[EM] hmmm. Maybe I missed something before SF passed IRV then called it RCV?

[Kristofer Munsterhjelm]

On 07/01/2017 08:46 AM, Sennet Williams wrote:
> If so, I conditionally apologize.
>   I had certainly never heard of the term "RCV" until an anti-IRV SF
> election commissioner suggested that RCV would be a better name and no
> one objected.  That was around 1990(?), after the SF measure was passed
> but before it was implemented.  The vote redistribution method is
> apparently exactly the same in SF, Oakland and Berkeley, all which were
> voted on as "I.R.V.", but are now usually referred to as RCV.
>
>   If the term R.C.V. had actually been used before that election
> commission meeting then the implication is that I.R.V. is merely a type
> of RCV, so I learned something new.

Consider what the term actually says:

Ranked: the voters are ranking something
Choice: what they are ranking are the choices (i.e. the candidates)
Voting: and what we're talking about is a voting method that uses these 
rankings to make a decision.

Does IRV fit this description? Yes. It uses rankings and makes a 
decision based on them. Does Range? Not really, because it uses ratings, 
not rankings. Does Condorcet? Yes. It, too, uses rankings and makes a 
decision based on them.

So if we're to go by what the term says, then both Condorcet and IRV are 
ranked choice methods, or instances of ranked choice voting.

> But if that day was actually the  first use of R.C.V., and now someone
> is applying it to competing counting methods, then that is CO-OPTION,
> with IRV's competitors trying to take advantage of IRV's huge success in
> the Bay Area.

That someone who were in favor of Plurality described IRV as ranked 
choice voting is no surprise, because IRV *is* (a form of) ranked choice 
voting. But so are other methods.

I suppose the point of contention is whether ranked choice voting is a 
term of art, specifically referring to IRV, or if it's a descriptive 
term, as I've explained it above. If it is the former, then I could see 
IRV supporters being annoyed at Condorcet/etc supporters "coopting" 
their chosen term. But to me, and probably also to Robert, "ranked 
choice voting" is descriptive. So from that perspective it is not 
Condorcet (or Bucklin, etc) supporters who are coopting FairVote's 
chosen term, but FairVote trying to set an equals sign between IRV and 
voting methods based on ranked ballots. And understandably, if you 
prefer Condorcet, and if you see the term as a description, you wouldn't 
want someone to say "the only way you can have Ranked Choice Voting is 
by using IRV".

> Personally, I think that some other ranked ballot systems (like
> Condorcet) are theoretically more logical but functionally more
> difficult, but either one would have the same effect on politics.   As
> for "weighted" systems, that would not be legal for official U.S.
> elections.  (Each voter can only have one vote)

Here are three very simple Condorcet methods if you're familiar with IRV:

1. IRV, but with Borda instead of Plurality.
2. IRV, but consider the two candidates with least first place support 
when you're to do an elimination. Eliminate the one of the two who 
fewest voters prefer to the other.
3. Order candidates in Plurality order, from least to most support. 
Compare the first and second candidate. Eliminate the candidate who the 
fewest voters prefer to the other. Now compare the winner with the 
candidate next in line. Continue until there's only one candidate left. 
He wins.

They're not very *good* Condorcet methods (e.g. neither the first nor 
the second is summable), but they're Condorcet. Number three is also Smith.

There's also Woodall: "eliminate candidates as in IRV until there's a 
Condorcet winner among the remaining candidates, then elect him". It's 
Smith too, and robust to strategy (see Green-Armytage's papers), but it 
requires that people know what a Condorcet winner is.

> I haven't seen the specific text of Don Beyer's. "Fair Representation
> Act," but I have seen it looks like a variation of Choice Voting being
> referred to as RCV, but that is not going to pass anyway.
>
> I predict that specific voting system that will be most successful is
>  "The Maine System" defined by the Question Five:  IRV used for partisan
> primaries followed by IRV used for the general election.  (The use of a
> coin flip to break ties is not legal, so scratch that part.)

That is assuming that IRV doesn't blow up like it did in Burlington, and 
that's part of Robert's point. To quote him: " if you want to reform 
elections with Ranked Choice Voting,  ***don't*** promote it with IRV. 
when it fails again (and it will) that will set back RCV reform for 
generations."

The blowup scenario could go like this:
- IRV is used in more and more places
- Voters think it's now possible for a third party to grow
- Third parties grow until they're competitive with the two main parties
- IRV gets confused in an election and elects the wrong candidate
- There's a backlash and IRV is repealed
- Voting reform in general suffers for a long time, because it's 
associated with IRV. E.g. in the fashion of "Oh, you want a better 
voting method? Like IRV? We all saw how that turned out".

If the blowup scenario is realistic, then it doesn't matter whether IRV 
has momentum. If anything, focusing on just one method is bad, because 
if it fails, the backlash will be all that more severe. It would be 
better to try a variety of methods: Range in one place, Condorcet 
somewhere else, IRV a third place; and see how they do before going 
national.