STAR Voting viewed pragmatically


A lot of the theoretical criticism of STAR Voting comes from the position that ideal STAR Voting strategy is what I’ll call “abacus voting”—i.e. pushing candidates to the extremes like beads on an abacus, which cannot actually overlap.

While I believe this is trivially true (and I’m sure Warren or Jameson or Andy Jennings can come up with a simple mathematical proof of this), the practical consideration is whether voters can effectively exploit this in the real world. I’m highly skeptical that they can.

To wit, I have yet to see any of our resident math geniuses cite a simple strategy heuristic that is better than fully honest normalized ballots. We KNOW such a strategy exists with Score Voting. Warren readily acknowledges it.

Given that, I would argue that even the most sophisticated voters (perhaps especially the most sophisticated voters) will tend to be honest with STAR Voting. And so, even if the runoff step reduces utility efficiency (as e.g. Warren Smith expects, but others believe the opposite), we should expect that to be more than compensated for by the increased voter honesty that STAR Voting incentivizes.

I am not saying that I am 100% sold on this argument, but I think it’s highly compelling.

We then of course have to ask, is this postulated improvement worth the additional complexity of STAR Voting? Well, two things:

  1. Complexity is mostly a concern of political viability. And there is a plausible narrative I hear from Mark Frohnmayer and other STAR advocates, that the “majority runoff” step actually helps with political viability, after having pitched the system to thousands of Lane County voters face-to-face (they got 16,000 signatures, so take this experience seriously). The argument is simply that voters often come up with the initial objection about bullet voting, but then generally seem satisfied with the counter about being honest to have an impact in the runoff step. Valid or not, what matters most is whether this rhetoric actually is effective with real human voters.

  2. Both Score Voting and STAR are much simpler than IRV. So STAR is still below an obviously critical complexity threshold.

Imagine that STAR Voting “tricks” tactical voters into being more honest, and then elects the first-round (honest Score Voting) winner 95% of the time. Then it’s essentially just a “better” way of getting the honest Score Voting winner—which should be the end goal that Score Voting advocates say they want anyway, right?

And even if you don’t totally buy all that, I think outright opposition or hostility toward STAR Voting is really unreasonable. It gets voters using a Score Voting ballot, and establishes that there are valid alternatives besides IRV. I feel that this should earn passionate support from any Score Voting advocate, if you’re really taking the long-term pragmatic view.

P.S. This is post ID 42. So you should really take it seriously.


If we were using STAR to vote on our voting system, I would do:
IRV = 0
FPTP = 1, a strategic vote because FPTP -> Score/Star is easier than IRV -> Score/STAR.
321 = 3
STAR = 4. This is a good system but I like Score better.
Approval = 3
Score = 5

I think that STAR is a good system but I am a little worried of two party domination if the third parties are never convincing. Score Voting is totally immune to clones and spoilers.

When it comes to things where multiwinner systems work:
Single winner FPTP = 0
PartyList = 1
STV = 1
Districts using Score or STAR or 321: 5
Asset Voting: 5


Re: complexity and political viability.

STAR voting and all the cardinal single winner systems are precinct summable. The idea that ballots would be counted centrally and audited centrally, and that they can’t be locally counted or audited for IRV and ranked systems is a huge deal-breaker.

I say if you can count and audit locally, and you can explain to people how it’s counted so that they can explain it back to you then that’s simple enough to be viable.

But there is a lesson here from STV, which is neither simple, explainable, or precinct summable and yet which IS politically viable. The takeaway there is that if a system is capable of inspiring people and really delivering on the goals that they care about the details don’t matter.

STAR isn’t perfect as measured by criteria, but it does deliver on people’s core goals. I think that’s what’s made it so politically viable. It’s come a long way in a short time.


Apparently, Discourse allows you to create polls. It’s approval only, can’t do score (yet!)

  • IRV
  • FPTP
  • 321
  • STAR
  • Approval
  • Score

0 voters

And here’s the source code:


Bullet vote for the win!