On 04/12/2019 21.06, Markus Schulze wrote:

I think the point is to convince people who support IRV but also
recognize the failure of Burlington 2009 as genuine. From such a
position, it makes sense to advocate for a small change that fixes the
problem of Condorcet noncompliance, instead of replacing IRV with Schulze.

In the category of smaller changes, I would prefer Benham (which I got
confused with Woodall earlier), but even that might be too large a
change (as I said in my initial mail to Robert).

Does Benham pass independence of clones?

On a longer term, I agree that Schulze is better than Woodall or Benham
(perhaps with the exception if the voters are very strategic), but it
doesn't seem Robert has the luxury of going for one of the advanced
Condorcet methods.

Lotsa people to respond to.  I feel I must begin with Markus because, as a signal processing algorithmist, I have so much respect for Markus and for the Schulze method.

I think I understand the Schulze beatpath method and agree with the consensus of the geeks that, technically, it is simply the best currently-known RCV method.  Most immune to all of these voting strategies.

But I consider cycles to be an extremely rare outcome of any Condorcet-compliant ranked choice system.  And, while it's uncommon (perhaps happened only with Burlington a decade ago) for an STV or Hare or whoever IRV to not elect the Condorcet winner, we found out that it can happen at least once.  Appears to be the only RCV election held by a government that did not elect the Condorcet winner, but my concern is that a voting system that does not always elect the Condorcet winner may become law again in the very city where this IRV failure occurred.

If I had my druthers, I would perhaps plug Ranked Pairs because RP will always elect the same winner as Schulze in the case of a Smith set of 3 or less.  A cycle in a governmental election will be rare, but a cycle involving more than 3 candidates happening in a governmental election is, I think, virtually astronomically unlikely.  Ranked Pairs is, for me, so much easier to describe than beatpaths. Markus, to be totally honest with you, I just don't think that Schulze has an ice-cube's chance in getting adopted for a governmental election. But maybe another, simpler-to-describe Condorcet-compliant method can.

So if I had my druthers, it would not be BTR-STV.  But here I am in Burlington Vermont of all places.  At least in this town, we should be aware of how election reform was set back a generation (assuming it takes another decade to finally get RCV to return to this 3-party town) when this IRV hiccup occurred a decade ago.  Now the Progs are actively trying to get the same-old IRV, repackaged as "RCV", but no different than what failed us in 2009.  They have already submitted language for how the question will appear on the ballot.  They sorta left this to the last minute to try to punch it through without argument to get on the ballot.  The city council must make the inclusion decision in 12 days. If they do not put the question on the ballot, it will delay the decision for a year, but charter change is also state law, so it will take another year to get enacted.  And they want this in place for the next mayoral election.  That's why the current push on this legislation.

My only hope and my only intention is to try to persuade some skeptical people in the Vermont Progressive Party that their submitted language can be changed without breaking RCV.  I know they will come back and say my language is too complicated and theirs is simpler.  I will tell them that theirs is simpler and wrong and point out exactly why in what happened in 2009 (I can post that letter to the list, if people are interested in how I do politiking).  But I want to show them language that is the least amount of change from their language, but will result in a Condorcet-compliant method.

I will feel like I succeeded and gained something if any Condorcet-compliant RCV method is adopted, but if they don't change the language, I will hold my nose and still support the old, crappy IRV, if that is what gets on the ballot.  That would be better than electing Mayor 41% and that is what we are potentially in danger of now.

So I can't have Schulze, I can't have Tideman, I can't have FAB-STV (and I don't think that Richard's 4a, 4b, 4c is an improvement), I can't have Instant Pairwise Elimination.  I can't expect to, in 5 days, walk into the room where the Charter Change Committee meets and tell them to totally replace their entire language with something completely different (except for the ranked ballot).  The best I can hope for is language that changes how the candidate is eliminated in STV, so that the Condorcet winner is never eliminated.  That's what BTR-STV is good for; it is a political increment (when RCV becomes Condorcet-compliant) and sometimes increments in reform is all we can hope for.

So, for reasons of political practicality, it's gotta be BTR-STV or it goes back to Hare STV.  And that will also be a fight.  RCV might lose in either case and we continue with FPTP with 40% minimum.  But also a bad outcome would be if the same IRV was returned to statute and we had another failure like we had in 2009.  Then voting reform will be set back for a lot more than a generation.  At least if we can get an IRV that is Condorcized, we won't have a repeat of 2009 and the repeal that happened the year after.

So the only thing that helps is good, concise language for BTR-STV that retains the form of the original language that I posted, has one part that is different, and I get to defend that one part that is different.  That's all I can get away with.

Thanks, y'all for your attention and comments.  So far, I am sticking with my language, but I can be convinced that some other is better, if the suggestion comes from understanding what political and time constraints I have at the moment.

bestest,

robert

--

r b-j                  rbj@audioimagination.com

"Imagination is more important than knowledge."

Yes.

Benham  is more resistant to Burial than Schulze.   A candidate who is
top-ranked on more
than a third of the ballots can't be successfully Buried.

Say sincere is

43: A
03: A>B
44: B  (or B>A)
10: C

A is the sincere CW:  A>B 46-44,  A>C 46-10  (or 90-10).

With the ballots cast thus all Condorcet methods and also IRV elect A.

But say the B supporters Bury against A:

43: A
03: A>B
44: B>C (sincere is B or B>A)
10: C

Now A>B 46-44,    B>C 47-10,      C>A 54-46.

Now Schulze  (or Ranked Pairs or Smith//MinMax) using either the
normally advocated Winning Votes
or Margins reward the strategists by electing B.  I think Losing Votes
(especially if we don't allow above-bottom
equal-ranking) is much better and it still elects A.

Benham in this example elects the sincere CW.  It just sees that C has
the fewest first-preference votes and is pairwise
defeated (by B) and so eliminates C and elects A.

Chris Benham

On 5/12/2019 10:04 am, Kristofer Munsterhjelm wrote:

On 05/12/2019 12.57, Toby Pereira wrote:

Wikipedia makes it unnecessarily complex. A beatpath from X to Y is
measured by the strength of its weakest link, and the path is
constructed so that the weakest link is as strong as possible.

If p[X,Y] is the strength of the beatpath from X to Y, and there's a
cycle from X to Y to W to X, then that means that
- the strongest path from X to W is weaker than the strongest path from
W to X (since W beats X by beatpath),
but also
- the strongest path from X to W through Y is stronger than the
strongest path from W to X (since X beats Y by beatpath, and Y beats W
by beatpath).

That's a contradiction, because "the strongest path from X to W" must be
at least as strong as the one through Y. So cycles can't happen, so
beatpaths are transitive.

Still complex, but not nearly as daunting as Wikipedia's "it can be
proven" makes it out to be.

That kind of follows from the transitivity. If there are no cycles, you
can imagine running a sorting algorithm with "is p[X,Y] greater than or
equal to p[Y,X]" as the >= operator, and you'd get an unambiguous order
up to ties. Suppose it's a descending sort; then whoever is first in the
sorted list beats or ties everybody else in that list by beatpath.

It's low, although I couldn't quantify it because I don't know the
distribution of a real-life election after people have become used to
Condorcet. Even the question "would there be lots of Condorcet cycles"
is unknown, and just extrapolating from current elections leads into the
IRV trap (IRV works well on current elections, but then Burlington happens).

I'd probably say RP is better than Schulze, at least as a stepping stone
to proportional representation. It passes LIIA and can thus be used as a
base method for elimination in an STV method that reduces to RP itself
in the single-winner case, yet obeys Droop proportionality in the
multiwinner case.

The drawback is that ties are much messier, and any recount would need
to know just how pairwise ties were broken the first time. But perhaps
that doesn't matter with a very large election; the chance of a perfect
pairwise tie between any pair of candidates could well vanish to zero.

And Schulze is more well known, inasfar as any Condorcet method is well
known :-)

On 05/12/2019 02.52, C.Benham wrote:

That's why I added the strategy exemption. I know that the methods that
generalize Minmax are easier to strategize against than are the
Smith-IRV hybrids -- Green-Armytage's paper is clear enough about that,
and that the main reason is burial.

I've been trying to construct a voting method that is just as resistant
(or nearly so) while also being monotone (and Smith). I'm pretty sure
what it would look like with three candidates without equal-rank
(fpA-fpC), but I've been completely stumped on four and more. But
recently I've had somewhat of a break: now I know, at least, how to go
about the process instead of fumbling my way by brute force.

Now if I just had the time, I could perhaps write a paper of my own
about it. But there's quite a bit of work to do yet before I have a
method, and I suspect its implementation will look very strange due to
the way the construction works.

No, of course you "can't have FAB STV", which is no more than a
blueprint of a book at present and quite likely for my life-time (as
with Meek).
You've mixed this Richard (Lung of FAB STV) with the other Richard,
Richard Fobes:  "(and I don't think that Richard's 4a, 4b, 4c is an
improvement)".

My (Richard L's) contribution to the debate, such as it is, is to see
the bigger picture. Going thru the posts, I think it fair to say they
are generally about elimination methods. But these generally are about
safest ways of getting rid of losing candidates, and that sacrifices
preferential information. The objective should be to make the most use
of the information. This is what my old statistics lecturer told me
about statistics fifty years ago. (FAB STV is a statistical election
method.)
Even more broadly speaking, the attitude of finding a winner is what
Mill called maiorocracy, the tyranny of the majority, not democracy.
I know you are dealing with practical politics, I don't want to get in
the way of that (apologies if needed) but do you know your ultimate
destination? Not IRV, with or without CW, surely?

Richard L.

On 05/12/2019 01:50, robert bristow-johnson wrote: