmic tradeoffs and emergent limits of systems.

But a lot of people just don't see why this issue is important. It just seems like an obscure focus.

> The various ordinal voting advocates (Equal Vote, CES, etc.

Cardinal. FairVote is ordinal (IRV).

Fargo ND recently adopted approval voting (it was approved by voters by a wide margin), and there's a campaign to get an approval voting ballot measure in St. Louis.

The Center for Election Science seems to be the main advocacy group that's pushing approval voting. They're a much smaller organization than fairvote.org, but they seem to be getting more funding and gaining momentum.

Fargo citizens passed Approval Voting by a 64% landslide majority in 2018. https://www.vox.com/platform/amp/future-perfect/2018/11/15/1...

St. Louis will likely adopt Approval Voting in 2020. https://www.electionscience.org/st-louis-polling-results

STAR Voting will be on the ballot in Eugene, OR in 2020 and will almost certainly pass, given a previous county-wide initiative passed among Eugene voters. https://www.registerguard.com/news/20190828/star-voting-retu...

Range voting is arguably the general case of approval voting and it has the same problem: candidates have no incentive to be honest about how the voting system works and most will campaign on "just mark 1" or "just give me the max score" in order to convert back to a FPTP.

This is not a hypothetical concern, something like this has already occurred in Australian politics. Queensland switched from exhaustive preferential (IRV where all candidates must be ranked) to optional preferential (1-n candidates may be ranked) and then ran "just vote 1" as their campaign slogan. In doing so they obliterated the centre-right and further-right parties that traditionally operate as a coalition in Parliament.

Relying on voters to be aware of the mathematics of a voting system and on candidates to represent those honestly is problematic. IRV is simple to explain, works well enough and can be efficiently hand-counted.

So, your argument is that because this system has a feature that allows it to revert back to exactly what we have now, but also gives options so that more informed voters can better indicate their preference, and is otherwise strictly better, we shouldn't do it and stick with what we have?

Arguments against IRV do not apply to SV/RV or AV. IRV and FPTP fall prey to shenanigans via Arrow's Impossibility Theorem because they require candidates to be ranked relative to one another (or you just pick the best one as in FPTP). Once they do not need to be ranked (as in two candidates can get the same vote from a given voter), then none of the Arrow problems apply.

SV and AV are also simple to explain, work well and can be hand-counted. And they don't have the huge problem that IRV and other ranked voting mechanisms have.

& how can IRV be hand-counted? It is the most difficult system listed here to compute and most difficult to explain to voters why the outcome happened.

> & how can IRV be hand-counted?

It's very easy. Australia regularly does it with millions of ballots.

Step one: you count the ballots by hand.

Phew, what an exhaustive list of instructions!

No it is not "very easy". IRV is not summable, so a two-way communication is needed between every round. The San Francisco city elections page once explained this as follows: “Due to the requirement that all ballots must be centrally tallied in City Hall and not at the polling places, the Department of Elections has not set a date for releasing any preliminary results using the ranked-choice voting method.”

So you either have the communication between every round (which is essentially re-counting) or you physically take all the ballots to a central counting location, which introduces delays and chain of custody issue.

Also, IRV leads to more spoiled ballots. http://scorevoting.net/SPRates.html

So while there are more important reasons to prefer Score Voting and Approval Voting (better outcomes, better resistance to strategy, etc.), complexity is certainly not a non-issue.

First preferences are summable and it is usually clear, on the night of the election, who has won the majority in Parliament. This is called the "indicative count".

As for chain of custody, again, Australia has solved this problem. Every box is numbered, every seal on the box is numbered. All of them are signed out by electoral officials, witnessed by at least two scrutineers from different candidates. When the boxes are opened this process is repeated.

Do you know how many times ballot boxes have gone missing in an Australian election? Once. In a century. Once. For which the High Court voided the election because there was a minute but non-zero chance that the outcome would've been affected. It was re-run from scratch.

IRV has potentially multiple vote transfer steps and passes whereas all the other mechanisms are a single pass sum. IRV is easily the most complex to both compute and understand compared to the other alternatives (AV/SV/FPTP).

Furthermore, doesn't Australia use it only for House elections where you don't have to compute across a particularly large population?

Your tone is rather dismissive here.

Australia uses it both for House and Senate elections. Millions of votes cast in a day. Indicative count on the night.

My tone is dismissive because I am tired of being told IRV is "too complex" or "won't work". Because it's simple and it has scaled flawlessly for decades. You just need to believe the experience of other countries is more valid than a theory.

The AU senate is STV, no? And even then, the largest polity is NSW with only 8M people. 12 US states are more populous than this.

But, I'm with you that IRV is workable (there's ample evidence) and better than the abomination that FPTP is. I guess my point is that AV/SV address these concerns through their simplicity (I find both of them less complex than IRV in both voter UX and result computation) and also that they avoid the unintuitive problems mentioned in the OP ncase link, which can potentially be even worse than the problems of FPTP simply because they are hard for even educated voters to wrap their heads around, whereas the problems of FPTP (vote splitting) are easier to explain (but still bad. this isn't a defense of fptp).

Range voting is an abomination. It's the perfect example of a system ranked as good by Arrow Theorem logic which horribly fails most actual requirements for a voting system - unlike Approval Voting, which despite the tactical voting non-issue is practically reasonable.

[Some practical requirements for/questions about a voting method:

* Does it always have a winner?

* Can the resulting winner always enjoy public legitimacy?

* Is the voting system simple enough for the public to understand?

* Can it map to useful public choices?

* Is it susceptible to cheating?

* Does it bias towards the center or not? (Note that some people might prefer biasing and some not).

* Can voters signal to politicians they must change course? ]

Score Voting is objectively the best of the commonly proposed voting methods, with any mixture of strategic or honest voters. http://scorevoting.net/BayRegsFig.html

It is objectively one of the simplest voting methods based on summability, rates of spoiled ballots, speed it takes the average voter to complete a ballot, etc.

http://scorevoting.net/Lorenzo

Score voting? Is that how they're calling it now?

However it is called, there are many reasons for why it is so bad, but they all boil to one point: it's asking voters for information they do not have while ignoring the human realities of the vote. As we code monkeys say: GIGO.

99% of humanity doesn't have a 'Score' for politicians* , but the reality of the vote is that whatever choice the voter has, the voter can push effectively by giving min/max scores - or at least is very likely to think it can be helpful, and feel guilty when they don't do it - imagine losing an election because you merely gave your preferred candidate an 8!

A Range Voting election where everyone gives min/max scores de facto may look mathematically identical to Approval Voting, but it is far worse in reality.

That's because voters will then convince themselves their choice is deserved - their politician is a 10 and the other is a 0, an enemy. The problem in other words is that voters will mostly not partially abstain, but be pushed into a false polarization beyond what is useful for society. Psychology, not math.

There are other points, but this message is long enough. TL;DR: "Score" voting is the triumph of mathematically optimized voting methods that ignores too many human considerations.

But look, I am willing to go half way: Approval voting is a perfectly good method and I can accept Borda. Both are strictly better than the current method even according to the weird measurement "The Center for Range Voting" uses. Approval can even be seen as a (sane) version of RV, so there's no reason at all for the center to object.

* Oh, and the 1% who actually has a score? That's almost worse, because these scores are nearly always the compilation of single issue organizations which tend towards the extremes and never look at circumstances.

Borda count may be the only ranked choice voting system worse than Instant Runoff. The simulations here are more compelling than any mathematical argument I could make. http://zesty.ca/voting/sim/

Approval voting is my preferred system, both for its intuitive appeal and quality of outcomes, but I'll settle for any change at all.

What about those simulations makes you prefer Borda to IRV? The weird shapes IRV produces seem to make a pretty compelling argument against it.

If you look at the IRV plots boundary by boundary instead of as a whole, you see that most of the boundaries are in the same places as they are for other methods, but the boundary for whichever candidate is in the center can become chaotic. If you look at borda however, all of the boundaries shift, and they all shift in ways that favor candidates toward the center of the distribution.

This says to me that Borda would not intuitively match people's ideas of what an election would do far more often, while IRV would have chaotic results mainly in exceptional cases.

(I like approval the best. The one argument I've heard for why IRV is useful is that you can adapt it to do multi member proportional representation in a clever way https://www.fairvote.org/fair_rep_in_congress#what_does_the_... )

> a third-party candidate doesn't spoil an election

that's enough of an improvement for me

Agreed. Let’s not allow perfect to be the enemy of the better.

After not making any significant changes to voting in over 200 years (besides expanding it to more people, thank goodness) in the USA, taking one step towards a better future is gigantic. Once we take that first step, future steps will be easier (even if they’ll still be difficult at first).

I can’t wait for RCV, any flavor, to come to more cities and states.

P.S. if you’re in Massachusetts and feel strongly about ranked choice voting, come join me helping out https://voterchoice2020.org/

If Nader takes enough votes from Gore, then Gore will have the fewest first place votes and get eliminated first.

Then Nader has to go up against Bush and it all comes down to who the Gore voters put as their second choice. It may only take a small percentage of them putting Bush second for Bush to win. https://youtu.be/JtKAScORevQ

Was this election spoiled? By Nader or by Gore? You can argue the semantics of what "spoiler" means, but to me it's clear that IRV doesn't handle this situation well.

It is safe to vote for a third-party only if they can't win. When three-way elections get close, you have to be more careful.

This hypothetical is so far from the 2000 election as to be unrecognisable. If I'd written "If Nader takes enough votes from Bush..." it would be about as realistic.

In any case, if Gore preferences flow to Bush over Nader, that's because those voters are expressing their preference for Bush over Nader. That's not spoiling: it's literally the entire point of providing a full ranking, is so that nobody's vote is ever discarded and so that any winning candidate must have accumulated an absolute majority.

The parent post said:

> a third-party candidate doesn't spoil an election, because your second-place vote will still count. (E.g. Nader wouldn't have taken votes from Gore, so Gore would have won instead of Bush.)

I was just trying to point out that this isn't always true. Instant Runoff Voting can eliminate the centrist, the best compromise candidate, first.

If we adopt IRV and third-parties grow, then we move into territory where IRV gets some elections obviously wrong. It has happened.

> Was this election spoiled? By Nader or by Gore? You can argue the semantics of what "spoiler" means, but to me it's clear that IRV doesn't handle this situation well.

In that scenario more people preferred Bush to Nader, all the bush supporters plus a few Gore supporters. So no it wasn't spoiled at all, the most popular candidate won and the system worked.

It seems you think the goal is to elect the centerist candidate or something?

Why not score voting instead? People can actually understand it and it gives even better results: https://ncase.me/ballot/

Score voting is approval voting, except that you are allowed to make the tactical error of assigning a score between the two extremes. This is always a tactical error, de facto the same as flipping a carefully biased coin for each candidate.

To be honest, I do personally think score voting comes the closest to what is truly most democratic and accurate, even better than Borda.

I hesitate to recommend it, however, because I'm not entirely convinced it's intuitive enough for the average voter to understand and apply properly. I'm also afraid it's too open to criticism of seeming overly "subjective", though I personally don't think it is.

I would be very, very, very happy to have evidence that my fears are unfounded, however.

The one thing I find iffy about score voting (or range voting) is that there's a decent chance people will just score everyone 5~8. This would give more voting power to people that deliberately only use 0 and 10.

So I'd advocate approval voting. Just vote for the options that you find acceptable, the option with the highest acceptance rate wins.

> there's a decent chance people will just score everyone 5~8

That sounds incredibly unlikely to me. I don't think I know anyone who would rank both Donald Trump and Hillary Clinton as a 7. To the contrary, it seems like people will almost always put either 0 or 10, making it not much different from approval voting.

That is a rather extreme example, but still I don't think many people will actually fill in a 0, plenty of 2s and 3s maybe.

Besides, like you said, if everyone always puts either 0 or 10 then it's the same thing as approval voting anyway so it can't hurt to just use approval voting.

A downside of score voting (aka range voting) is that people are likely to have less influence if they vote honestly.

If we image the 2016 election replayed under range voting, I can easily imagine a lot of Clinton voters rating her 6/10 or whatever, and most of the Trump voters rating him 10/10. Even if you have a lot more voters that prefer Clinton, Trump still wins. An advantage of approval voting in that scenario is that it forces everyone to take an extreme position, yes or no, which is what a strategic voter would do. So, you'd have strategic voters and honest voters voting the same way and having the same amount of influence.

Range voting would be great in a primary, though.

Those who dilute their approval across potential winning candidates are under-represented relative anyone with an extreme view who would apply a dis-proportionate approval level to their smaller (maybe singular) set of approved candidates.

The approval or not of a given candidate might best depend on who supports them _at all_ out of the entire population, over any position in the rankings. That would mathematically simplify to ignoring rank order, considering all rank voting sets as approval tracks, and always eliminating whoever had the least votes until a single winner is left.

However that poorly reflects the preference of selecting the most desired winner, which is why other slightly more complicated algorithms were designed.

I get your concerns, but this IS an instance where I think progress over perfection is key.

The mere idea of changing a voting process is a big hurdle, and if it is shown that "yeah, we can do that" and that FPTP is not the only voting system in history, we have achieved a victory.

And as others have said, IRV is a huge win by allowing people to vote "FOR" something, rather than strategically vote against something.

So, even if the third parties rarely break out, it would more accurately poll the public's sympathies.

> the proposed system is specifically Instant Run-Off Voting (IRV), which does not make a centrist third-party candidate likely to win.

I agree with you that it's not as nice as something like Borda count, but I think I only agree with you that it wouldn't help a centrist third-party candidates chances if we assume, for the sake of argument, that switching from first-past-the-post to IRV wouldn't change who voters pick as their top candidate.

And that seems like a very difficult assumption to support, because the typical public discussion around third party candidates makes it explicit that people habitually choose to vote for someone other than their top choice in first-past-the-post elections.

I can speak somewhat from experience here. Some of my local elections use (non-instant) runoff voting, and my decisionmaking process in the first round of those elections is markedly different from the one I use in elections without a runoff. For starters, it doesn't typically involve having to hold my nose while I fill out the ballot.

A third party running against two polarizing candidates is a pathological case (as are two polarizing candidates alone).

IRV gets better when there are more candidates, because they can form alliances and trade preferences. It goes further than that by /encouraging/ new candidates too because they can align themselves with other candidates and band together against competition.

FPTP voting doesn’t encourage new candidates, because they’re competing against the big two. That’s why American politics sucks so much, both in its staleness and how the opponents treat each other.

> a third-party candidate doesn't spoil an election

Without the fear of throwing their vote away, more will rank the 3rd party candidate first.

But IRV still does allow you to throw your vote away. Once third parties become competitive, the best strategy in IRV regresses back to "your first choice vote should be whichever of the two most viable candidates you prefer".

IRV solves the problem of people irrationally throwing their votes away on candidates who have no chance of winning, but it absolutely still allows competitive third-party candidates to spoil an election. This is not just theoretical, it happened in Burlington in their 2009 mayoral election, which subsequently resulted in IRV being repealed there.

I feel that the reason IRV was repealed in Burlington, was because an independent won, which was not palatable to either the Democrat, or the Republican political machines.

The system worked exactly as intended, the people who usually hold power felt threatened by it, and engaged their political machinery to restore that power.

I don't live in Burlington so I may not have as much background on this as a native would, but my impression is that the Burlington election showed that strategic voting it's still a problem with IRV, contrary to what its supporters generally claim. The Republican voters would have preferred that the Democrat won over the Progressive, and they could have caused that outcome by dishonestly ranking the Democrat higher on their ballots. Thus, the purported benefit of IRV of allowing you to honestly vote for your favorite candidate first while still taking your second choice into account was discredited.

The problem with that kind of strategy is that you may be giving away the election to the Democrat.

You don't actually know, at the time that you cast your ballot, how everyone else will vote. This kind of strategic voting is difficult to reason through, and incredibly likely to backfire.

Regardless of whether anyone actually votes strategically in practice though, it seems pretty understandable that people would feel disillusioned when they see an outcome like this. Most IRV supporters say "you can honestly rank your preferences among all candidates without worrying about throwing away your vote"...and then the Republicans in that election ranked their preferences honestly, and it turns out they threw away their votes.

As you point out, it's harder to predict whether you're throwing away your vote because it's harder to reason through the effects, but I think that's just going to make even people unhappier with IRV and generate even worse election outcomes. At least in FPTP you can usually tell what the best strategy is in advance, so if you throw away your vote you only have yourself to blame.

In FPTP, the strategy is only obvious if your candidate is likely to win... Which is a bit of an ouroboros.

As a former Burlington resident that participated in that election this whole case study is a travesty - the baby was thrown out with the bath water.

Essentially what happened is that the person who was the 2nd most popular in the first round ended up overtaking the Democrat, so it wasn't a wild swing or anything... but after that happened the winner turned out to be a terrible person who was removed from office, then the two major parties lobbied to get IRV thrown out with the bad mayor and managed to get the majority on board.

As I've mentioned elsewhere, it's essentially like looking at Nixon/Clinton/ye favorite impeachment person and saying... "Welp, we're impeaching Nixon, so I guess this Democracy thing is a bust, I wonder if George II would be willing to take us back."

The two major parties warped what was just a bad candidate into a reason to regress the voting system because it allowed non-two party candidates to win easily (they still can win because 802 is serious about quality over party - hence the republican governor, and socialist senator/former mayor of BTV)

https://en.wikipedia.org/wiki/2009_Burlington_mayoral_electi...

I don't see the problem?

The third party won. That's not a spoiler. And several of the criticisms listed on that page are really dumb.

Edit: Okay, so someone else would have won using ranked pairs... that's a different problem from "a spoiler" though.

In this election, the spoiler candidate was the Republican. If the Republican hadn't run, the Democrat would've beaten the Progressive, and the majority of voters would've preferred that outcome (including the majority of Republican voters who assumedly thought that their honest rankings would be properly reflected in the election result).

It's a confusing example because we're used to spoiler candidates coming from third parties, but the politics in Vermont are a little bit unusual.

IRV isn't perfect, but it's a huge improvement over the current state of affairs.

I would point out that the "won't elect a centrist" perspective might be a bit of an American perspective and a perspective that might change with IRV.

For example, let's look at Canada. You have approximately from right to left, Conservative, Liberal, NDP, and Green. There are districts where the Conservative party won with around 30-40% of the vote while left-of-center parties gained 60%+. I'm guessing that most people that voted Green would prefer a Liberal candidate to a Conservative one. Similar for NDP voters. In that case, votes would have been transferred toward a more centrist party.

I think in the US there's also the trap of the primary system. Primaries often have extremely low turnout compared to the electorate and so finding an energetic base can swing a primary in a way that it can't swing a general election. Many people who are more centrist don't vote in primaries.

Frankly, you might just be underestimating how many centrist voters there are in general elections compared to primaries. Even as the Democrats have lots of great candidates, Biden is still topping the polls so far.

IRV certainly isn't perfect, but it's an incredible improvement, easy to explain, easy to trust, etc. It might not end up with the candidate that is the most acceptable by some weighting system, but it does favor people with broad appeal.

Honestly, if we got rid of the primary process and had a hypothetical general election that was Trump, Romney, Biden, and Warren, I think it's very likely that Romney or Biden would win. I think the big issue drawing parties away from the center is partly that primaries are decided by the small number of people who show up to them. If there was one general election with many candidates, it would likely be a different story.

Heck, if the Republican primaries had been IRV, Trump likely wouldn't have been the candidate. Even if he was popular on the extreme, more centrist votes would have coalesced around someone else.

IRV is simple and trustworthy. Borda, frankly, seems difficult to explain and trust. Ok, let's say that I have 4 Democrats and 1 far-right Republican with a uniform distribution of the 50% of democratic votes for the democrats. With 10,000 voters, the Republican gets 5,000 points. Each Democrat gets 1,250 * 1 + 1,250 * 0.8 + 1,250 * 0.6 + 1,250 * 0.4 which is only 3,500 points. The Republican wins a landslide victory despite an evenly split electorate. Or do you force people to list all candidates?

Heck, let's go with Bush, Gore, and Nader. Let's say Bush gets 48%, Gore gets 37%, and Nader gets 15% (and assume that Nader voters aren't going to vote Bush). With IRV, it's easy. 13% protest-vote for Nader and Gore gets elected once their votes transfer. With Borda, it's more complicated. Bush gets 48% of the points and then Gore gets .37 * 1 + .15 * .67 which is only 47% of the points. Bush wins - but that's not what the public wanted to happen. They thought they could vote for Nader first without hurting Gore!

IRV does have tactical voting considerations, but they seem really simple to understand and trust by comparison.

IRV isn't perfect, but replacing a voting system requires something that people easily understand and trust. People easily understand the implications of ranking and eliminating the bottom person, redistributing those votes. They understand how it eliminates spoilers and lets them vote for who they want in the order they want. A points based system is a lot harder for people to understand and trust. Again, IRV isn't perfect, but if we're going to have change, it has to be easy for people to understand and accept. Personally, I prefer single-transferrable-vote and multi-seat constituencies, but that requires radical change to our system. IRV is a simple change that people intuitively understand, accept, and that so many people have wanted for decades now.

IRV isn't perfect, but it's a huge improvement over the current state of affairs.

I disagree. In theory, yes it's an improvement, but it still doesn't eliminate the spoiler effect, and in practice it's much harder to understand why it produces the election results that it generates.

I think most people who study this topic would agree that approval voting is better than IRV. Many people then go on to say "but at least IRV is better than FPTP". My concern is that IRV is effectively worse than FPTP, because it may poison people's enthusiasm for alternative voting systems when they see the strange outcomes it sometimes yields.

I don't think my fear is just theoretical either, as this has already happened before when the 2009 Burlington IRV election produced a strange outcome, and they subsequently switched back to FPTP.

I'm concerned as you are. I've come to believe that IRV in a world of Cambridge Analytica spinoffs could be even more dangerous than FPTP. Not only would strategic voting be possible (using algorithms to target swing-voters), it'd be almost impossible to detect, and nearly impossible to do general audience journalism about.

Intuitively, I'd expect ordinal methods to be much harder to game, but I'm not even convinced this is the case. I've done some research trying to dig into this question, but a lot of the distributed-systems-game-theory-economic-math is hard for me to understand, and isn't focused on the messy problem of consensus between humans as much as it is on the problem of consensus between computer-like agents.

In a world of Cambridge Analytica spinoffs, I believe this is an extremely important issue, and that FairVote's take is outdated.

> which does not make a centrist third-party candidate likely to win.

The goal to ranked-choice voting schemes isn't to promote centrism, it's to eliminate the possibility of electing candidates without majority support due to factional vote splitting.

As far as fixing polarization of factions within the electorate, that's not something an election scheme is going to be able to do at all. Modern Fox viewers and vegan college students are simply never going to see eye-to-eye, and that has nothing to do with what choices they're presented with every year in November.

I highly recommend CES's video on voting methods that highlights the non-monotonicity of IRV [1]. I would prefer the adoption of simpler models that can be implemented and understood by voters easily, such as approval or score voting.

[1] https://www.youtube.com/watch?v=2q_eMUGCU5U&feature=youtu.be...

The point of getting rid of FPTP is absolutely not to “make a centrist third-party candidate likely to win”.

I'm not sure I follow why IRV means the centrist would be ignored. That assumes people actually _want_ to vote for the extreme polarizing candidates.

People may currently only be voting for the more polarizing candidates because they get the most press and they seem the most "electable".

With IRV people could actually vote for the boring centrist candidate without worrying that their vote isn't going to count.

In your thought experiment, the entire country is extreme one way or another. If this country used the Borda method, it seems like a moderate would win every time, despite the fact that there are no moderates in the entire country. Under IRV, the country would probably regularly rotate between extremes, which seems to correspond more closely to the actual makeup of the electorate. If a country is mostly moderate, it should have a moderate government. If a country is polarized, both sides should be represented in government.

This is especially pronounced if we suppose that the "moderate" party isn't just the middle of two extremes, but actually has some distinctive characteristics of its own. For instance, imagine that there's a liberal party, a conservative party, and a single-issue party that advocates for the interests of farmers. Supposing a split like you proposed, with the farmer party as the "moderate centrists" it seems like the government would perpetually dedicate most of its money and energy towards agriculture, even though no voters particularly cared about that issue.

It really seems like your preference for Borda is based on your own preference for moderate politics, not on any desire to faithfully represent the electorate.

> So to really get away from the poisonous political polarization of our times, ranked choice is necessary, but it needs to be Borda or similar, not IRV.

You are definitely better off, but none of these solve polarisation issue. All deliver the tyranny of the majority.

In the current system we generally elect one person per electorate. If you have 4 polarised groups, the split is 40%, 30%, 20% and 10%, and they are roughly evenly distributed among the electorates, then most of the people elected will be from the 40% group simply because they will win most electorates. The 40% will run the place, the other 60% will get very little say in what happens.

This doesn't change much if you use one of the other voting systems. Lets 30% and 20% groups mostly agree with each other. It's true that with First Past Post the place is run by the 40%'s. But if with one of the others the only change is it the place will likely run by a coalition of the 30% + 20% because they are the only ones that will be elected. So yes the people in power change - but again 50% the people get no say in parliament. We are a whole 10% getter off. Whoopee.

One way to get a balance is to have bigger electorates and have 10 people elected from each. Since 10 are elected from each the likely outcome is 4 of the 40%'s, 3 of the 30%'s and so on. Since none have a majority in parliament, they all have to compromise to get what they want. And that nice outcome happens regardless of which of these voting systems you use.

I think all this focus on voting systems distracts from the real issue - tyranny of the majority.

While I agree with you in principal, in the context of the state of NY, and especially within NYC, ranked choice voting will do nothing more than hand the results of every future election to those who control the political machine.

Think of it this way, in 2018, a top 2 primary in NY-14th district could very well have resulted in both AOC and Crowley, just like it did in the case of various districts in states like California and Washington which have top two primaries, but unlike in AOC's case, the incumbent largely pulled in those who would otherwise be Republican voters.

This same exact story would've played out in the Queens DA race as well. In the democratic primary, there were 3 candidates who could be in some ways marked as left, center and right. The left and center candidates both got around 40% of the vote +- 100 votes, but the right leaning candidate got around 15% of the vote. If there was a second round of voting among this block of voters, the center candidate would've likely done much better in said second round.

The issue you describe is really only likely to happen, when you have something like 10+ competitive candidates, when you have two candidates pulling 30-45% of the votes each, this being a problem becomes way less likely.

> So to really get away from the poisonous political polarization of our times, ranked choice is necessary, but it needs to be Borda or similar, not IRV.

All these fancy vote counting methods will only give you some scattered "third party" (actually more like independent, no party) candidates here and there. They will not give you a multi-party system. Only proportional voting can do that.

It's actually kind of hilarious. You're totally right.

So what it means is third-party candidates still have zero chance to win whatsoever, but the votes cast towards them would no longer be lost.

You're still allowed to pointlessly vote for an independent, but now you can do that AND still influence the actual battle between Rep and Dem.

> So what it means is third-party candidates still have zero chance to win whatsoever

Third-party candidates have won in many elections, just not the presidency.

Maybe it takes time for people to adjust. In SF it’s netted dissatisfaction in many voters in my neighborhood.

People are treating it like a trifecta rather than most favorite to least favorite of top three. So you end up with people few people had as their main favorite. And do it’s very underwhelming, at least first go round.

Borda seems to assume every voter ranks every single candidate. That seems pretty impractical -- it's asking way too much of voters, unless the total number of candidates is somehow limited to ~5 (and even 5 would be pushing it IMO). Am I missing an obvious workaround to avoid voters essentially needing to rank every single candidate for their ballot to be considered valid?

Edit: Wikipedia has a discussion on this issue [1], and none of the solutions seem great. The obvious one (giving unmarked candidates the lowest number of points) devolves into anyone being strategic only ranking their single favorite candidate as #1, which is the current system.

[1] https://en.wikipedia.org/wiki/Borda_count#Truncated_ballots

If I understand your example, then one of the extreme candidates would actually have a super-majority of first-place votes, right? I'm not sure elevating the centrist there is the right move. I mean, it sounds like a pleasant outcome, but at some point you have to accept the will of the voters.

Condorcet [0] is widely recognized as the best voting system to address voting issues raised by Arrow's impossibility theorem [1]. No voting system is perfect, but Condorcet gets as close as possible.

[0] https://en.wikipedia.org/wiki/Condorcet_method [1] https://en.wikipedia.org/wiki/Arrow%27s_impossibility_theore...

I prefer the Schulze (path voting/winner) method, and I don't care if the most moderate candidate wins, I care that voters are able to express their desires and if a centralist candidate happens to be the overall best choice that's fine; I just disagree that it's always the best option to fully compromise.

https://en.wikipedia.org/wiki/Schulze_method

I don't see why it should be taken for a given moral good that a third party cannot spoil an election (Ranked Choice doesn't actually prevent that, at any rate, it just lessens the effect near the midpoint). If your political values are that a multiparty system is better than a binary choice, the risk of an upstart candidate being a spoiler gives him and by extension potential voters more bargaining power.

> the risk of an upstart candidate being a spoiler gives him and by extension potential voters more bargaining power.

bargaining power how? if you have a situation like this:

- 55% of people vote left - 45% of people vote right

The left party has a clear but narrow majority. Now a third party, further to the left enters the race! Say they can pull 15% of the vote, largely from the left (quite popular!)

- 15% of people vote farther-left - 40% of people vote center-left - 45% of people vote right

now the election likely to be won by the right-wing party - even though 55% of voters still favour a left-of-center result. Voters feel pressured to vote for the center-left party, despite having farther-left views, because they don't want the right wing party to win. Under first-past-the-post systems, there is no bargaining power short of "I will drop out of the race if you do x" which does not seem to happen much

If they are unable to pull enough support to "spoil" the election result, they are considered a "throwaway vote" and marginalized. No power at all.

Because they don't have to win in order to have power: they can extract concessions from one of the stronger candidates in exchange for dropping out and possibly endorsing him. In Ranked Choice, especially with IRV, third party candidates can be more easily ignored if their 1st choice votes will be assimilated.

In your example, under Ranked Choice, the 15% who really wanted far left policies can be safely ignored after the primary. In traditional voting, the majoritarian center left now has to take their concerns seriously if they don't want to lose the whole thing.

> they can extract concessions from one of the stronger candidates in exchange for dropping out

While technically true, I can't think of a single example of this happening outside of the US two-party system. I.e. you'll see democrats stepping down during the primaries but it's not like Ralph Nader offered to pull out if Gore adopted various policies.

> the 15% who really wanted far left policies can be safely ignored after the primary

The idea is that you wouldn't need to have a "primary" - you could have multiple parties running candidates in the ACTUAL election. The whole reason for the primaries is to get the in-fighting out of the way so you can run a single candidate - because under FPTP anything other than running a single candidate is suicide. With Ranked Choice you could have multiple candidates on the left and right - and people could vote for a first and second choice.

E.g. in 2016 both bernie and hillary could have run. as long as people picked both as part of their ranked choice there would be no negative consequences and people would be free to choose whomever they believed in the most without worrying about their perceived "electability".

Political parties have four jobs:

Build the party.

Elect their endorsed candidates.

Forge a platform.

Lobby for their platform.

RCV mostly helps with #1. No more spoiler effect enables 3rd parties to slowly, incrementally build awareness, socialization, support. At the bottom of the pyramid (locally). To eventually mount challenges to higher offices.

Think of politics as a marathon (or maybe a horde invasion), with elections being periodic opinion polls.

Wonder how this would have worked out during the two elections where Ross Perot pulled in a decent share of the vote.

Perot pulled roughly evenly from across the political spectrum. There is no reason to assume it would have changed any result.

https://www.jstor.org/stable/2111651?seq=1#metadata_info_tab...

Anything for a least objectionable candidate winning majority of support.

See the stupidity of the result of Maine Governor election of 2010. How can a person with a mere 37% win?!?!?!

I agree with your criticism, but this still seems like an improvement.

> BUT, the proposed system is specifically Instant Run-Off Voting (IRV), which does not make a centrist third-party candidate likely to win

Perhaps not the third party candidate, but what about a same-party candidate that garners votes? Perhaps for example, lots of democrats would have selected Bernie as their number one choice last time around, then Hillary, then Trump. Perhaps Bernie would have had a better chance in that situation.

Then again, I'm not sure this suggested system in NYC would even allow you to select someone who is not the nominee of a party without a write-in.

Couldn't then a completely unknown person win?

Is "the middle-of-the-road candidate wins" necessarily a feature when you're in a political environment that's been generally dragged far to the right of the rest of the world by decades of big business donating to right-wing candidates? Most "centrist" Dems feel more like the Repubs of my youth these days.

The duopoly parties are both right wing parties, and increasingly extremely so. Third-parties tend to be more centrist or leftist.

Often (but not always) when people talk about a party or politician or political position being right or left, they are talking about its relative position compared to the distribution at hand.

Or, I guess sometimes they also might mean, like, some ambiguous combination of an idea of an objective left/right axis, along with the idea of a relative axis?

Regardless, people generally agree that, currently in the US, that the Democratic Party is “to the left of” the Republican Party. As such, for the purpose of practical communication, it is often useful to understand what people are talking about in terms of this way of distinguishing between the two.

I think the confusion here is mostly coming out of comparing US stances to globally expressed stances - both US parties are pretty right leaning on the global scale, but within the US left is democrat and right is republican - while people at the more extreme ends of the spectrum complain that, on the right, the republicans are too liberal and, on the left, democrats are too conservative.

The sort of interesting thing is it's really hard to be objective about this, if you were to measure where parties have stood on minority representation over history both parties today are astronomically to the left of any parties in the 1800s. I think the main thing we can sort of examine is where policies lie in relative comparison to the voting base and under that measure I'd expect that we have far more conservative and corporate representation today from both parties than the people voting - in a large part due to the political corruption existent in our system.

"The sort of interesting thing is it's really hard to be objective about this,"

And beyond that, words do not mean the same things over time.

The following quote from the 1972 Democratic Party platform would be read quite differently in 1972 versus 2019.

https://www.presidency.ucsb.edu/documents/1972-democratic-pa...

Family planning services, including the education, comprehensive medical and social services necessary to permit individuals freely to determine and achieve the number and spacing of their children, should be available to all, regardless of sex, age, marital status, economic group or ethnic origin, and should be administered in a non-coercive and non-discriminatory manner.

Your second paragraph is closer, though in fact most people don't think about it at that level. They typically don't know or understand that the position is relative to the current distribution. When they say "democrats are on the left" they mean it in an absolute sense, and both the duopoly parties play that ambiguity up to their collective advantage.

The kind of voting system proposed isn't likely to change that.

> But you're still stuck with one of the polarizing two-party candidates who will win.

In New York City most people just vote for the candidate that most closely matches their own ethnicity / gender, so none of the candidates have any reason to run especially polarizing campaigns.

Citation needed

Reply on news.ycombinator.com