ClayShentrup

i co-founded the center for election science, which promotes approval voting.

my alternative proposal to simple sortition is election by jury. https://www.electionbyjury.org/

do approval voting now.

fight for election by jury tomorrow. https://www.electionbyjury.org/

approval voting gives you MORE information about opinions, even if this is counterintuitive.

this is because:

1. the IRV tabulation algorithm discards a lot of useful information. (there are better ways to tabulate ranked ballots.)

2. strategic exaggeration of rankings is more distortionary than with approval voting.

https://medium.com/@clayshentrup/expressiveness-6ef8c034bc65

we call that a voting _method_ not system.

approval voting is also just more accurate and resistant to tactical behavior.

https://www.rangevoting.org/BayRegsFig

> Score voting calls for ambiguous, proven-to-be-culturally-variable ratings that have no natural relationship to preferences

utter nonsense. the scores are just normalized utilities.

or, if you're being strategic, you're just min-maxing based on expected utility. https://www.rangevoting.org/RVstrat6

> (and approval does the same thing in an opposite way

utter nonsense. approval voting is just a binary 0-1 scale instead of e.g. 0-5. the math in no way changes.

> while score voting relies on ambiguous distinctions that don’t have a clear mapping from actual preferences

utterly false. there is nothing "ambiguous" whatsoever.

> approval forces voters to rank with artificial ties that don’t have a clear mapping to actual preferences

of course it has a mapping to actual preferences. what you're _trying_ to say is that the transform from utilities ("preferences") to scores is LOSSY. the problem is that transforming to ordinal rankings is even MORE lossy, and this is why approval voting still outperforms ranked methods in general, especially with strategic voting taken into account.

we have several pages explaining this common myth to laypersons.

https://medium.com/@clayshentrup/expressiveness-6ef8c034bc65

https://www.rangevoting.org/Expressiveness

> the most significant difference within democratic systems in quality of government by popular engagement, popular satisfaction with government, and a number of other measures is with the degree of proportionality of the main legislative body

there's absolutely no evidence for this. you're making the "proportional representation fallacy". i address this here.

https://clayshentrup.medium.com/the-proportional-representat...

so virtually everything you just said was incorrect. which is why it's good to study this issue and talk to experts before forming opinions.

there is no good argument for IRV over score/approval. score and approval blow IRV out of the water on literally every single metric that matters. they're simpler, more accurate (with any ratio of strategic or honest voters), and more transparent and fair to minor party candidates.

https://clayshentrup.medium.com/later-no-harm-72c44e145510

it's not even close.

there's nothing special about the condorcet criterion. it's mathematically proven that the most favored candidate can even be the condorcet _loser_. you just want the voting method with the highest voter satisfaction efficiency.

https://electionscience.github.io/vse-sim/

> As the article mentions, in the real world score voting would just be approval voting where you put a max score on some choices and 0 on others.

utterly false.

https://www.rangevoting.org/HonStrat https://www.rangevoting.org/Honesty https://www.rangevoting.org/RVstrat6

> And in approval voting you need to think about how others will vote and pick your cutoff point based on who you think has a chance

1. this is true of _every_ deterministic voting method. it's mathematically proven.

2. even approval voting is extremely accurate. see voter satisfaction efficiency calculations from harvard stats phd jameson quinn, who served with me on the board of the center for election science. https://electionscience.github.io/vse-sim/

> RCV isn't perfect, but in all but the smallest elections there's really no practical strategic voting considerations. You just state your true preference order.

ludicrous. IRV is extremely vulnerable to strategic voting. my aunt who voted biden even tho she preferred warren (to stop trump) would do the same thing with ranking: rank biden 1st to prevent warren from losing to trump.

something similar happened in the first alaska house special election. palin was a spoiler. she caused the democrat mary peltola to win, even tho fellow republican nick begich was preferred to peltola (and to palin) by a large majority. in other words, palins supporters hurt themselves by honestly ranking palin 1st. they would have been strategically better off burying palin and ranking begich 1st.

https://rcvchangedalaska.com/

and the VSE metrics show that even 100% honest IRV performs worse than approval voting with all voters being strategic.

IRV is simply one of the worst voting methods in existence. not to mention being radically overcomplicated and opaque.

i'm sorry that you posted this before having even an inkling about how any of this works.

no, it's just a trivial minor semantic difference.

https://clayshentrup.medium.com/oop-vs-fp-182475457a01

if you want the oop style you can just change "foo bar baz" to "bar `foo` baz" which is effectively the same thing as doing bar.foo(baz), because for some reason people are obsessed with passing an argument by putting a dot after it.

Haskell even on a bad day is radically more sane then all of these languages doing this bizarre obsession with foo.bar() instead of bar(foo), because they only allow overloading on the first arg. semantic indenting. type inference. automatic laziness. it's just another level.

what matters is satisfaction with the election outcome, not the five minutes you spent filling out a ballot.

https://rpubs.com/Jameson-Quinn/vse6

yep.

https://www.rangevoting.org/RVstrat6

this is a zero sum fallacy. social choice is not zero sum. voters who use intermediate scores donate more utility to society than they give up by not being strategically normalized, leading to an increase in net utility. with enough honest voters, even the "naive suckers" are better off than they would have been under a normalized approval-style ballot. https://www.rangevoting.org/ShExpRes

also see the "no math skills" argument.

https://www.rangevoting.org/RVstrat6

> When it is changed from IRV complexity is never the reason

it's a reason why voters vote YES to repeal it.

> The argument proffered in favour of returning to FFP is always "simplicity". It is a lie.

you're talking about politicians. the more crucial side of the equation is _voters_, who have to vote on whether to repeal it.

we need election by jury, not "democracy".

https://www.electionbyjury.org/

i co-founded the center for election science. this is superficially reasonable, but IRV could paradoxically make things worse by squandering election reform resources (namely, money) on a suboptimal and overly complex fix, at the cost of progress on approval voting, score voting, etc. then, if IRV gets repealed often due to its complexity, we end up worse off. alaska is plausibly going to repeal IRV this november.

tactical voting is normally what we call it when you DON'T vote for your favorite, e.g. a green party supporter votes democrat.

with approval voting they'd obviously vote for green too.

some of the people who normally vote green under the current system might _still_ only vote green with approval voting, but very few would do it unless green actually had a chance to win.

https://www.rangevoting.org/BulletVoting

> Even first-past-the-post seems to be doing a reasonable job of that.

utterly false. https://www.rangevoting.org/PConsumer.html

> Arrow's theorem also doesn't guarantee that you will have problems. It just says that for some votings systems you can construct voting populations with preference that can't be captured well.

no, it has nothing to do with capturing preferences. it simply says that no ordinal social welfare function can simultaneously satisfy these criteria:

    There is no dictator.
    If every voter prefers A to B then so does the group.
    The relative positions of A and B in the group ranking depend on their relative positions in the individual rankings, but do not depend on the individual rankings of any irrelevant alternative C.

> Arrow's theorem also doesn't apply when you allow bargaining, or people compensating each other.

that doesn't make sense. the result you get after bargaining would just _be_ one of the options.

well, yes. https://www.rangevoting.org/ArrowThm

but technically it only applies to social welfare functions, not voting methods.

i had a chance to visit kenneth arrow at his home in palo alto circa 2015 and we had a nice little chat about this.

cardinal (rated) voting methods are better than ordinal (ranked). so score voting, approval voting, and star voting are all good.

approval voting is just score voting on a 0-1 binary scale. star voting is score voting on a 0-5 scale, followed by an "instant runoff" between the top two highest rated candidates.

> public understanding and trust make a voting system more optimal.

star voting: score the candidates from 0 - 5 and the winner is the majority favorite between the two highest rated overall.

there's no risk with public understanding here. i've worked on this subject since 2006, and conducted exit polls with score voting ballots within months of getting into this issue. people understand it just fine. it's radically simpler than ranking.

this is a common intuition-based fallacy. i'm an expert with almost 20 years of experience working in this field with various phd's in political science, math, and game theory. i explain this here:

https://medium.com/election-science/expressiveness-6ef8c034b...

tl;dr - for any two candidates, X and Y, the voters who approve only one of them express a "ranked preference" between them. and because those voters are, on average, a good statistical sample of the X-vs-Y preference of the entire electorate, approval voting turns out to measure voter preferences very accurately.

whereas instant runoff voting can actually have much bigger problems in preference detection because it only looks at one layer of preferences at a time. for example:

35% left center 33% right center 32% center

in this example, center is preferred to both the left and the right by a LANDSLIDE 2/3rds majority, yet is the first candidate eliminated. this happens because of "vote splitting" (spoiler effect) between the two more partisan candidates. so it might _feel_ like you expressed your preferences better. but you didn't really.

and what about STRATEGIC VOTING?

if right wins, then the left supporters will wish they had strategically voted for center to prevent that. and if left wins, then right supporters will wish they had strategically voted for center to prevent that.

approval voting satisfies the "favorite betrayal criterion", which means it cannot possibly hurt you to support your sincere favorite candidate.

for these and many other reasons, approval voting is vastly superior to ranking. https://electionscience.org/library/approval-voting-versus-i...

there's no evidence that proportional repreesntation can achieve a higher level of voter satisfaction than good single-winner methods like score voting or approval voting.

and besides, you won't get PR at any scale beyond a handful of cities without first escaping duopoly, which requires score voting or approval voting as a prerequisite.

http://scorevoting.net/PropRep

> approval voting doesn't allow you to express your preferences very well

yes it does, in the aggregate. because the people who approve only X or only Y will statistically have a pretty similar X-vs-Y preference as all other voters (who approved both or neither).

https://medium.com/election-science/expressiveness-6ef8c034b...

and because approval voting uses the preferences it does have more accurately than IRV, it ends up being much more accurate, as measured by voter satisfaction efficiency simulations.

https://rpubs.com/Jameson-Quinn/vse6

i wrote a page on this very issue.

https://www.rangevoting.org/RVstrat6

strategy affects all deterministic voting methods, and an optimal ranked ballot can be quite complex. see an example from andy jennings, who did his math phd thesis on voting methods and co-founded the center for election science (with myself and some others).

http://www.youtube.com/watch?v=JtKAScORevQ

optimal approval voting strategy is simply to approve everyone you prefer to your expected satisfaction. to an approximation it's, "vote for who you normally would, plus everyone you like better." so it's rarely any more taxing than now, and often easier.

hundreds of thousands of voters have now used approval voting in fargo and st louis, and we've heard no reports of difficulty. approval voting is also simpler in essentially every way we can objectively measure. most of the points here apply to approval voting even more so.

https://medium.com/election-science/star-voting-is-simpler-t...

Yes you're correct. but cardinal methods are just generally better all around.

i didn't say your 2nd can't hurt your 3rd, but it can't *IF* your 2nd is eliminated before your 1st. which often happens. for instance, a simplified version of what just happened in alaska:

35% peltola begich

33% palin begich

18% begich peltola

14% begich palin

-- https://electionscience.org/commentary-analysis/rcv-fools-pa...

begich is eliminated first, even though he's preferred to each of his rivals by a sizable majority.

but more notably here, palin is a spoiler. her supporters get peltola—their least favorite. BUT if they insincerely/strategically rank begich in 1st instead of palin, then begich wins. i.e. they get their 2nd choice instead of their 3rd.

so they have to LIE ABOUT THEIR FIRST CHOICE in order to prevent the spoiler effect. whereas this is mathematically proven never to happen with approval voting.

> I still disagree that Approval's lack of later-no-harm is a feature, though. In practice, what it means is that voters have to think strategically about how well they think the various candidates will do, in order to decide whether they have to "risk" giving an approval to a candidate who they genuinely don't like, but who might be the only chance of preventing a detested candidate from winning.

you can't argue about strategic voting because:

1. as i just showed you, IRV/RCV has an even more severe vulnerability to this—you have to strategize about whether to support your FAVORITE, not whether to support your 2nd favorite.

2. approval voting gets better results with strategic voters than IRV does with HONEST voters, so even in the worst case scenario, approval voting is still better. see these VSE calculations by harvard stats PhD jameson quinn.

https://electionscience.github.io/vse-sim/VSEbasic/

my link already discussed all of this quite clearly.