Hacker News Reader

Comment by ggggtez | Hacker News Reader

No there are no real valid arguments against ranked choice voting. Ranked choice has the benefit that the there is no point in "gaming" your vote, so your optimal strategy and nash-equilibrium is to provide your real candidate preference order.

The key benefit of ranked choice is that voters will tell you what they actually want (even if their favorite candidate has no chance to win).

View on HN

Ranked choice has the benefit that the there is no point in "gaming" your vote, so your optimal strategy and nash-equilibrium is to provide your real candidate preference order.

This is very demonstrably false, both theoretically and in the real world. Ranked choice voting only solves the spoiler effect in cases where the spoiler candidate has a small enough amount of support as to be rendered non-competitive, but if you have more than two candidates who are competitive then the spoiler effect comes back in full force. This is not just theoretical; it happened in the Burlington 2009 mayoral election, and fallout from that election led to ranked choice being repealed shortly afterward.

So, not only does ranked choice still incentive strategic voting sometimes, but it does it in a way that is harder to see and understand. There are other non-ranked voting systems like approval voting and score voting that are much simpler and better in this respect.

Ranked choice also has other drawbacks - like violating the monotonicity criterion - as well as more practical issues like making it impossible to distribute the counting process across precincts.

No there are no real valid arguments against ranked choice voting.

This is also quite incorrect. Our current system of first-past-the-post is bad, yeah, but there are plenty of valid arguments against ranked choice and - in my opinion - there are much better systems we could choose instead.

Incorrect.

From the article you discussed on wiki (Burlington 2009 mayoral race)

> In IRV, there is no tactical incentive for a voter withhold or falsify their second choice.

We can get into the weeds about mathematical details like Monotonicity and Condorcet, but the truth of the matter is that nearly all the time (much more often than First-past-the-post) these measures end up being identical anyway.

I don't think anyone would argue that first-past-the-post would have produced a better candidate in this election than RCV did.

The statement "there is no tactical incentive for a voter withhold or falsify their second choice" says nothing about a voter falsifying their first choice, which is exactly where IRV broke down in that election. If Republican voters who preferred the Democrat to the Progressive candidate had falsely specified the Democrat as their first choice, they could have changed the election outcome in favor of a candidate they would have preferred over the winner that IRV elected.

I don't think anyone would argue that first-past-the-post would have produced a better candidate in this election than RCV did.

I don't think so either, but I also think that's a straw man argument. The Condorcet winner would've been the Democrat. IRV elected the Progressive, but according to that same Wiki page FPTP would have likely elected the Republican, whereas the Democrat would have likely won under systems like approval or score.

IIRC, the Burlington election as evidence of a failure isn’t universally agreed. Some point to it as an example of it working.

It's a tricky example because the Republican was effectively the third party candidate in that election. But once you realize that, the spoiler effect is very easy to see; the Republicans would have mostly preferred the more centrist Democrat candidate over the progressive Independent candidate, and if they had strategically ranked the Democrat first on their ballots the Democrat would have won.

By voting honestly and putting their favorite candidate first, they screwed themselves over and threw away their vote, and were understandably bitter about it given that ranked choice is frequently claimed to solve this exact problem.

It's also maddening that a single mayoral election is apparently the slam dunk counterexample.

Meanwhile, Australia has been voting this way for a century. And over that time period, I think there have been fewer ugly paradoxes than in the US.

I agree that Burlington is overweighted, but I also wonder: we know about Burlington because in the US, jurisdictions typically make data available down to the level of how many times each rank-order was voted. Does Australia do this? Or is it possible that Burlington-like paradoxes have happened that we just don't know about, because we've only seen the tabulation-level data?

Australia also uses proportional representation (STV) which is quite different from the winner-takes-all instant-runoff elections that are being referred to as "ranked choice" in the US.

Only in the Senate. The House of Representatives is single-member electorates.

Ah thanks, my mistake - I didn't realize that.

The advantage that RCV -- at least exhaustive RCV -- has over other systems is that by forcing a ranking, you prevent a collapse back to FPTP.

In approval voting, score voting, range voting etc, parties can and will push their voters to effectively "just vote 1".

This isn't a hypothetical -- it happened in Queensland, Australia when they switched from exhaustive preferential to optional preferential voting (you could rank any number of candidates, rather than having to rank all candidates). The Labor party campaigned on "Just Vote 1" and destroyed the Liberal-National Coalition at that election. Those parties in Queensland then had to merge to offset the effect on their primary votes. In turn that caused the separate Liberal and National party caucuses in the Federal parliament to merge into a single caucus (otherwise there'd be three party rooms). That in turn greatly strengthened the power of hard-right Liberals in the selection of Prime Minister. It's been fairly fucked ever since.

There are better voting systems on paper, but typically those theories don't account for political parties actively misleading voters about how the voting system works.

This is unfortunately exactly what Maine is doing, as per their example ballot linked above.

That is disappointing.

As others are saying, there are certainly drawbacks to IRV. I think there are theoretical results showing that all of the mainstream proposals for voting systems are susceptible to strategic voting aka “gaming” your vote, among other issues.

What they aren’t acknowledging is that the situation is even worse with our current system.

My personal preference is approval voting, but I’m down with all sorts of systems that are better than the status quo. I expect that anything that lets you express more of your preferences will increase buy in and participation. Increased participation with preferences closer to people’s true preferences goes hand in hand with increased legitimacy.

We shouldn’t let perfect be the enemy of better.

We shouldn’t let perfect be the enemy of better.

I mostly agree with you, but I do fear that some of the hidden flaws of ranked choice could end up poisoning the well for better systems like approval voting. After 2009, Burlington switched back to FPTP, and as far as I know there have been no serious efforts to try any other alternative systems there.

If massive money and effort is put into adopting RCV while people falsely claim that it solves the spoiler effect, then once the problems with RCV start to become more widely understood, people are going to be a lot more skeptical next time around when advocates of approval/score/STAR voting come around and say "okay but this time it really fixes it!"

This is completely false. Pretty much all voting systems are susceptible to tactical voting, e.g. gaming the vote. This was formally proven via the Gibbard–Satterthwaite theorem[1], which showed that the only way for a voting system to be ungameable was to either have a dictator (a certain special voter who always picks the winner), or to pre-constrain the output space to only two alternatives.

This critique certainly applies to Instant-Runoff Voting, the algorithm in question here. It is known that IRV often incentivizes voters to not rank their favorites first [2].

[1] https://en.wikipedia.org/wiki/Gibbard%E2%80%93Satterthwaite_... [2] https://electionscience.org/library/tactical-voting-basics/

Not only is Gibbard–Satterthwaite only applicable to ordinal voting systems, it is only applicable to single winner electoral systems. A popular example of a multi-winner system is STV[1], which I suppose could be used to elect a President and Vice President, similar to the situation in the US before the Twelfth Amendment was ratified.[2]

[1] https://en.wikipedia.org/wiki/Single_transferable_vote

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

Gibbard–Satterthwaite only covers ordinal voting systems: it doesn't apply to, for example, approval voting.

Yes, you are right. I should have written more carefully. I am a huge fan of approval voting. Anyone reading this -- you should look more closely at approval voting, which surpasses ranked choice voting in many ways.

https://electionscience.org/library/approval-voting-versus-i...

This is incorrect: people do better voting strategically even with ranked choice. To take an example, 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 ranked choice 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. Staying home and not voting would similarly have helped them.

I said there is no real argument against it.

I'll go ahead and make the unfounded claim that situations like the one above are mathematical curiosities, and nothing more. True voter preferences do not align in a circle.

Reply on news.ycombinator.com