Fork me on GitHub
#off-topic
<
2021-08-20
>
sova-soars-the-sora02:08:15

Score Voting > Ranked Choice

Elliot Stern15:08:26

Score > Instant Runoff, definitely. Score vs Schulze, though, is much more of a toss up.

Elliot Stern15:08:39

At least so long as you don’t care about being able to explain it to people who don’t have much of a math background. It’s pretty easy to explain to most CS people, though.

sova-soars-the-sora18:08:16

I don't know what Schulze is. I assume there are several possible "seats" so I think Score Voting would make the most sense.... it's just like the olympics and slots on the podium

Elliot Stern19:08:20

Schulze/beatpath is a Condorcet method, which means that if someone beats every other candidate in a head-to-head election, they win.

Elliot Stern19:08:27

In particular, in Schulze, you first make the complete weighted directed graph of all the candidates, where the weight and direction indicates how many votes they win by, then calculate the all-pairs widest path problem. It turns out that one candidate will have a wider path from them to each other candidate than the other candidate has back to them.

Elliot Stern19:08:52

And in the single winner case, that’s the winner.

sova-soars-the-sora17:08:34

That is super fancy and there is no practical way to explain that to the laity 😅 ... would the results would be different for multi-seat results (taking the n top winners) when compared w/ Score Voting ?

Elliot Stern15:08:00

Potentially, yes.

didibus04:08:21

Is Shultze like ranked, but you can put many people in the same rank?

didibus04:08:52

Like: Rank:

``````B
C
A``````
Shulze:
``````B C
A``````

Elliot Stern22:08:43

Schulze allows tied rankings, yeah.

Elliot Stern22:08:28

It’s a ranked system, but is quite different from “Ranked choice”/Instant Runoff in terms of how it behaves.

didibus23:08:01

Is the difference more than just that it allows same rankings? Like there's also something special about how it tallies up?

didibus23:08:20

Or to be more precise, are we saying that there are multiple ways to tally up a ranked choice that allows equal rankings? And not all of the ways to tally give the same outcome? Even though the input data is the same?

Elliot Stern16:08:46

There’s a lot of ranked methods, both ones that allow equal rankings and ones that don’t. Many of them will agree much of the time, but they generally have different outcomes

Elliot Stern16:08:22

Borda, IRV, and Condorcet methods are the big ones to know about.

Elliot Stern16:08:16

Condorcet methods will all elect the Condorcet winner if one exists. A Condorcet winner is one who wins against every other candidate in pairwise contests. IRV isn’t a Condorcet method, because while a Condorcet winner will win the last round if they make it that long, they can be prematurely eliminated because IRV ignores a lot of information on the ballot.

Elliot Stern16:08:30

“Ranked Choice”/IRV in particular has a lot of nasty corner cases because of the way it works. For example - in 2008, Republican voters in Burlington helped elect a Progressive. If the right number of them stayed home or even voted for the Progressive, the Democrat would have won instead.

didibus19:08:47

Ok, so the differences are in the way the data is tallied? But the voting is the same? Like you just rank candidates in order (and allow equal rankings)

Elliot Stern19:08:28

Yeah. There’s a ton of different ways to take rankings and pick a winner.

didibus19:08:30

That's interesting ya. I didn't know that.

seancorfield03:08:09

Oh, wow, I see I'm on the list... How did that list come together @U051KLSJF? Clearly not just folks applying for funding! 🙂

danielcompton04:08:51

Nope, these are people that our members nominated or are well known open source developers. There's no strings attached to the funding and you can decline it if you don't want the money 😀

seancorfield05:08:57

I should be so lucky... Interesting list of names... all nominated, or are some of them applicants?

dpsutton05:08:18