[EM] Open letter to STAR voting promoters

[robert bristow-johnson]

> On 06/04/2024 5:26 PM EDT Toby Pereira <tdp201b at yahoo.co.uk> wrote:
> 
> 
> I think it does make common sense that rated ballots are easier to fill out above a certain number of candidates. I think some people may have previously cited some studies, but just from the common sense view, I think I would find it easier to rate with more than a handful of candidates. I was watching the Eurovision Song Contest last month, and gave the songs scores out of 10. It would have been way too much effort to rank all 25 or so songs. With scores, you can just work your way down the list (perhaps after identifying your favourite and least favourite to calibrate your scale), whereas with ranking, you have to do them in order with the potential risk of missing one out and messing it up.
> 

Yes, I agree about this type of contest.  But you might end up changing your score on some song when you hear another one that sounded just a little better (or just a little worse) and you need to bump one or the other over because you don't think they're equal merit. 

Now, again, here is my beef:  If you're taking a poll and your intent is to just communicate to the pollsters how well you like the songs for them to get a measure of popularity, Score voting is the correct way to go.

But if the point of polling consumers is for these consumers to get some song (and band and nation) elected as EU's best, and if there is some pride or satisfaction or some other other social or political benefit you derive from your crew winning, *then* I don't think Score is the way to go.  Or if there is some other song/band/nation that is crap (like Donald Trump and the influencers that follow him) if your other intent is that these assholes (who have some followers) are shut out and are *least* deserving of election as EU's best, then that contest should not be decided by Score.  Because now, as a partisan, your intent in marking your ballot is not to provide accurate information for measuring collective utility, your intent is to accomplish a political goal and that leads to tactical/strategic voting on a Cardinal ballot.

Cardinal ballots cause the voter to make tactical voting decisions the minute they walk into the voting booth whenever there are 3 significant candidates or more.  They have to decide how much they're gonna score their second favorite (or lesser evil) candidate.  That is inherent to Cardinal methods.

Ordinal ballots, you know what to do with your second fav.  You rank them #2.  If the election method is working well, it shouldn't harm your favorite if the contest ends up being between the two (LNH).  But you also don't want to worry about helping your second-fav beat the greater of evils (favorite betrayal).  Scoring your second-fav is an inherent tradeoff between LNH and FB.  But, unless Arrow and Gibbard are ruling the day, it need not be with Condorcet; remove any loser and the winner remains the same.


> But as has been said by Lime, this is a very different matter from whether you actually use the numerical values of the scores.

In my opinion, the semantics are meaningless if you do.  Or, at least, a Score ballot, with discrete score values, that is used only to infer relative preference, is a ranked ballot with equal ranking allowed.  There is no meaningful difference between Score Voting and Borda Count with equal-ranking allowed and that closes gaps in rank.  Now some people might insist that Borda cannot have equal ranking except for unranked candidates (who get a score of zero), then that's a little different from Score, but still comparable.

> You can just infer the rankings from them and run a Condorcet method if you so wish.
> 

But if the scores are not used numerically, except with a sgn(a-b) function, then it's essentially a ranked ballot that allows for equal ranking and that closes gaps of unranked candidates (which is the law in a few IRV towns, so an unused rank doesn't necessarily truncate the ballot).

I really don't want to conflate Cardinal vs. Ordinal.  Ordinal is about ordering, Cardinal is about scoring.  Borda seems to have a foot in both camps.

Lastly, I might reflect a little on my own politics regarding ballot access laws for governmental elections, it's the gubmint that running these elections.  Just like in other functions of gubmint in which people compete (because of limited resources) for access to some benefits that not everyone gets, there need to be some objective requirements (age for drinking and for health benefit, income for welfare, fee for license, money for bids on contracts) to get your foot in the door.  Eligibility.  For elected office or a ballot question there should be appropriate ballot access requirements (normally valid signatures on a petition) to get something or someone on a ballot in a public election.

For a single-winner office, if there are often more than 5 candidates on the ballot, that jurisdiction needs to increase the number of valid signatures.  At least in my opinion.  5 candidates is perfectly precinct summable (not for Hare RCV), provides the voter with a variety of choice, doesn't suffer from the Paradox of Choice, and, assuming Write-In is another option, needs only 5 ranking levels.

For multi-winner, I dunno.  I wouldn't expect much more than 4 + number of seats or double the number.

--

r b-j . _ . _ . _ . _ rbj at audioimagination.com

"Imagination is more important than knowledge."

.
.
.