[EM] Some ways to extend social rankings to scores

[robert bristow-johnson]

> On 02/10/2026 2:07 PM EST Kristofer Munsterhjelm <km-elmet at munsterhjelm.no> wrote:
> 
> 
> I was talking about scores as outputs, not inputs :-)
> 

My point is, with STAR, that if your input is A>B>C, unless you anticipate the Center Squeeze effect (something that happens less than 0.5% of the time in RCV elections in the U.S.), there is no reason whatsoever to mark your STAR ballot any differently than A:5, B:1, C:0 .

> The voters would still use plain old ranked ballots.
> 

So this is some Borda-ish scheme that turns ranked ballots into scores.

> The method would just return more detailed information -- like how FPTP 
> shows the strength of victory on a scale (as percent of first 
> preferences gained), even though FPTP isn't a cardinal method.
> 
> As for STAR strategy, it makes sense that STAR would behave that way if 
> everybody does 5-1-0, because that's more or less reproducing the logic 
> of IRV/TTR (first preferences count a lot, other preferences don't). I 
> suspect that STAR strategy is less unbalanced, though; that some times, 
> 5-4-0 is good, other times 5-1-0 is good, and you don't get skewered the 
> same way you can in Approval if you get it wrong.
> 

The *only* time (assuming 3 candidates, or 3 significant candidates) that you ever want to mark your STAR ballot 5-4-0 is when you actually *want* your 2nd-choice (or lesser evil) candidate to actually *beat* your favorite candidate to get into the STAR final runoff.  This would be because you fear your favorite cannot beat the candidate you hate in the runoff but your lesser evil candidate *can* beat your greater evil candidate.

That's my spin on it.

--

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

"Imagination is more important than knowledge."

.
.
.