Hacker News Reader

Comment by wyldfire | Hacker News Reader

I think it would be excellent if it found more adoption in the US (or approval voting or similar).

What drawbacks/surprising consequences, if any, are there?

Most surprising consequence is probably that IRV is unlikely to break 2 party dominance, even though it this is an often claimed advantage of IRV.

The frequently imagined scenario with IRV is that, in a two party system where most people will rank 1 of 2 parties first, then a small percent of voters can safely rank a third party first. This party is then eliminated, but not before a surprisingly strong showing (e.g. 5% first rather than the .05% we currently have). This would likely happen.

The problem arises in the next election, when the third party is now strong enough to command a large percent of first place votes. Since the party with the least first place votes is eliminated (even if they have a ton of second place votes), the third party now stands a real risk of cannibalizing votes from the next most similar "mainstream" party. If this cannibalization were to occur, both parties could lose to the other mainstream party that many voters might not have preferred! Due to this risk, it's likely voters would strategically rank the third party below the mainstream party even if that wasn't their true preference. In this way, IRV can still lead to 2 party dominance.

Assume for the sake of argument that the policy of the "third" party is just like that of the "second" of the two "main" parties. In that case, it's true that the second party could be eliminated in the initial round of voting. However, all its votes would then transfer to the third party. In that case the third party very well could beat the "first" party, satisfying supporters of both the second and third parties.

It's true that third parties would be unlikely to map to the current political spectrum in such a "pure" fashion. However, that just means that voters reject the political choices currently on offer. Voters really should have that option.

I think you have two incorrect assumptions here. You're assuming A) that parties and political ideologies exist in 1 dimension, and that B) the second party is closer to the third party than the first, such that its votes will go to the nearest neighbor, which is then the third party.

Even granting A, B is definitely not true. There are many similarities between the two major parties, and removing one in favor of a third party would likely split voters between the third party, and the other major party.

Also, A is not true either, so the distance between parties becomes even more unclear.

I for sure believe voters should be able to reject the current political choice on offer! I support IRV over our current system for sure. But, my main point is that IRV is actually not what many people believe it to be, and is far worse than other, simpler systems, such as approval voting.

This whole thread is a thought experiment. I made a single simplifying assumption, for the sake of clarifying the particular qualities of that thought experiment: that voters would understand the policies of the second and third parties to be identical. In that case, IRV isn't "unlikely to break 2 party dominance", as you claimed above. So, that claim isn't universally true. It might be true in some circumstances, but we don't know what those might be. You'll need to explain more carefully why we should reject RCV in favor of some more perfect alternative.

IRV fails the favorite betrayer condition and isn't monotonic.

In Canada it’s been used by the Conservative party for their leadership contests. Both times, the response to the winner has been “Who?”.

So in a field with a lot of contenders it can favour marginal candidates.

You also end up with an absurdly big ballot. Past, say, five ranks the choices are hardly meaningful.

I’m favourably disposed to this type of voting but those are a couple of downsides.

> You also end up with an absurdly big ballot. Past, say, five ranks the choices are hardly meaningful.

How does the type of voting affect the number of people on the ballot?

Right now if you vote for a third party in the US your vote is wasted.

Under a ranked ballot system you can vote for a super marginal party confident that your more mainstream preference will ultimately get your vote if your marginal candidate loses.

This encourages votes for smaller parties

Are you saying more parties crop up? I still have third parties on FPTP ballots.

My vote in MA is not mathematically meaningful today, as you could correctly call the 2020 presidential race for blue back all the way back in 2016.

Change the voting process and I could vote for Party Q, then F, then D and have something closer to my true preferences be measurable vs today and give the parties a sense of where they stand with the populace. I’m still not convinced it breaks the binary star in US politics.

Right now, if those are my prefs, I have a choice of voting Q, voting D, or staying home. If Q is a third party, there’s an argument that vote might as well have not been counted or cast (because from today’s voting you can’t infer the true level of support for Q because some Qs cast their ballot for D just on the off chance that R would otherwise win; for the same reason, you can’t tell as much about the true R or D support either).

My question was around the size of the ballot, not the weight of a person's vote.

More third parties become reasonable contenders.

You have to rank them against one another. 10 candidates, 10 comparisons. 100 candidates, 100 comparisons.

Alternatively cardinal systems scale trivially (you express how much you like an individual candidate -- independently-- not how much you like a candidate -- comparatively -- to another candidate). You can bullet vote a party line if you really want.

That doesn't affect ballot size. You have the same number of candidates with boxes next to their name you put a number in, unless someone is making the implementation more complicated.

I agree that there are some downsides, but regarding the conservative party leadership contests, it's probably fairly likely that the "who?" response would've happened under FPTP.

2017, likely that Bernier would've won, but I'm not convinced that he was more well-known than Scheer. Happy to be proven wrong here.

The 2020 result would likely have been the same as it is now. A small contingent of Sloan/Lewis supporters strategically voting would have tipped it to O'Toole.

It can have strange results when there isn’t a clear moderate.

Canada uses it for a lot of leadership elections. What often happens is that a candidate nobody knows or really wants wins because they got picked so their vote wouldn’t go to the main competing candidate.

So you get party leaders who nobody seriously examined as the more hardcore members just needed to park their votes.

this is a direct consequence of switching from "most popular wins" (FPTP) to "least unpopular wins" (IRV), which is a subtle but very important distinction

The drawback is that it is more complex than a traditional ballot. It's not especially difficult, but given how many people have difficulty with even our current ballots it will cause some additional issues.

It would hurt the current duopoly so you can expect a significant pushback against it.

If MA passes theirs it could pick up real mainstream momentum though.

> It would hurt the current duopoly

I get that this is conventional wisdom, and a lot of smaller parties support rank choice voting because they think it helps them. But neither of the major parties like "spoiler" candidates, it seems like the ability to "collect" votes from the green or libertarian parties would be in their interest.

I suspect the reason it's not more widespread is more mundane. It isn't incomprehensible, but it's more complicated than what we have. If you're explaining you're losing. Also people are (perhaps rightly) cautious about messing with how elections happen. Finally, it doesn't ignite the same passions as abortion or police reform. The interest groups pushing most strongly for it are, by definition, small and not really powerful.

That's why I'm a bigger fan of approval voting. No ballot updates and people just check next to who they like.

Actually I think this is why we see widespread support for RCV, because it still favors the major parties. On the other hand, cardinal systems (approval or STAR), do not favor mainstream parties. Really they don't favor any party.

Reply on news.ycombinator.com