Software - Single Transferable Vote

Hello,

My branch of my political party is conducting a STV election in December. Could you please recommend a software which we can use for counting ballot papers.

Best,
Oliver

There are many many different versions of STV. Could you please specify?

We are using STV for the election of a ranked list of candidates:

http://www.rosenstiel.co.uk/stv/orderstv.htm

We are thinking of using Meek STV with backwards tie-breaking: Is there a software which can do that?

Best,
Oliver

You might not have a lot of luck with such a request here. There is a pretty solid consensus here that STV is terrible. Would you consider using a modern score system?

Does OpaVote work for your purposes?

1 Like

We have to use STV. But I’m interested which electoral system you recommend for electing an ordered list of candidates.

OpaVote doesn’t sell software as far as I know.

Best,
Oliver

I would recommend sequential proportional approval voting.

It should be possible to input the votes from the ballot papers into an OpaVote poll and then get the results that way.

This software might work as well: https://github.com/OpenTechStrategies/openstv

That is a good choice. Or RRV. They are both simpler and give fairer results than STV. I could write code for them in like 30 min.

There are some modern method which look to give better results but people tend to not want to be on the cutting edge.

Or just use Asset Voting. It is simpler and gives fairer results than RRV. You do not need to write code for it.

I found that hard to follow. Is that the same as RCV with multiple winners?

If by “that” you mean STV and if by RCV you mean IRV not condorcet then yes. STV is the multi winner version of IRV. And before you ask, STV inherits the nonmonotonicity, polarisation and zero sum properties of IRV. Its not a good system.

It’s proportionality does mitigate it’s flaws. STV (the method Olivia is talking about technically isn’t STV but rather a modification of STV that outputs a proportional ranking, but it’s very similar) is much better then IRV, though I’d still rather use something like SPAV, PAV, SMV, harmonic, etc.

Would an STV with equal ranks allowed fix these issues sufficiently to make it a good system?

Would at least help but why bother?

We can’t expect single-winner methods to really represent every voter, and ranked methods generally are the best interface for allowing voters to decide what they’ll support (little to no coercion on how a voter must vote.) In a time where polarization and traditional institutions like the media are failing to cover what people really want, the best way to know is to represent all of them, and while I can see why non-monotonicity is unappealing, the worst I’d expect to happen because of that is a slightly different member of a party winning or something like that, not for quality of representation to significantly diminish. I think the issue with criticizing PR for diminishing consensus is essentially, if a majority of your society is voting for extremists and not consensus-makers, then they get what they asked for; if that same majority would seek to implement a consensus-biased PR method, I’d expect them to also choose consensus candidates in traditional PR systems too.

But there are good systems so why would you choose a bad one?

1 Like

Give me a realistic example where STV clearly underperforms relative to a cardinal PR method; I think in practice, STV elects candidates that most voters would say are their favorites of the bunch, and that’s about all you can ask from the representatives. Some mix of consensus and proportionality is also possible, but I really do wonder why voters wouldn’t just vote for consensus candidates of their own accord under STV.

My issue with STV is somewhat that it inherits IRV’s flaws, but the real problem is how difficult it is to count, not the unrepresentativity issues. Consider:

STV Asset
(1) Voters have to cast a rank-order ballot of potentially 20+ candidates. Truncating is a must. Ballots have to be checked for if they are legit or not. (1) Voters have to choose either one or a small number of potentially 20+ candidates. Ballots still have to be checked but the process is simpler.
(2) STV cannot be counted in precincts. (2) Asset is additive, at least in the first part.
(3) One option is for all votes to be shipped to a central location. This has security issues.

Another is to have precincts publish their totals, then the correct candidates are eliminated, and so on. But this makes recounts a huge pain.

(3) There are only two rounds in Asset Voting. The vote totals must be published within a specific timeframe, and then negotiations can kick off to decide the winners.

It might take a little bit more time to decide the winners, especially if there are recounts, but this is nowhere near as difficult as STV is.

Do you have an example of an STV failure?

To some extent, allowing equal-ranking makes filling out a ranked ballot as easy as filling out an Approval PR ballot.

Regarding precinct-summability, most proposals for STV seem to focus on doing it multi-member districts, where the difficulty of centralizing the counting would be much less significant or a security risk. I think Party List is also a good option if there are many winners, or even some sort of biproportional method to maintain districts but add proportionality. Overall, if you can convince people to go for Asset, fine, but why throw STV completely out as an alternative?