[Bob Kuczewski]

I just found this bit of history from the HGAA voting discussion. I had saved it in my email system under the title "Misc Posts from Voting Committee up to page 48".

This appears to be a partial history of the posts to the very long "Voting" discussion. In that discussion, Jack Axaopoulos was strongly advocating the use of the "Range" voting system, and I was very much against it. Jack felt that the "Range" system would be used as if the votes were cast by impartial "Olympic Judges". But I argued that when people really care about an issue, they will vote for it based on the outcome they want most and will not be impartial at all. Jack countered by claiming that anyone voting with "extreme" values (which he could not define) would be seen as corrupt and somehow removed from office (again by a mechanism that he could not define). I argued that a voting system should reflect what people actually WANT and not what they somehow objectively "rate" as if they were an impartial observer. Voting is not Scoring.

In the end, I was proven right. Here's the actual vote by the HGAA Transition Team to Ban Scott Wise. Note that nearly everyone on Jack's team (wanting to ban Scott) voted 100 to ban, and 0 to not ban:

Vote_Count_Ban_Scott_from_HGAA.png (13.06 KiB) Viewed 1572 times

Note also that there were 5 members (shown in red) who wanted to ban Scott, but there were 6 members (shown in blue) who didn't want to ban Scott. Those 6 votes should have won (in Scott's favor) using any reasonable voting system (Plurality, Instant Runoff, or Condorcet). But the "Range" system pushed by Jack Axaopoulos and his cronies caused Scott to be banned by the minority vote of 5 against 6.

Take special note of how that happened. The final "Range" point score was 552 points to ban Scott and only 551 points not to ban him (he was banned). While there are many ways (among 11 voters) to account for the 2 points needed to reverse that outcome, the most obvious is to look at the votes by HGAAflyer and PilotGuy:

HGAAflyer:	1 to ban	100 to not ban
PilotGuy:	1 to ban	100 to not ban

As you'll read below, those two "1" values were obvious concessions to the continual threat by Jack Axaopoulos that anyone scoring with extremes (100s and 0s) would be seen as "corrupt" and "fired". Here are the direct quotes from Jack Axaopoulos during that discussion on the HGAA forum and prior to the vote to ban Scott:

And if you voted 100,0,0,0,0,0 with voting transparency, many would think you are gaming the system and corrupt.

If someone in a real vote, votes: 100,0,0,0,0 I know they are dishonest people with a serious character flaw.

As I pointed out before, I think its GREAT that it allows someone to show their corrupt character by scoring 0,100,0,0,0.

That argument is dead in the water. Go ahead and vote 100,0,0,0 using range if you arent too bright, but you WILL be judged on your honesty by other HGAA officials and members. You will get fired. Range allows us to detect who the immoral are. Thats a huge advantage of the system.

Now look at how Jack himself ("SG"), and his cronies voted:

SG:	100 to ban	0 to not ban
KK:	100 to ban	0 to not ban
JB:	100 to ban	0 to not ban
Dan:	100 to ban	0 to not ban

Right down the line, Jack and his supporters voted the full extremes of 100 for what they wanted and 0 for everything else. They did this regardless of Jack's many statements (above) about how this would be seen as "corrupt" and such voters would be "fired". In other words, "do as I say, not as I do." Did any of them get "fired" as Jack Axaopoulos claimed? Of course not. Jack Axaopoulos himself voted 100 to ban Scott and 0 to not ban Scott.

Here's the discussion from many weeks before that vote. Read it carefully to see who was lying and who was telling the truth:

Re: Interim Voting Committee - Session #1
by ThinAirDesigns » Thu Jun 24, 2010 1:33 am

One position that I am WITH Bob on:

IF (and it's a BIG if IMO) we wish to use a system that is not based on the age old and widely understood plurality runoffs, I would *most definitely* go with the Condorcet system that Bob has proposed. It has been widely tested and researched, has few gotchas and it has only two drawback as I see it:

A: it's unfamiliarity to the Average Joe and the inability for said Joe to quickly grasp how it works. After all, it's perception as a "black box" has kept it from being widely accepted as an established voting method.

B: it requires a second *non-Condorcet* backup system for the times there is no Condorcet winner. (it happens) This means that If you want what Bob is selling, you also have to buy some complete version of something that Scott or Scare or I am selling or you don't have a functioning system.

I am not against it because I feel it is flawed, but for reasons "A" and "B" above.

"Transparency" does little good if what you show isn't understood.

JB

Note that John Borton's objections are easily overcome by simply explaining that (A) the "Condorcet Winner" is the candidate who would defeat all other candidates if the ballots were evaluated for each pair of candidates, and (B) any voting system can produce a tie and needs a backup system in that case (flip a coin, take a second vote, use an alternative system, etc).

Re: Interim Voting Committee - Session #1
by ThinAirDesigns » Thu Jun 24, 2010 6:16 am

Hey, I consider this the way that good group decisions get made -- take smart people who respect each other but aren't afraid to bang it out and call it as they see it. In the end it brings things to light that people nodding politely around a table because they are afraid to hurt someone's feelings can never bring to light.

JB

Re: Interim Voting Committee - Session #1
by bobk » Thu Jun 24, 2010 4:40 pm

:
:

If voters prefer A over B,
And if voters prefer A over C,
And if voters prefer A over D,

Please tell me how you would justify naming either B or C or D as the winner. No matter which one you pick, I can demonstrate that voters would prefer A. So how can you justify B, C, or D?

Re: Interim Voting Committee - Session #1
Postby SeeMarkFly » Thu Jun 24, 2010 5:35 pm

So how can you justify B, C, or D?

Here is the justification, somebody HAS to win.
All these voting "systems" have problems when the votes are near a draw.
I seriously doubt it will be an issue and if it turns out to actually be a problem, can we deal with THAT in a targeted committee on that particular subject?
You know, a committee of people looking out for the best interests of hang gliders?

Mark's "justification" was that somebody HAS to win. Let's assume that's true. That still doesn't explain why that "someone" should be B, C, or D, and not A who can defeat each of the others in a runoff against them. In some ways, this was so simple, that it's hard to understand how people like Mark didn't get it. If any system picks candidate B as the winner (among A, B, C, and D), and it can be shown that the voters (using the ballots they've already cast) prefer A over B, then why would B be elected over A? That's simply a bad system.

Re: Interim Voting Committee - Session #1
by bobk » Thu Jun 24, 2010 7:17 pm

OK, I'm ready to cast my advisory vote as well.

I vote 100% for Condorcet/Range.

Now some might argue that's "selfish" since I didn't give the others any weighting. But I am 100% convinced that Condorcet is the best solution. There's no doubt in my mind. So that really does reflect my vote.
:
:

Re: Interim Voting Committee - Session #1
Postby sg » Thu Jun 24, 2010 7:49 pm

Bob,

In a real voting scenario, you couldnt see how everyone else is voting up front.
And if you voted 100,0,0,0,0,0 with voting transparency, many would think you are gaming the system and corrupt.
The fact that this system would uncover that is a BIG PLUS and the membership would have concrete evidence to vote you out.

Re: Interim Voting Committee - Session #1
Postby sg » Thu Jun 24, 2010 7:54 pm

I would also propose, any voting system we vote in, we try for a trial period of fixed time.
Then revisit to see if it really works well or not. I dont think any vote on a voting system should 100% close the door forever and forever.
Thats a principle I listed in my original slide deck. Lets try stuff... if its wrong... we can admit that and try something different.

So as I see it... when we vote for a voting system.... we are voting for a voting system experiment, that if proves out, then and only then does it "stick"

Again Jack Axaopoulos is demonstrating his lack of understanding of elections. People VOTE for what they want. They will vote any way they can (hopefully within the rules, but sometimes pushing them) to get what they want. If it's not "illegal" to cast a ballot with 100,0,0,0,0,0, then you can bet that someone will vote like that. If you outlaw the two extremes (100 and 0), then people will vote 99,1,1,1,1,1 for their favored choice. If you outlaw the new extremes (99 and 1), then you'll get 98,2,2,2,2,2. That's the whole purpose of elections. Elections are not the Olympics and they're not a beauty contest. Elections are a means for the voters to try to get what they want. It was never clear whether Jack simply didn't understand this or had some ulterior motive. But one way or another Jack Axaopoulos ended up killing the HGAA while the U.S. Hawks is about to celebrate its 16th year of continuous support of hang gliding.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Thu Jun 24, 2010 8:04 pm

I would also propose, any voting system we vote in, we try for a trial period of fixed time.
Then revisit to see if it really works well or not. I dont think any vote on a voting system should 100% close the door forever and forever.
Thats a principle I listed in my original slide deck. Lets try stuff... if its wrong... we can admit that and try something different.

So as I see it... when we vote for a voting system.... we are voting for a voting system experiment, that if proves out, then and only then does it "stick"

Totally agree.

JB

Re: Interim Voting Committee - Session #1
Postby bobk » Thu Jun 24, 2010 8:07 pm

I would also propose, any voting system we vote in, we try for a trial period of fixed time.
Then revisit to see if it really works well or not. I dont think any vote on a voting system should 100% close the door forever and forever.
Thats a principle I listed in my original slide deck. Lets try stuff... if its wrong... we can admit that and try something different.

So as I see it... when we vote for a voting system.... we are voting for a voting system experiment, that if proves out, then and only then does it "stick"

I agree with your sentiments here 100%.

But I disagree with your conclusion that my 100, 0, 0 ... vote is corrupt in any way. I really do believe that's the best system. I think it's the only system that should be used. Any other system may end up picking a candidate who could be beaten by another candidate (whether "candidate" is a person, or a mission statement, or a voting method).
:
:

Re: Interim Voting Committee - Session #1
Postby sg » Thu Jun 24, 2010 8:20 pm

But I disagree with your conclusion that my 100, 0, 0 ... vote is corrupt in any way.

I never said that.

Read my quote again:

many would THINK you are gaming the system and corrupt.

The fact is, you may be corrupt or a true believer. Its totally ambiguous, but many MAY still THINK you are corrupt, rightly or wrongly. In any case.... it says something about you. You are very black and white, with no grey scale in your mind. People could then decide if thats the type of person they want representing them.
:
:

Re: Interim Voting Committee - Session #1
Postby sg » Thu Jun 24, 2010 8:23 pm

Interesting quote about Olympic voting:

Normally, we think of Olympic judges as using range voting – the highest average score wins. (Albeit with some alterations such as dismissing the two highest and lowest "outlier" scores each performance, but the principle remains just averaging.)

I keep forgetting they are averaging and not totaling. DOH.

But its interesting that they throw out the 2 highest and lowest. Bobs "true believer" vote would get flushed in the olympics. And to further the analogy.... all HELL would break lose if you gave a skater a flat 0 and you would be fired as a judge for life [laughing] Seems like the system works [wink]

Re: Interim Voting Committee - Session #1
Postby bobk » Thu Jun 24, 2010 8:24 pm

In a real voting scenario, you couldnt see how everyone else is voting up front.

Yes, I know that. And if this was a real issue that I really wanted to win (like making me King of Hang Gliding forever - that's a joke, so don't quote me), then my tactic would have been to wait until the very last minute to spring my 100,0,0... on everyone. I did it early here because I wanted to demonstrate how the assumptions of "true voting" are not a good foundation for a voting system.

And if you voted 100,0,0,0,0,0 with voting transparency, many would think you are gaming the system and corrupt.
The fact that this system would uncover that is a BIG PLUS and the membership would have concrete evidence to vote you out.

Could I have given 1 or 2 percent to IRV? Maybe to show a preference over the others. But do I think IRV should be used instead of Condorcet? No. Just No. Just 100% No.

It is improper for anyone to assume that such a choice is somehow "corrupt" or does not reflect what I truly believe. You guys know that I've been standing up for things like open voting on the USHPA Board for a long time. You know that I've always stated what I believe is true. I've coded up an IRV system and shown how it fails. I've coded up a Condorcet system and seen how it handled a variety of cases. Has anyone else done that? I believe I have more grounds to be 100% sure of my vote than any of you do, and I currently believe Condorcet is the best way to go. Period. It is simply improper for anyone to ascribe the notion of "corruption" to my firm and unwavering convictions.

Re: Interim Voting Committee - Session #1
Postby sg » Thu Jun 24, 2010 8:31 pm

Bob... drop the corruption stuff. AGAIN.... im saying people may PERCEIVE you as corrupt *rightly or wrongly* if you voted straight 0's. Thats very different than actually being corrupt.

Also.... I find it very very hard to believe that you do dont have a single point of confidence more in say the scare method versus the scott method versus plurality versus runoff.

Surely you think one of these 4 other methods which you have gave all zero's to, are at least slightly superior to the rest?

If that is the case.... then straight 0's does not represent what you really believe, d and you voted incorrectly.
Unless you believe they are all 100% equal AND 100% completely worthless as voting systems.

Is that the case in your mind?

Re: Interim Voting Committee - Session #1
Postby SkyPilot » Thu Jun 24, 2010 9:45 pm

I think the most important question is:

What voting system are we going to use to vote on the voting system we are going to use?

Seriously, ALL of you guys are really very smart - and I say that with much respect! The discussion on voting methods is really fascinating and I'm not sure yet which one I prefer. Is there a deadline for this discussion and a time we will choose. Should we list up the current methods - with a short description of the algorithm - and then take a straw poll to see where we as TOs stand?

I suppose there is no perfect algorithm but the best one would produce the most "agreeable" outcomes on the any test cases to demonstrate how an algorithm will react. I think anything where it shows: A better then B, A better than C, A better than D - oh D wins would not be a good algorithm.

In regards to Bob's suggestion that there may be a need for 2 algorithms - I've been thinking about it. Is it possible that theoretically it is not possible to have a single perfect voting system that (a) adheres to the wishes of the majority (and I believe this is a desired requirement for a voting system) and works when there is NOT A CLEAR MAJORITY. Like I said, you guys are way smarter on this but I'm trying to think it out too.

BTW SG: In regards to voting on poll type questions can there be an ability to correct a vote before the voting deadline. There were some occasions when I voted a certain way but after "seeing the light" wanted to change my vote but couldn't.

Re: Interim Voting Committee - Session #1
Postby bobk » Thu Jun 24, 2010 9:50 pm

I think anything where it shows: A better then B, A better than C, A better than D - oh D wins would not be a good algorithm.

Finally, someone is paying attention. That's all I'm saying. If we run simple Condorcet first, it will tell us if there is one candidate that beats all others. If there is such a candidate, then it should be the unambiguous winner. If there isn't, then we've got a more ambiguous situation on our hands, and we'll have to use an algorithm to handle that ambiguity.

This is very simple. Thanks SkyPilot!!

Re: Interim Voting Committee - Session #1
Postby SkyPilot » Thu Jun 24, 2010 10:07 pm

I would also propose, any voting system we vote in, we try for a trial period of fixed time.
Then revisit to see if it really works well or not. I dont think any vote on a voting system should 100% close the door forever and forever.
Thats a principle I listed in my original slide deck. Lets try stuff... if its wrong... we can admit that and try something different.

So as I see it... when we vote for a voting system.... we are voting for a voting system experiment, that if proves out, then and only then does it "stick"

Just recording my concurrance on this. It wouldn't be fishy but an honest attempt to adopt the best system before it gets locked down in the HGAA founding documents. A voting system that chooses correctly AND maintains transparency in the voting process.

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 12:28 am

:
:
But most importantly, if more voters would rank the Condorcet winner above the Range winner when matched one on one, how do you justify not picking the Condorcet winner at that point? This isn't a rhetorical question. I'd really like to know how we (the HGAA) will justify picking B over A when the open ballots clearly favor A over B by 55% to 45%. What will we say? Will our answer be "Because that's our system"?

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 4:57 am

Yes, and we'll call it the "Ranondorality System (Range + Condorcet + Plurality)
I LIKE it!!

It sounds like you're violating the sincere voting constraint there JB!

I think you get my point though. By definition, the Condorcet winner (if it exists) will beat any other candidate in a one on one race (the simplest and easiest kind of election to understand and explain to our members). In other words, my candidate will beat your candidate any day of the week and twice on Sundays.

:
:

OK, enough damage has been done in these 23 pages. Let's see the candidates and vote on them.

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 10:13 am

:
:

Third, one of my fears with Range voting is that when people really have strong feelings, it will break down to the old plurality system where everyone votes 99 for what they want and a token 1 or 2 for what they don't want. With Condorcet the ordering means that once I've rated A above B on my ballot, then my ballot will always contribute to A beating B. So there's no incentive to leave B out of the rest of the ordering just because I want A much more than B. That's also true of C, D, E, and so on. So as long as I have my first choice (A in this example) listed first on my ballot, then I can honestly rank the other options without harming that first choice. That goes for the 2nd, 3rd, and other choices as well. I don't think that's true of Range voting since every vote for an alternative is actually being tallied against the first choice at the same time (they are competing against each other). So I believe there's a greater chance for Range voting to break down into a 99,1,1,1 type of plurality than there is for Condorcet to do so.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Fri Jun 25, 2010 1:37 pm

... one of my fears with Range voting is that when people really have strong feelings, it will break down to the old plurality system where everyone votes 99 for what they want and a token 1 or 2 for what they don't want.

A: I understand why you would fear this -- its the very first impulse you displayed when it came time to vote the mock ballot (and you've since defended it) - but I wouldn't project that on others. Right out of the gate on the mock ballot you voted 0,0,100 (meaning you considered IRV and Range EQUALLY valueless), even though just a few pages earlier you were wildly upset that I wouldn't change a personal and perfectly accurate post of mine to reflect your STRONG preferrence for Range over ANY of the other Condorcet backups. I think if most people strongly prefer "Range" as you do as a second choice, they will give "Range" an appropriately higher ranking, but that's just an opinion of mine.

B: It's nice to know that if folks DO game the system by exaggerating their vote (with nefarious intent or not) rather than rank their preferences according to their true feelings, the Range system is demonstrated to actually be MORE IMMUNE than Condorcet to the very worst of that behavior.
:
:

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 2:44 pm

Right out of the gate on the mock ballot you voted 0,0,100 (meaning you considered IRV and Range EQUALLY valueless)

If I'm standing on launch with two fully assembled gliders and one of them has a downtube that's been visibly bent and restraightened with obvious deformation, then I'm going to fly the undamaged glider 100% of the time. I don't say that the damaged glider is "valueless". I'm just never going to choose it over one that's undamaged. Never. In my opinion, these other systems all have bent downtubes. They're going to fail us when we need them the most.

[sarcasm alert] But keep on quoting from "rangevoting.org". Oh yes, that's a real unbiased opinion you're quoting from that source. If they say it, then it must be true!! [/sarcasm alert]

Can we vote yet? We might as well get it over with, drink our Kool Aide, and hope the dose is low enough to survive.

Re: Interim Voting Committee - Session #1
Postby sg » Fri Jun 25, 2010 2:58 pm

Right out of the gate on the mock ballot you voted 0,0,100 (meaning you considered IRV and Range EQUALLY valueless)

If I'm standing on launch with two fully assembled gliders and one of them has a downtube that's been visibly bent and restraightened with obvious deformation, then I'm going to fly the undamaged glider 100% of the time. I don't say that the damaged glider is "valueless". I'm just never going to choose it over one that's undamaged. Never. In my opinion, these other systems all have bent downtubes. They're going to fail us when we need them the most.

I don't say that the damaged glider is "valueless".

Actually, thats EXACTLY what you said when you scored it a zero.

Anyway pretty horrible analogy. You choose it to be as binary as possible but it doesnt reflect reality at all.

That would be like saying, you voted for Obama, therefore McCain didnt have ONE SINGLE quality about him you agreed with in ANY way, so you gave him a zero.

A valid analogy would be: Score the following sports cars Bob:

Porsche model X, Ferrari Model Y, Lambo Model Z
How would you answer?
Porsche: 100
Ferrari: 0
Lambo: 0

Because you like the porsche? Of course not. That would be absurd, they are all functional cars.

But what you did is disingenuous and has no bearing on what were discussion. A valid hang glider analogy would be:
Score the following: Falcon, Freedom, Horizon

But your analogy is equivalent to: Score the following: Falcon, BROKEN FREEDOM, BROKEN HORIZON, therefore 100, 0, 0

Makes no sense at all.

Re: Interim Voting Committee - Session #1
Postby sg » Fri Jun 25, 2010 2:59 pm

[sarcasm alert] But keep on quoting from "rangevoting.org". Oh yes, that's a real unbiased opinion you're quoting from that source. If they say it, then it must be true!! [/sarcasm alert]

That is a blatant poisoning the well fallacy. Is this how were gonna debate now?

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 3:01 pm

You may have missed my previous edit where I mentioned Kool Aide.

I think we're about to do something very foolish, and it appears that I am not going to be able to stop it.

Let's just vote and get it over with.

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 3:27 pm

JB, arguing with you and SG about Range Voting is like arguing about religion. It's only going to make all of us angry at each other.

If we had a year to debate this, then I'm sure I could convince you with theory and demonstrations and experiments. But we don't have a year, so we're going to do what we're going to do. I'm just hoping we can fly it to safety before it folds up on us.

Can we please vote?

Re: Interim Voting Committee - Session #1
Postby holger » Fri Jun 25, 2010 3:29 pm

If we pick a method that not all of us agree on, how is this as a compromise:

We use this method and if there was a different winner using the other method a runoff can be requested with only these two candidates, which we can handle without a discontent. Nothing will tell more about the different voting methods then our actual results on HGAA stuff. That should put Bob's concern at rest that we are about to make a cardinal mistake early on.

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 3:36 pm

If we pick a method that not all of us agree on, how is this as a compromise:

We use this method and if there was a different winner using the other method a runoff can be requested with only these two candidates, which we can handle without a discontent. Nothing will tell more about the different voting methods then our actual results on HGAA stuff. That should put Bob's concern at rest that we are about to make a cardinal mistake early on.

Well said Holger!!

The problem is (and you can refer to my post on page 22) that all systems are DEFINED to lose against Condorcet in a head to head, one man one vote runoff when a Condorcet winner exists. That's the Condorcet criteria!! So the people who want any other approach will NEVER agree to a one-man one-vote runoff because they've already lost ... by definition.

Re: Interim Voting Committee - Session #1
Postby sg » Fri Jun 25, 2010 3:40 pm

I think mixing stuff would complicate things.

You might have missed it, but in a previous post, I stated this should be a vote for an experiment, not the FINAL choice.

If Range were to win a vote right now, I say we use it for X months, and see if anything jumps out at us.
It may also be interesting to run other methods along side to see how it goes.

If we find that Range works and condorcet matches it for all of our votes, it would be a pleasant surprise, etc...

We may find Range totally sucks.... then we can try the next most voted for method or something....
But using one method, and another as a back up, or something like that, violates the KISS/ease-of-understanding principle some of us feel strongly about here.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Fri Jun 25, 2010 3:41 pm

If we pick a method that not all of us agree on, how is this as a compromise:

We use this method and if there was a different winner using the other method a runoff can be requested with only these two candidates, which we can handle without a discontent. Nothing will tell more about the different voting methods then our actual results on HGAA stuff. That should put Bob's concern at rest that we are about to make a cardinal mistake early on.

I would have no issue with that proposal being put to a vote, but I would suggest that we see the results of the selection vote before we do so. We may (or may not) find that there is a large majority agreement on one system or another in which case I would vote against such a proposal (even if I'm in the minority).

I don't think we need to consider bogging down what is already a herculean task with many, many votes ahead by counting votes different ways and having runoffs unless we truly do have a closely split vote on which system to use.

It's just a for a trial period anyway. If it turns out we don't like it, we can change it. That should be one of the Hallmarks of the HGAA -- we aren't afraid to admit mistakes.

JB

Re: Interim Voting Committee - Session #1
Postby holger » Fri Jun 25, 2010 3:47 pm

I think mixing stuff would complicate things.

You might have missed it, but in a previous post, I stated this should be a vote for an experiment, not the FINAL choice.

Haven't missed it, but I was offering a way to help us assess the experiment, and a second chance for those (few?) cases different methods produce different winners.

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 3:47 pm

I think mixing stuff would complicate things.

SG, whichever method we use, you know that you and I are going to be calculating and probably posting the results of the other system anyway. So that's not complicating things any more than will happen anyway.

Furthermore, if we use the Simple Condorcet, it will tell us instantly whether it can find an absolute winner or not. If it cannot, then the program simply calls the other method. My previous Java program could already do that since it had Plurality, Simple Condorcet, and IRV already programmed. It should be very simple to add Range Voting as well. Then we could compare all four.

All I'm asking is that we make our actual decisions on something that's not considered experimental. I still haven't seen any code for the Range system yet. Can you post it?

Re: Interim Voting Committee - Session #1
Postby sg » Fri Jun 25, 2010 3:50 pm

To clarify... by "mixing stuff" I did not mean that we should not run other systems along side. Thats perfectly fine if someone wants to do the work. Im sure some will, cool.

Im saying, once we pick the first system to try, it will be "range" or "condorcet" not some weird hybrid of "range followed by condorcet followed by X in case of a bad result" or whatever. We should only test things we are actually considering to use for real in the end.

Re: Interim Voting Committee - Session #1
Postby holger » Fri Jun 25, 2010 3:52 pm

I don't think we need to consider bogging down what is already a herculean task

Don't want to bog it down. We do business with the method we picked by vote. Say it's not Bob's. Bob determines his winner, and requests a runoff if he gets a different winner. May never happen, and until then no extra work for the team.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Fri Jun 25, 2010 3:54 pm

Haven't missed it, but I was offering a way to help us assess the experiment, and a second chance for those (few?) cases different methods produce different winners.

I think ANY system we pick other than the tried and true (and also known to be seriously flawed) plurality system should be considered merely a test and should be compared as we go along to other systems (I'm sure the proponents of the varios systems will be happy to keep us appraised as a matter of course).

I agree with Jack -- after we make this selection, schedule a revote on the voting system for some reasonable date in the future and then let the proponents present their cases based on the actual action of the Tteams votes.

JB

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 3:54 pm

I don't think we need to consider bogging down what is already a herculean task

Don't want to bog it down. We do business with the method we picked by vote. Say it's not Bob's. Bob determines his winner, and requests a runoff if he gets a different winner. May never happen, and until then no extra work for the team.

Bingo!!

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 3:58 pm

I think ANY system we pick other than the tried and true (and also known to be seriously flawed) plurality system should be considered merely a test and should be compared as we go along to other systems (I'm sure the proponents of the varios systems will be happy to keep us appraised as a matter of course).

Then you're saying that we "pick" plurality and allow all others to develop a track record as we move forward. I endorse that.

This means that our ballots would contain the needed information to satisfy both methods (which is easy enough), but only their top choice would be used for actual decisions until we choose an alternate system. I think that's a reasonable way to go.

Can we vote on that?

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Fri Jun 25, 2010 4:01 pm

Then you're saying that we "pick" plurality and allow all others to develop a track record as we move forward. I endorse that.

Nope -- you're misinterpreting my use of the word "test".

My suggstion is that we use the system that wins the upcoming vote for a given test period and then revote to make sure we are still comfortable with it (or not).

JB

Re: Interim Voting Committee - Session #1
Postby S C Wise » Fri Jun 25, 2010 4:05 pm

Stop debate!

Re: Interim Voting Committee - Session #1
Postby sg » Fri Jun 25, 2010 9:11 pm

Here is my feedback Scott.
:
:

The problem with this is that it introduces a problem that condorcet currently has.
The "youre throwing away your vote" issue. One of the GOOD things about Range is, it allows you to freely vote for who you really want without the worry that you are wasting your vote and causing the candidate you are most against to win by not giving your vote to the candidate most likely to win (who is not your favorite).

So if I were libertarian who hates democrats, I could vote like this in the last election:
Ron Paul:100
McCain:90
Obama:10

With our current election system and with condorcet, by voting for Ron Paul, I could accidentally cause Obama to win by robbing McCain of votes. That REALLY sucks not being able to actually vote the way you want to.

Re: Interim Voting Committee - Session #1
Postby bobk » Fri Jun 25, 2010 11:04 pm

They might think A is barely better than B, but that A is complete garbage, thus they weighted extremely low to communicate that.

They might think anything, and you're reading what you want into it. I think they didn't like any of the choices but felt that A was their best bet and that's why they voted 1000 rather than 0000. Are you telling me that you can read their minds now with Range voting?

But you vote A??? Bzzzzt.

That's the will of the many not the will of the few. Your method is letting a small vocal minority run the show against the clear and rather deliberate preference of a large majority. Your method violates "one man, one vote". Are you sure this isn't USHPA?

Scott? Can we decide how we're going to present our options and conduct the vote? Thanks.

Re: Interim Voting Committee - Session #1
Postby SkyPilot » Fri Jun 25, 2010 11:13 pm

Condorcet and the Transition Team

There have been some complaints that Condorcet might be difficult to explain and difficult to check. So I've come up with a real world example using the 17 members of the Transition Team voting to select one of four items (A, B, C, and D).
:
:

This really is not that hard to understand at all - rather simple and I think most people who CARE will understand. Those who DON'T CARE, and that may be of the type SG says have the attention span of a fruit fly will not understand ANY system - show me a fruit fly that knows basic arithmetic and I will amend my post

SG: Could it be possible that there is a fundamental difference between sports competition voting (i.e. therefore Condorset not good) and election voting (there Condorset good) and the argument is in regards to the applicability of a voting method for different "kinds" of voting?

Re: Interim Voting Committee - Session #1
Postby holger » Sat Jun 26, 2010 1:39 am

SG, how does the Range method deal with extremists? Imagine a range of centerline candidates but one extremist that gets strong votes from those that think in extremes (or have been bribed).

Re: Interim Voting Committee - Session #1
Postby sg » Sat Jun 26, 2010 2:10 am

SG, how does the Range method deal with extremists? Imagine a range of centerline candidates but one extremist that gets strong votes from those that think in extremes (or have been bribed).

2 answers to that

1) The studies and computer simulations done on range voting show thats it quite resistant to gaming. All systems can be gamed to some degree including condorcet, but range seems to have good resistance. Bob in fact tried to game a sample vote here and it failed. (and thats with a very small voting set, which should make it a lot easier, only 3? or4? of us voted). So we even have evidence its difficult to game right in this thread.

2) Since we will be using full transparency, and showing all the votes after they have all been cast, if someone *IS* trying to game this system, they get caught red handed, and thats something I consider a big plus. Range voting + transparency tends to highlight who the gamers are.
So if Bob again voted 100,0,0,0 in a real vote (instead of gaming to try and make a point), I wager Bob would lose a lot of support from constituents quite quickly. Imagine the Russion ice skate judge giving the American skater a perfect 0 :rofl: All hell would break loose.

Also, at the olympics, for example, they throw out the bottom 2 votes... so Bob's 0 vote when trying to game the system wouldnt count at all. It makes it such that you would need a lot more collusion to get any effect. But the bigger the vote pool, the more difficult it becomes to game. So if we really have a big concern about gaming, you could use a similar tactic, but maybe as a percentage of the vote pool size.

In any case.... a pro's con's summary of each system needs to be posted.

Re: Interim Voting Committee - Session #1
Postby sg » Sat Jun 26, 2010 2:28 am

SG: Could it be possible that there is a fundamental difference between sports competition voting (i.e. therefore Condorset not good) and election voting (there Condorset good) and the argument is in regards to the applicability of a voting method for different "kinds" of voting?

Nope. Range Voting was used during the longest running democracy in history.

History: http://rangevoting.org/OlympHist.html

Approval voting is a form of Range Voting... but if I remember correctly, the inventor of Approval Voting switched to the Range Voting camp.

Re: Interim Voting Committee - Session #1
Postby holger » Sat Jun 26, 2010 3:07 am

SG, how does the Range method deal with extremists? Imagine a range of centerline candidates but one extremist that gets strong votes from those that think in extremes (or have been bribed).

2 answers to that

Thank you, SG. I should read the literature you guys digested before asking questions, but if I may for now, I concerned about a different thing. We may find that the differences in the voting systems are not so much how robust they are against gaming or how fair they are, but what tendencies they support and which they dampen. The Range method seems to emphasis people with strong opinions. People with moderate choices carry less voting weight then those that are all for one direction and really dislike another. I find myself often in the moderate camp, thus my voting will have less influence.

IOW, the tendency goes towards solutions that have gang-ho supporters or that radically oppose an alternative. When I look at politics, that makes me uncomfortable.

BUT: This is a gut feeling concern. I really need to read up and understand this better.

Re: Interim Voting Committee - Session #1
Postby sg » Sat Jun 26, 2010 11:17 am

Whoa whoa.... the ONLY person here who tried to game a sample vote was BOB... not me.

Re: Interim Voting Committee - Session #1
Postby sg » Sat Jun 26, 2010 7:22 pm

Finally, I think SG mentioned that with this method people could still be influenced by a F.U.D. factor. But I don't see how this system/method has any more tendency toward that more than any other system we're considering.

It absolutely does. You are changing the system in a way that exactly causes the "throw away vote problem".
Your new version would have this problem.

Ill give you a concrete example.

Imagine you want to vote for Ron Paul, and you lean Republican.
If you pile your chips into the Ron Paul bucket, it leaves you with very few or none for the McCain bucket.
Ron Paul has very little chance of winning but YOU REALLY WANT TO VOTE FOR HIM, its who you believe in.
The Republicans will remind you that if you actually vote for Ron, you are THROWING AWAY YOUR VOTE, and you could cause Obama to win.
(Previous analysis of elections has shown this has actually happened in US races)

With the original Range Vote, you could vote: Ron Paul:100, and McCain:100.
It allows you to vote your conscience.

With your system, you would to vote:
Ron Paul:80, McCain:20 Obama:0, or something like that. Maybe:
Ron Paul:100, McCain:0 Obama:0 if you are extreme.

You CANNOT vote your true conscience. Your new system, like condorcet, virtually prevents you from voting for who you really want to vote for, because youre throwing away your vote on someone unlikely to win. Of course, maybe the very reason he is unlikely to win is the very same F.U.D. factor of throwing away your vote.

A system that doesnt allow me to actually vote my conscience, out of fear of causing the worst opponent (in my mind) to win, is very flawed.

Note: The previous post by Jack Axaopoulos shows either how poorly he understands the Condorcet system being proposed or how dishonest he is. Jack claims that Condorcet "virtually prevents you from voting for who you really want to vote for, because youre throwing away your vote on someone unlikely to win". That's simply not true. In fact, one of the strengths of the Condorcet system is that each candidate is run against every other candidate on a one-on-one basis. Using Jack's example, he could easily cast his ballot as:

        Paul    McCain    Obama

So when Paul was compared against McCain, Jack's ballot would choose Paul. When Paul was compared against Obama, Jack's ballot would choose Paul. When McCain was compared against Obama, Jack's ballot would choose McCain. If those were Jack's preferences, then that ballot would express his wishes completely in all pairings of all candidates. The Condorcet Criterion states that if any one of the candidates should defeat each and every other candidate in all pairings, then that Canididate should be the winner. It's as simple as that. Jack Axaopoulos was either speaking without understanding what he was saying or outright lying. Neither are good qualifications for leadership in any organization.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Sun Jun 27, 2010 12:37 am

JB, Simple Condorcet simply states that any candidate who can defeat all other candidates in a one-on-one election (simple majority) would be the winner. In each of those elections, you can vote for A>B, A<B, or A=B. Does that help?

I don't need any "help" Bob -- I'm quite confident that I understand how the simple Condorcet system works. What I need is for you to stop saying that it does things that it doesn't.

Here is my assertion that you have said was wrong, but continue to prove it's correct with every post you make:

By using 1-7 you still are giving me the the ability to rank choices as ties if I wish (7, 3, 3) - can't to that with Condorcet, ...


Call it whatever name you wish, but the Condorcet system you have proposed and demonstrated DOES NOT ALLOW ME TO RANK CHOICES AS TIES -- it has no mechanism whatsoever use that information and a vote in the #3 spot will beat a vote in the #4 spot EVERY TIME -- that's the very definition of "not a tie".

Now, that's fine - that just one of the differences between the Condorcet system and other systems, SO JUST STOP PRETENDING IT CAN!

************************
Scott, if the "simple Condorcet" system as Bob has proposed it is so complicated that someone as familiar with the system as Bob can be innocently confused by the above point and claim it wrong when in fact it is correct, then:

B: It's far too complicated for Average Joe to ever understand

A: I have some waterfront property ...

Note: The proposed Simple Condorcet voting system does indeed allow assigning of choices as "ties". In my proposed notation, a voter would express a tie with an "=" sign. So if a voter wanted A first, B second, C, D, and E third, and F last, their ballot would be: [A,B,C=D=E,F]. The pairwise comparisons between candidates can easily handle those cases. You can find that software at either of these two locations:

https://ushawks.org/voting/
https://bobkuczewski.github.io/Voting/Voting.html

Re: Interim Voting Committee - Session #1
Postby sg » Sun Jun 27, 2010 1:32 am

What I seriously reject is allowing individual voters to concentrate or dilute the power of their individual say in community decisions. To me
that's either dishonest (being able to effectively vote twice) or irresponsible and negligent (to effectively vote less than one "full" vote).

Point 1) They are no more concentrating their votes than olympic judges are when they give 5 different competitors 9.5+ ratings.
I feel like youre completely missing the philosophy behind range voting. Its like you are only considering the GAMING cases only, which should be rare, especially when each representative has their voting record shown and will look like a douchebag if they vote 100,100,0,0,0,0,0. [laughing]

Again.... there is NO double or triple voting, etc, voting in Range. You are independently giving a SCORE to each candidate based on how much you like them, or they match your positions or whatever. That would be the equivalent of claiming an olympic judge who scores 9.5,9.23,9.87,9.77 got quadruple votes. They did not. They score each candidate independently JUST LIKE every other judge.

Point 2) Its not dishonest, irresponsible, or negligent. Its the TRUTH of the situation. A voting system that allows you to ***accurately express your choices*** is really important IMO.
If, like MANY voters, I lack the time to research each candidate 110%, and I vote 100, 95, 0,0,0,0, or 100,95,abstain,abstain,abstain,abstain, then that is a very accurate representation of what I want. I mean, come on... do you REALLY believe everyone who voted for Obama knew every other candidate on the ballot inside and out in the last election? I bet there were people that knew almost absolutely nothing but the other candidates.

Here is the really important point. If you do NOT give me a way to express my vote accurately, by forcing 100 or 0 on everything, or not using weighting at all, you are forcing me to be dishonest/irresponsible/negligent. You have it exactly backwards. Think about it.

Imagine if what I REALLY BELIEVE is: 100,95, abstain,abstain,abstain. For example, by using condorcet, and making me order the candidates, they get equal weightings and therefore *DO NOT* reflect my beliefs at all. If we force me to use 100 or 0 everywhere, again, im being forced to plug in numbers that DO NOT reflect my voting wishes. If you give me a limited set of points to you, again, you force me to score in a way that does NOT reflect my beliefs.
THEREFORE: the vote is NOT accurate because you are literally preventing me from voting correctly.

The real power of Range Voting, is that im free to actually express what I really believe without being punished for throwing away a vote or whatever.
Condorcet and the other systems here fall flat in this area. No other system I have seen so far allows me to ACCURATELY EXPRESS MY VOTE. They lack the data to do so.

Re: Interim Voting Committee - Session #1
Postby SeeMarkFly » Sun Jun 27, 2010 1:30 pm

We need a "voting system".
I had already made my decision by page 17 and twenty more pages of "how simple the other system is" did not seem to help.
I see a need to get everyone "out of the clouds' and back on this planet.
I do not care to custom build a new internet voting system (at this time).
let's pick one system and USE it.....Task one completed.

At this point, if you still want to try to "impress" me, I have a "down to earth" proposal.

Let's take the minutes of the last meeting of the USHpA (Bob should have a copy).
These are real time decisions that have to be made in THIS sport.
Now explain how any one of the "voting systems" that we are looking at:
Condorcet/ANY combo
IRV
Plurality
Range
Runoff
Scare's custom combo
Scott's custom combo
would have changed any one of the votes (other than one system taking more time).

Most voting is Yea or Nay.
I do find it admirable that you managed to solve the "ambiguous paradox" that seems to show up every other decade.
We are already miles ahead of our competition by just making votes public.

Re: Interim Voting Committee - Session #1
Postby sg » Sun Jun 27, 2010 9:48 pm

PS: If you are so convinced everyone would game the system with pure Range, you can limit their voting to 100 or 0. OR 1 or 0.
Thats called Approval Voting, a form of Range voting.
http://en.wikipedia.org/wiki/Approval_voting
Of course, it introduces a bunch of problems you can read about in that link, but it solves your main issue.

That said.... I think youre blowing the gaming thing way out of proportion.
Condorcet is totally gameable too.

But more importantly... do you REALLY think people here are going to game with transparency???

I like the Range+abstain method right now. So people are not forced to score a 0 or whatever.

So using that model. Do you TRULY believe that someone here is going to game and vote like this:

100,0,0,0,0
instead of:
100, abstain,abstain,abstain,abstain, which is still pretty douchey, but not as horrible as the first choice???
or maybe even vote honestly: 95,76,45,30,28 ???

Because if you REALLY think that..... then that makes me want to have Range even more. WHY?
Because it will reveal peoples character. If someone in a real vote, votes: 100,0,0,0,0 I know they are dishonest people with a serious character flaw.

With condorcet.... this is hidden from me.

And that, my friend is, EFFECTIVELY, just what we have with the USHPA Executive Committee.

Not even close. They couldnt get away with this if their voting was made public using Range. They CAN get away with it using Condorcet by always putting who they want to lose in last place and it would be nearly undetectable.

Re: Interim Voting Committee - Session #1
Postby sg » Mon Jun 28, 2010 3:14 am

Scott, here is a simple question.

Lets say you, JB and I are using Range to vote for a single candidate.
Just one. We only get one option.

I give the candidate an 53.
You give the candidate a 97.
JB gives the candidate a 68.

Did you really vote with more voting power than anyone else? Or did we all simply rate this candidate?

SG,

Without choices you simply are not voting - period.

Exactly... you are SCORING this one candidate. So we (you, me, JB) simply rated this candidate.
Just like an olympic judge scores a single skater. We have not voted... YET.


Now if I ask you to score a 2nd candidate, and you give them a 57, you are again simply individually SCORING how good of a candidate you think they are.

Now if I take all of your individual skaters ... errr... candidate scores together, NOW I have your vote.

But who cares..... your mind is clearly already made up.

The problem I have is your assertions that this will be horribly gamed, even with transparency but the fact is, we HAVE NO EVIDENCE OF THAT. Its never been tried. Do you have any actual data or are you simply guessing???

As I pointed out before, I think its GREAT that it allows someone to show their corrupt character by scoring 0,100,0,0,0. If anyone were to vote like that in a test period of Range Voting, id lose respect for them and wouldnt want them to be part of this group. Range Voting transparency speaks VOLUMES about the person voting. Im explicitly talking about Range+Abstain in this post.

Unfortunately, no other method has the ability to reveal this of someone.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Mon Jun 28, 2010 3:29 am

To game Condorcet, I must merely list my opponents proposal #4 each and every time (in a 4 way race). I can always explain it away perfectly well by saying "his proposal really was excellent, there were many things I really liked about it -- I just honestly thought the others were just a tiny bit stronger". In truth, my #4 vote for the proposal has actually given it a big fat ZERO (it can't beat any other choice).

To game Range, I must actually give those proposals a VERY low score and with this transparency a pattern emerges that give insight into the voter.

Now, I truly currently have no opinion on how easy it is to game the system you have proposed Scott -- I'm looking forward to your treatment of that mock ballot that Bob gamed so easily when normalized. If your proposal is different then that, it will be rather easy to get my support.

JB

Re: Interim Voting Committee - Session #1
Postby S C Wise » Mon Jun 28, 2010 4:05 am

. . . I like the Range+abstain method right now. So people are not forced to score a 0 or whatever.

I would use abstaining in connection with not voting at all.

So using that model. Do you TRULY believe that someone here is going to game and vote like this:

100,0,0,0,0
instead of:
100, abstain,abstain,abstain,abstain, which is still pretty douchey, but not as horrible as the first choice??? . . .

To tell you the truth, I can see many people voting honestly with your above pattern. If they like the idea of picking ONE definitive favored choice (as they are used to doing with traditional voting systems) then it makes complete sense simply to "put your money" on the horse you want to win. They may even feel that it's just too much bother to waste 10 minutes assigning point values between 0 and 100 to 3, 4 or more possible choices.

I, myself, would consider voting that way. Partially because I may want to express my solid (and honest) support for one choice. But, in addition, I don't like the idea that one or more voters may load the vote against my choice. Therefore, the best course is to fight fire with fire. BTW - I've flown a couple hang gliding comps and a lesson I was taught more than once was to use the rules to your advantage. The rules define the limits and oftentimes the limits must be pressed to win. I think that applies to voting as well.

Also, I think it's more telling to notice what people vote for as compared to how they vote for it.

Because if you REALLY think that..... then that makes me want to have Range even more. WHY?
Because it will reveal peoples character. If someone in a real vote, votes: 100,0,0,0,0 I know they are dishonest people with a serious character flaw . . .

I don't think one particular pattern can necessarily identify any character flaws at all. As I just described, there are many good reasons for people to place their full bet on their favorite choice. That's not dishonest at all, it's In Your Face Honest.

Re: Interim Voting Committee - Session #1
Postby sg » Mon Jun 28, 2010 11:43 am

So using that model. Do you TRULY believe that someone here is going to game and vote like this:

100,0,0,0,0
instead of:
100, abstain,abstain,abstain,abstain, which is still pretty douchey, but not as horrible as the first choice??? . . .


To tell you the truth, I can see many people voting honestly with your above pattern. If they like the idea of picking ONE definitive favored choice (as they are used to doing with traditional voting systems) then it makes complete sense simply to "put your money" on the horse you want to win. They may even feel that it's just too much bother to waste 10 minutes assigning point values between 0 and 100 to 3, 4 or more possible choices.

The ABSTAIN vote exists exactly for this case. "Its too much bother", or "I dont know anything about this candidate".

Why would you lie and score someone a 0? If an ice skating judge decided to take a nap, he wouldnt give a 0. He would be SHOT for this. He could abstain instead. Yes, it makes complete sense to CHEAT in any voting system. Thats what you really are saying here. After all, it could get you a win [smile]

I, myself, would consider voting that way. Partially because I may want to express my solid (and honest) support for one choice. But, in addition, I don't like the idea that one or more voters may load the vote against my choice. Therefore, the best course is to fight fire with fire. BTW - I've flown a couple hang gliding comps and a lesson I was taught more than once was to use the rules to your advantage. The rules define the limits and oftentimes the limits must be pressed to win. I think that applies to voting as well.

Scott... if you read the literature, you just defined "strategic voting". Its *IS* being dishonest. You arent voting the truth, you are gaming with purpose.

I don't think one particular pattern can necessarily identify any character flaws at all. As I just described, there are many good reasons for people to place their full bet on their favorite choice. That's not dishonest at all, it's In Your Face Honest.

You perfectly described dishonest voting. The reason to place a full bet on one candidate, and knock down every ther on purpose is because you want to game the system. Thats as dishonest as you can get in Range Voting.

Its the same as if I change the order of my condorcet vote just to force one particular candidate to lose as many head to head comps as possible by putting him at the end of my list.

I think were arguing semantics at this point. What im calling gaming/cheating, you are calling strategy/pressing-the-rules.

If your TRUE HONEST ranking of candidates is 97,76,54,45 and you place a vote of 0,1000,0,0, you *ARE* being as dishonest as Range Voting allows you to be by DEFINITION and they have have a term for this in the literature. Strategic Voting.
You will also be called out on it, and there is NO WAY you could defend a zero vote for the same reason an olympic judge could never defend a zero vote. You couldnt convince anyone that 4 of the candidates were perfectly equal AND 100% purely horribly.
Youre a gamer/dishonest-voter/stragic-voter, plain and simple.

I guess a real easy question for you should be: How WOULD you game Rank Voting?? What would it look like? You cant say 0,100,0,0 [wink]

Note: Jack's insertion of "abstain" into the Range voting system is undefined in this discussion and in general. In order to compare summed Range scores for two different candidates, some score would have to be assigned. Would an abstained candidate simply get no score on that ballot? Isn't that exactly the same as giving them zero points? Maybe it should be 50? What if other "non-abstained" scores were lower than 50? Would the "abstain" candidates then be scored higher than the non-abstained candidates? Maybe the abstained candidates would get a score equal to the lowest score on that ballot, or half the lowest score on the ballot? Maybe 1/3 or 1/4 the lowest score on the ballot? Obviously Jack hadn't considered this and many other important aspects of what he was proposing.

Re: Interim Voting Committee - Session #1
Postby sg » Mon Jun 28, 2010 12:23 pm

Bob,

You made one really important point that needs addressing.
That if we fight this hard over something like a voting system, all is lost.

I actually think there is no truth to that. I actually think you perceived something that isnt there at all.

What REALLY happened here is, we stepped into a tar baby from the start.
We all assumed picking a voting system would be easy and ran down that path.

What we fell into is an EXTREMELY COMPLICATED and subtle subject matter that has been fiercely debated for centuries.

I think it has very little to do with the people involved, but mostly to do with the subject matter at hand.

We might as well have decided that we were going to solve the philosophical problem of subjective versus objective morality in the universe before moving forward :rofl: Or settling the gun debate :rofl: Or when life actually starts in the abortion debate. Any subject matter of this sort would cause a long drawn out and heated debate.

THATS what I believe happened here. So I dont think you need to lose any faith at all.

Note: Bob ended up being correct. All was lost. The petty ego of Jack Axaopoulos drove this issue into the ground. The evidence in this thread is clear that Bob continually called for a vote to get it over. But Jack Axaopoulos dragged it out and used it as an excuse to hold a witch hunt to ban a number of key people from both the HGAA and HangGliding.org. Jack's actions killed the HGAA movement that Bob had started. But the HGAA was reborn as the U.S. Hawks and continues to this day.

Re: Interim Voting Committee - Session #1
Postby hgxc » Mon Jun 28, 2010 2:11 pm

Bob,

You made one really important point that needs addressing.
That if we fight this hard over something like a voting system, all is lost.

:
:

THATS what I believe happened here. So I dont think you need to lose any faith at all.

I agree it is a difficult topic and worse it has the threat of sapping the energy out of the enthusiasm of developing the new club. A side committee at the right time should be designated to sort out the pluses and minus of different voting systems and make a recommendation. The forum notes should be available to all so that we can see the debate but to place this as the main event can only be distracting. We have already heard the lier lier pants on fire accusations and we just begun.

Just my take on this,

Dennis

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Mon Jun 28, 2010 4:26 pm

I'll make a generic comment on this topic - I don't feel that this issue is one of 'make or break' HGAA. I really don't.

I suggest the Chair simply declare that we will use a long established voting method - say plurality, or runoff if Chair desires wins to be by majority (plain runoff, not IRV). This would be a 'just for now' decision and then a sub-committee could in time put together a good report for all the proposed systems.

It's clearly not going to be an easy or quick decision (as many thought) so I trust Scott enough to just make the declaration and move us forward.

But that's just me.

JB

Re: Interim Voting Committee - Session #1
Postby bobk » Mon Jun 28, 2010 5:15 pm

Bob, I read your last post over like 10 times trying to identify exactly the condorcet method you are using.

I *THINK* you are using MiniMax(the technical name for simple condorcet) with ties, with a "winning votes" scoring implementation. Since I doubt you invented a new one.

Please verify here:
http://en.wikipedia.org/wiki/Minimax_Condorcet

Thanks for the reference. But I'm not proposing that system either. I'm just using the Condorcet criteria itself. That's why it's so simple. I believe the MiniMax method will find the Condorcet Winner if it exists, but will find a winner with similar properties even if there is no candidate that meets the Condorcet Criteria.

Let me say it another way. The Condorcet Criteria is that the winner must defeat all others in all pairwise run-offs. If there is no Candidate that meets the Condorcet Criteria, then all of the methods I've seen will try to find a winner that's close. So, for example, if I would have to disregard (or even change) 1 vote to get A to be the Condorcet winner, but I'd have to disregard (or even change) at least 2 votes to get all other candidates to be the Condorcet winner, then I could make the case that A is CLOSER to being the Condorcet winner than any of the other candidates. But A is still NOT the Condorcet winner because A does not satisfy the Concorcet Criteria. A might be the Condorcet MiniMax winner or the Condorcet Schulze winner or some other winner, but A would still not satisfy the basic Condorcet Criteria.

So what I'm calling "Simple Condorcet" is just enforcing the Condorcet Criteria itself. It might be better to call it "Strict Condorcet" because it requires the winner to strictly meet the Condorcet Criteria.

Does that help?

By the way, I do agree that this is a topic that's been debated (and not resolved) for centuries, but I also think we'll have other topics that will be just as heated (if not more so) which are directly related to the sport of hang gliding. We can be split along many lines, and there are people who want to see that happen (imagine the delight on some faces when they're reading this topic). I don't believe SG is one of them, and I don't think JB is either. But they're out there, and we should not play into their hands by being so hard on each other.

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Mon Jul 12, 2010 10:52 pm

I would like to suggest that the Chair call for a straw pole vote on the topic of the voting system to use during the formation phase of the HGAA.

If the straw pole shows a clear majority preference for one system over the other, then I would suggest he quickly call for binding vote.

If no majority preference is evident, I would suggest that the Chair send the discussion to sub-committee and allow decisions to be made for the time being through simple plurality vote.

JB

Re: Interim Voting Committee - Session #1
Postby bobk » Mon Jul 12, 2010 11:10 pm

I would like to suggest that the Chair call for a straw pole vote on the topic of the voting system to use during the formation phase of the HGAA.
...
JB

I'd like to second JB's request. Now is the time to make a decision and accept the consequences. If we find that we don't like what we've done, then we can change it and exchange the appropriate "I told you so's" at that time.

Onward and upward!!! Thanks JB.

By the way, I've been doing a lot of thinking about the Range system, and I think its main weakness is that it assumes a degree of impartiality in the ratings. I don't think we can assume that in general, and the result is that someone who rates choices dishonestly is rewarded over someone who rates choices honestly. Here's the simplest example - 2 choices with 2 voters.

Voter #1 votes A[80] B[70]

Voter #2 (if honest) would similarly vote the opposite: A[70] B[80]

But if Voter #2 wants to win, then they can "tilt" their honest opinion to be slightly less honest: A[60] B[90].

No one will question a 60/90 vote. It's not as obvious as a 100/0 vote. But it will tend to win. And that's the biggest problem I see. With the Range system, every point you give to a competing choice is actually hurting your first choice. This rewards people who are willing to "tilt" their votes and punishes people who really do try to be honest. That's just not a good system. It may be fine for a tie breaker or a back up, but I think it's too easy for people to play games with their numbers and that's why I don't think it should be our primary system.

But let's just choose something!!! Anything!!!!!!!

Re: Interim Voting Committee - Session #1
Postby ThinAirDesigns » Tue Jul 13, 2010 12:17 am

I find it pretty freakin' hilarious Bob that you claim the Range system is not a good system and yet from a myriad of options you think highly enough of it to consider it as your SECOND OVERALL choice as evidenced by you picking it to backup the Condorcet system.

In fact, it's such an IMPORTANT second choice to you that you threw a hissy fit and eluded to taking your toys and starting a *third* competing HG org when I wouldn't list it as YOUR specific choice on a list of generic options which already included it.

Can't rationally have it both ways BOB -- if it's good enough to be part of your overall preferred system (which you scratch and claw for) , you can hardly claim it's a bad system.

JB

Note: What actually happened was that JB (John Borton) stepped in as Transition Team Chairman when Scott was on vacation. JB unilaterally posted a set of choices for the vote which he picked out of (coincidentally) "thin air" with no discussion among the Transition Team. JB's choices didn't include the Condorcet/Range system (Condorcet with Range as a tie breaker) that I had been advocating. I asked JB to include that option and he refused. I did (and still do) believe that Range voting is a very bad system, and that was proven correct in the banning of Scott. However, when there is a tie, almost ANY system (including a coin toss) can be acceptable, and I was supporting the Condorcet/Range system in an effort to be inclusive of those (especially Jack Axaopoulos) who were adamant fans of Range voting.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 2:01 am

Third..... agree with JB. Lets do the quickie vote and see if there is an obvious consensus.
Time to get the show rolling again.

Once the voting system in place.... things can really start to move. Even during the summer with lots of people busy

(Ive been in vegas since last wed, im back now)

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 5:43 am

I find it pretty freakin' hilarious Bob that you claim the Range system is not a good system and yet from a myriad of options you think highly enough of it to consider it as your SECOND OVERALL choice as evidenced by you picking it to backup the Condorcet system.

In any decision space, there are broad areas where the best decision is clear. In between those broad clear areas are the gray zones. I don't mind making mistakes in the gray zones because they're already gray. But I don't want to take risks with the broad clear areas in order to somehow get better results in the gray areas.

When there is a Condorcet winner, then we're in a broad clear area and we should go with it - always. But when we're in the gray area, then we can flip a coin. Even if it isn't a perfect coin, we're already in an area where we know the results are less meaningful because we're near that 50/50 border. We're going to have about 50% unhappy either way. When elections are close, we just have to live with that reality.

There was a big fuss about hanging chads in the 2000 election, and that created a big push toward automated machines where every single vote would be counted perfectly. What they didn't understand (and still don't) is that by trying to get the most accurate results for that last 1/10th of a percent, they're opening themselves up to the possibility of massive electronic election fraud on an order of 100 times what they're trying to correct. I would rather sacrifice some small scale accuracy to have a system that's hard to fool when the results are decisive than the other way around. I hope that helps explain my position ... even if it isn't quite as hilarious. [smile]

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 6:28 pm

Again, I don't think you can infer dishonesty or strategic voting from a set of numbers. You're effectively saying that it's impossible for someone to want one choice with 100% certainty and to not want all the others with 100% certainty. That's not fair, and to bust them for that is also not fair. If it's really not possible to want 100,0,0,0, then the voting system should not allow 100 or 0. But then you'll say that 99,1,1,1 isn't possible for an honest voter and we have to eliminate 99 and 1 as well. Where do you stop?


This is where we disagree the most. 100,0,0,0 is blatant strategic voting, so is 99,1,1,1.

The olympic judges example shows exactly how well range voting works ***WHEN YOU HAVE FULL TRANSPARENCY***.

Olympic judges have strong conflicts of interest. The USA judge votes for Russian skaters and the Russian judge votes for USA skaters.
According to the arguments against Range I see here, the Russian judge would give the USA skater a 0, and the USA judge would give the Russian skater a 0 too. Or maybe a 1 in a silly attempt to be covert

***BUT IT DOESNT HAPPEN***. There vote is in full view, and its obvious that they are strategic voting and would lose their positions as judges.
The claim that "100,0,0,0,0" is how I REALLY feel is simply nonsense. If an olympic judge used that excuse he would be laughed out of the building and so would an HGAA representative.

That argument is dead in the water. Go ahead and vote 100,0,0,0 using range if you arent too bright, but you WILL be judged on your honesty by other HGAA officials and members. You will get fired. Range allows us to detect who the immoral are. Thats a huge advantage of the system.
Its also the most expressively accurate of any system. It NAILS exactly how a voter rates a candidate. Condorcet only gives us order.

A100 B98 C95 D0
and
A100 B15 C20 D19

are VERY VERY different opinions.Yet condorcet flushes this valuable data and pretends they are EXACTLY the same. They are not. Thats a horribly inaccurate and does not truly represent what the voters actually believe. Isnt that the WHOLE GOAL of a voting system to accurately discover what the voters want? I think so. Only a Range system allows for this. Condorcet captures too little data.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 6:40 pm

Who should run Torrey? The choices are: Rob McKenzie, David Jebb, Gabe Jebb, or Robin Marion.

Here's my vote: 100 / 0 / 0 / 0

You can call me dishonest if you want, but I would vote 100% for Rob and 0 for the others because I'd want to maximize Rob's chances of winning.

When an issue is important, that's how people will vote. They may throw in a 1 or a 5 or a 10 for window dressing, but that 100/0/0/0 is probably more reflective of their views than 100/1/5/10. So which is really more dishonest?

P.S. You wrote: "Its easy to think up of many cases where the vote would be dead wrong with condorcet."

I'd honestly like to see a few examples of this. Please post your best example that shows where Condorcet would be dead wrong. Thanks.

Note: Neither Jack Axaopoulos nor John Borton nor anyone else ever produced a single example where the proposed Condorcet method would be "dead wrong". It's impossible because the Condorcet criteria only selects a winner that would beat every other candidate in one-on-one runoff elections using the same ballots.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 7:04 pm

Who should run Torrey? The choices are: Rob McKenzie, David Jebb, Gabe Jebb, or Robin Marion.

Here's my vote: 100 / 0 / 0 / 0

You can call me dishonest if you want, but I would vote 100% for Rob and 0 for the others because I'd want to maximize Rob's chances of winning.

BAM! Strategic voting, right there, by definition. Youre fired! [laughing] See, the system worked perfectly [smile]

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 7:17 pm

Who should run Torrey? The choices are: Rob McKenzie, David Jebb, Gabe Jebb, or Robin Marion.

Here's my vote: 100 / 0 / 0 / 0

You can call me dishonest if you want, but I would vote 100% for Rob and 0 for the others because I'd want to maximize Rob's chances of winning.

BAM! Strategic voting, right there, by definition. Youre fired! See, the system worked perfectly

So I'm fired for voting for what I wanted? Where are the "rules" that tell me which voting patterns will get me fired and which won't?

How about showing me an "acceptable" vote that expresses my admiration for Rob and my disgust for the other 3? Please show me what's "legal" and won't get me fired. Thanks.

Note: The reply by Jack Axaopoulos (immediately below) demonstrated his naieve confusion between "voting" and "scoring". When voting, people are attempting to get a result that they (or their constituents) find most favorable. That's not the same as scoring.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 7:22 pm

No, your not fired for voting what you wanted.
Youre fired because you openly admitted to strategic voting. You literally told us you gamed the voted with:

...because I'd want to maximize Rob's chances of winning.

If an olympic judge said: "I gave everyone else really low scores because I wanted my Russian skater to win", he would be immediately fired too.

Perfect analogy. Its so simple. The instructions are: "Vote as if youre an olympic judge."
You independently give each candidate a rating. If you cheat, people will notice as you just proved, and you will suffer the consequences.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 7:29 pm

As you pointed out, your example boils down to a choice between A (Condorcet winner) and D (Range winner).

Voter #1 A3 B2 C1 D0 (prefers A)
Voter #2 A3 B2 C1 D0 (prefers A)
Voter #3 A3 B2 C1 D0 (prefers A)
Voter #4 A3 B2 C1 D0 (prefers A)
Voter #5 A3 B2 C1 D0 (prefers A)

Voter #6 D80 A3 B2 C1 (prefers D)
Voter #7 D80 A3 B2 C1 (prefers D)
Voter #8 D80 A3 B2 C1 (prefers D)
Voter #8 D80 A3 B2 C1 (prefers D)

So if you held a runoff election between A and D, then A would get 5 votes and D would get 4 votes. Just because the D supporters yelled louder does not give them the right to dominate the election.

Imagine David Jebb running against Rob McKenzie. We all know that Jebb would get his clipboard army to vote 100/0 right down the line. Rob might not be so ruthless and would end up losing. Sure you can wag your finger at all those Jebb voters and say they were "gaming" the system, but that doesn't change the fact that Jebb wins and rules the roost. How do you propose to invalidate all those 100/0 votes? This isn't a theoretical question, I'm really asking what criteria are you going to use to keep that from happening?

Note: Again and again Jack Axaopoulos refused to answer the question as to how he would keep people from voting with extreme scores in order to "game" the Range voting system.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 7:34 pm

So if you held a runoff election between A and D, then A would get 5 votes and D would get 4 votes. Just because the D supporters yelled louder does not give them the right to dominate the election.

Totally inaccurate. No one yelled louder. Thats like claiming that an olympic judge who gaves the USA skater a 9.2, yelled louder than when he gave the Russian skater a 9.1. Doesnt make sense.

A group of voters rated D highly. No group of voters rated A highly. Cant get any clearer than that.

Imagine David Jebb running against Rob McKenzie. We all know that Jebb would get his clipboard army to vote 100/0 right down the line. Rob might not be so ruthless and would end up losing. Sure you can wag your finger at all those Jebb voters and say they were "gaming" the system, but that doesn't change the fact that Jebb wins and rules the roost. How do you propose to invalidate all those 100/0 votes? This isn't a theoretical question, I'm really asking what criteria are you going to use to keep that from happening?

And what prevents the same group from purposefully putting Rob McKenzie dead last in their condorcet vote, to ensure he loses as many possible head to head comparisons? Every system is gameable.

But this raises another question.... are we voting on a system for HGAA officials to make decisions, or on a voting system for the new HGAA membership to select candidates? I sorta assumed it was the former.

Note: The hypothetical by Jack Axaopoulos doesn't add up. If the true desires of the voters were that Rob would defeat Jebb by 60% to 40%, then it wouldn't matter if that 40% (supporting Jebb) put Rob last on their ballots. The Condorcet results would still show that 60% of the voters had placed Rob ahead of Jebb while only 40% had placed Jebb ahead of Rob. The pairwise comparisons are evaluated without regard to any other candidates on the ballot. This is a further demonstration of either (1) Jack's inability to understand what he was talking about or (2) Jack's outright dishonesty. Either trait was a disaster for the HGAA and that's why it vanished shortly after Jack's takeover.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 7:41 pm

No, your not fired for voting what you wanted.
Youre fired because you openly admitted to strategic voting.

Here are two votes from two different voters:

Voter 1 votes: A[95] B[10] C[8] D[6]
Voter 2 votes: A[95] B[10] C[8] D[6]

One of these voters is being completely honest, and the other is intentionally trying to game the system so A will win. Please tell me which is which.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 7:46 pm

Here are two votes from two different voters:

Voter 1 votes: A B C D
Voter 2 votes: A B C D

One of these voters is being completely honest, and the other is intentionally trying to game the system so A will win. Please tell me which is which.

Oh, and you have even LESS data to detect the cheater with. Score one for Range. (in a real vote, the range numbers would almost never match like in your example, the data would be more realistic)

So here we see, condorcet has the same issue, BUT WORSE.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 7:57 pm

Totally inaccurate. No one yelled louder.

They yelled louder by giving the larger number. The 5 voters who wanted A over D will realize this in the next election and are more likely to vote 100/0/0/0 because they (the majority) got stuck with their LAST choice. Why would we want a voting system that would give the majority their LAST choice anyway?

:
:

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 8:07 pm

Here are two votes from two different voters:

Voter 1 votes: A B C D
Voter 2 votes: A B C D

One of these voters is being completely honest, and the other is intentionally trying to game the system so A will win. Please tell me which is which.

I don't have to because the Condorcet doesn't rely on people "playing nice" with their votes. The way you've described the Range system, it relies on some outside mechanism to somehow penalize people who try to "game" the system. Maybe they get called bad names, or maybe they get ostracized. But that still doesn't solve the problem that they will end up winning the elections because they're willing to post 100/0/0/0.

So once again, please tell me how you're going to decide when a particular voter is "gaming" the system and what you're going to do about it? What is the threshold you're going to use? Will you throw out votes that exceed that threshold? Really, I'm asking honestly.

Oh, and you have even LESS data to detect the cheater with. Score one for Range. (in a real vote, the range numbers would almost never match like in your example, the data would be more realistic)

The system I've proposed would use the same scoring format:

A[90] B[80] C[55] D[12]

So the same information is available. But I'm just proposing that if one candidate would defeat all the others according to that ranking, then that candidate would win any runoff and should be declared the winner.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 8:14 pm

Here are two votes from two different voters:

Voter 1 votes: A B C D
Voter 2 votes: A B C D

One of these voters is being completely honest, and the other is intentionally trying to game the system so A will win. Please tell me which is which.

I don't have to because the Condorcet doesn't rely on people "playing nice" with their votes.

Yes you do. You have NO idea if someone artificially changed their order to punish a particular candidate.

The way you've described the Range system, it relies on some outside mechanism to somehow penalize people who try to "game" the system. Maybe they get called bad names, or maybe they get ostracized. But that still doesn't solve the problem that they will end up winning the elections because they're willing to post 100/0/0/0.

We have SOLID evidence, that the system works with olympic judges.
Can you show me a single example where a Russian skating judge gave the Russian skater a 10.0 and all other skaters a 0.0???
It cant happen because of transparency.

So once again, please tell me how you're going to decide when a particular voter is "gaming" the system and what you're going to do about it? What is the threshold you're going to use? Will you throw out votes that exceed that threshold? Really, I'm asking honestly.


How do you decide when someone is gaming condorcet? Same problem.

Oh, and you have even LESS data to detect the cheater with. Score one for Range. (in a real vote, the range numbers would almost never match like in your example, the data would be more realistic)

The system I've proposed would use the same scoring format:

A[90] B[80] C[55] D[12]

So the same information is available. But I'm just proposing that if one candidate would defeat all the others according to that ranking, then that candidate would win any runoff and should be declared the winner.

Other than being extremely complicated, how does it the solve the issue I recently posted where condorcet picks A as the winner, when A is a HORRIBLE choice of winner??? It doesnt. Range chose the correct winner, and condorcet lacks the data to make the correct choice.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 8:19 pm

Wait - here's one ...

[a]=95, [b]=90, [c]=35, [d]=30, [e]=10, [f]=10, [g]=5
[a]=0, [b]=0, [c]=100, [d]=0, [e]=0, [f]=0, [g]=0

One Voter publically voiced a VERY STRONG AND REPEATED preference for "B" over all other non "C" alternative. The other voter publically voiced very little preference between the "A" and "B" with all others distant.

One of these voters is being completely honest, and the other is intentionally trying to game the system. Please tell me which is which.

[smile] [smile] [smile]

JB

I don't really understand your example here. I can't tell if you've got two voters (one per row) scoring 7 alternatives (a through g) or if you've got 7 voters (a through g) scoring two alternatives (one per row).

But it doesn't matter because the Condorcet system does not have to levy any judgement on the voters intentions. You guys are basically telling me that the Range system is vulnerable to people posting extreme (or dishonest) values, but you haven't told me how we're going to keep that from happening. You can't even give me a criteria for detecting it. Is 100/0/0/0 a "bad" vote? How about 90/10/10/10? What about 80/15/12/8? There's a huge judgement call that needs to be made regarding voter's intentions and I don't know how you're going to do that without being mind readers. Can anyone tell me what grounds you're going to use to declare someone dishonest when voting in the Range system?

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 8:20 pm

Can anyone tell me what grounds you're going to use to declare someone dishonest when voting in the Condorcet system?

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 8:28 pm

Simulation of Condorcet Gaming
---------------------------------------------------------

I was curious what would happen if 10% of voters gamed condorcet voting, by simply moving choice #2 to the end of the list out of 5 possible candidates. 1000 simulations were run.

Results: 13+% of election results were effectively GAMED/CHANGED by simply moving one of the candidates to the end of their voting list. (total # of elections gamed is listed at the very bottom... divide by 1000)

You can run the simulation here: http://www.hangglidingassociationofamer ... dorcet.php
The numbers on the left side are how many head to head comparisons each candidate won. On the right, the same count but with 10% gamers.

CONDORCET is easily gameable and with even LESS data to work with than Range, its virtually impossible to detect. It allows a voter all the wiggle room in the world to explain their voting order.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 8:32 pm

Can anyone tell me what grounds you're going to use to declare someone dishonest when voting in the Condorcet system?

There's no need because Condorcet doesn't rely on people being honest or dishonest. Everyone votes for what they want, and the winner is what most people want.

Maybe we should really examine your example here:

Voter #1 A3 B2 C1 D0 (prefers A)
Voter #2 A3 B2 C1 D0 (prefers A)
Voter #3 A3 B2 C1 D0 (prefers A)
Voter #4 A3 B2 C1 D0 (prefers A)
Voter #5 A3 B2 C1 D0 (prefers A)

Voter #6 D80 A3 B2 C1 (prefers D)
Voter #7 D80 A3 B2 C1 (prefers D)
Voter #8 D80 A3 B2 C1 (prefers D)
Voter #8 D80 A3 B2 C1 (prefers D)

The first 5 voters rated D as last. So after this election is over, they're going to be kicking themselves because the majority got their last choice (D). What do you think they'll do in the next election? As people get used to this system, they're going to start voting 100 for what they like and 0 for what they don't like. And then we'll have to have the thought police telling them that they're gaming the system because it's not possible for them to vote 100/0/0/0. But if they're Jebb's clipboard army, they'll just laugh as they take over the HGAA.

The bottom line here is that the Range system rewards bad behavior because there's no way to draw a line. So anyone who's willing to push the extremes will be able to do so unless we can define some "standard of voting behavior" to throw out their votes. Can anyone define that?

Note: The previous ballot example given by Jack can be reformatted to work in the Javascript versions of the Voting program at these locations:

https://ushawks.org/voting/
https://bobkuczewski.github.io/Voting/Voting.html

This is the same data in the appropriate format for those Javascript programs:

A[3] B[2] C[1] D[0]
A[3] B[2] C[1] D[0]
A[3] B[2] C[1] D[0]
A[3] B[2] C[1] D[0]
A[3] B[2] C[1] D[0]
D[80] A[3] B[2] C[1]
D[80] A[3] B[2] C[1]
D[80] A[3] B[2] C[1]
D[80] A[3] B[2] C[1]

The results show that 3 of the 4 algorithms (Simple Plurality, Instant Runoff, and Condorcet) all choose candidate "A" as the winner. Only Range voting chooses candidate "D". When candidates "A" and "D" are compared via condorcet, it is clear that 5 of the voters prefer "A" over "D" while only 4 voters prefer "D" over "A".

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 8:55 pm

And what would happen in Range if 10% gave their candidate of choice a score of 100 and gave candidate A a big fat zero? I know, that will be obvious and you're going to say "naughty naughty". What are you going to do after that? Are you going to disqualify ballots or not? And if so what criteria are you going to use? I've asked this many times with no answer.

Same question applies to Condorcet. But condorcet has LESS data to detect gamers with.

With Range, the membership can look at the actual numbers and decide for themselves, and let it influence their future voting.
I know what I would do if I saw someone voting 100,0,0,0

But when someone votes with condorcet, I really cant conclude anything, im running blind.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 9:08 pm

Gaming = strategic voting = voting in a way designed to harm a specific candidate, or voting in a way that doesnt really reflect your true views because you have an agenda.

The easiest way to game condorcet is to simply put as many candidates ahead of the candidate you want to "ding". This causes them to lose more head to head competitions.

Maybe you don't understand Condorcet. By putting a candidate last, you only end up casting your one vote for each of the other candidates when those other candidates are running against your least favored candidate. If you want that candidate to lose, then that's exactly what you should do. I don't see that as gaming the system at all.

The notion that people have an "agenda" (or not) is false from the start. We all have an agenda. We want to fly. We like logo A over logo B. We want one candidate over another. We like loose ratings or we like strict ratings. These are all "agendas", and I don't think it's bad for people to vote for what they want. That's the foundation of a democracy. It seems that you want to somehow remove people's preferences from voting, and I don't think that's possible - or even correct.

But let me ask this a different way. Let's say there's a group inside the HGAA that just wants to kill the HGAA (unfortunately, this is a very likely scenario). They don't care if people like them or not, because they're just here to destroy this organization. They're willing to vote 100/0/0/0 on everything to get what they want. What are we going to do about it?

I see there are only 3 choices:

1) We let them take over the organization because they're willinng to scream (100/0/0/0 is screaming)
2) We draw some line to throw out ballots that exceed some threshold of "niceness" (they'll end up right on that line)
3) We end up fighting fire with fire (everyone votes 100 for what they want and 0 for what they don't, i.e. plurality)

Do you see any other options?

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 9:15 pm

Same question applies to Condorcet.

No it doesn't. The reason this comes up in Range is that people can exaggerate their positions by using extreme values (100/0/0/0). You can't do that in Condorcet. In Range if you vote 60/40, and I vote 100/0, then I've excercised 5 times more influence on the election than you have. It's as if my vote counted 5 times as much. Now the Range proponents like to say that's good because maybe the difference isn't as important to you as it is to me. But that's a double edged sword because it means that I'm outvoting you by 5 to 1 because I'm willing to post extreme values while you're trying to be more "fair and balanced". The problem is that my extreme behavior would be rewarded and your "polite" behavior would be punished. That establishes exactly the wrong reward/punishment system.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 9:17 pm

Maybe you don't understand Condorcet. By putting a candidate last, you only end up casting your one vote for each of the other candidates when those other candidates are running against your least favored candidate. If you want that candidate to lose, then that's exactly what you should do. I don't see that as gaming the system at all.

I understand it perfectly, ive coded it, and what you just described *IS* strategic voting. You are being a gamer. That is one of the "Bad" things all voting systems try to address.

The rest is a semantics argument I have no interest in.

If you are voting in a way to support one candidate while doing maximum damage to some particular candidate, youre rigging the system.
Range and Condorcet are gameable. I dont need it to be proven for Range, I already KNOW its gameable.
I only wrote the simulation to show that Condorcet is easily gameable too.

Youve been running an anti-Range campaign while ignoring all the issues with Condorcet, so I have shined the light on the problems with condorcet.

The glaring example I gave where Condorcet picks a winner A, that everyone thinks totally sucks, shows just one of its big flaws.
All voting systems have issues. Strengths and weaknesses.

But for HGAA officials, Range Voting (With an abstain), is the best system ive seen so far because its the most expressive, provides the most data, and gives us insight into what people REALLY think about candidates, or ideas, or whatever we are voting on.

Anyway... is Scott MIA? Seems all the active people have asked for a vote several times now.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 9:24 pm

OK, Let me propose a solution just to get us going.

Let's calculate both Range and Condorcet for every vote. When the votes are the same, we're good. When they differ we can have a simple majority vote to decide whether people like the Range result or the Condorcet result.

Can we do that? If so, we can start voting right now and get the HGAA moving again.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 9:25 pm

Were stuck on the chair. Ill accept almost anything at this point to get things moving.

JB's idea of a strawpoll to do a quick check on consensus would be quick and easy.
There may be NO debate at hand and its an easy win.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 9:28 pm

OK, Let me propose a solution again ... just to get us going.

Let's calculate both Range and Condorcet for every vote. When the results are the same, we're good. When they differ we can have a simple majority vote to decide whether people like the Range result or the Condorcet result. This will also give us some hands-on experience with the two systems and might help us to choose one or the other for the long run.

Can we do that? If so, we can start voting right now and get the HGAA moving again.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 9:31 pm

Still stuck waiting on the chair, regardless.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 9:36 pm

Still stuck waiting on the chair, regardless.

That's true, but can we agree to proceed with that plan so that we're all ready to go when Scott comes back?

Are you good with that SG? How about you, JB?

Hopefully they'll agree most of the time and we can accomplish something. And when there is no Condorcet winner, then Range will win automatically. Is there anyone who has a problem with that as a way to get us moving forward?

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 9:41 pm

I doubt anyone will have an issue with tracking alternative methods in parallel if someone wants to do the work.

But the decision to use both would just be another voting option for the straw poll or whatever.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 9:47 pm

I'm not sure I understood your reply, but let me be more formal.

Motion:

I move that we begin voting using the generally accepted shared notation listing candidates and scores between 0 and 100 (inclusive):

A[100] B[73] C[13] D[0]

Then we apply both the Strict Condorcet method and the Range method. If there is no Strict Condorcet winner, then we declare the Range winner as the final choice. If there is a Condorcet winner and it agrees with the Range winner, then we also declare the shared winner as the final choice. If the two methods produce different winners, then we hold a second vote by simple majority to decide which method's winner to use.

Is there anyone willing to second that motion? Thanks.

Re: Interim Voting Committee - Session #1
Postby sg » Tue Jul 13, 2010 10:07 pm

Ummmm..... why would we debate this to death with thoughts of finally taking a vote, and then at the end simply accepting what one person wanted all along???

The vote is going to determine what voting system we officially go with. Just need Scott to start the voting.
We need an official list of choices that are NOT ambiguous. Like....

a) Pure Range Voting
b) Range voting with abstaining allowed (details posted later...)
c) condorcet as implemented in Bobs code listed HERE (provide link)
d) Scares method 1
e) Scares method 2
etc
etc

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 10:40 pm

Ummmm..... why would we debate this to death with thoughts of finally taking a vote, and then at the end simply accepting what one person wanted all along???

Actually, it's not what I wanted all along. I would first prefer a pure Condorcet method (with Condorcet completion). But even without that, it's not what I proposed all along because my current motion requires a second vote when the two disagree.

Now that I think about it, this second vote is actually a very good thing because it gives us a warning when something is controversial enough to be judged differently by two different methods. It gives us all a chance to take a second look at what it is that we're doing when we're in the gray area. People may vote differently the second time around. It will also shine a light on those cases where the two methods disagree - because that's when we'll be taking that second vote. That will force us to see how the two methods differ in real life. So given that we're trying something new, I think having a second vote is a good safety net for us.

The vote is going to determine what voting system we officially go with. Just need Scott to start the voting.
We need an official list of choices that are NOT ambiguous. Like....

a) Pure Range Voting
b) Range voting with abstaining allowed (details posted later...)
c) condorcet as implemented in Bobs code listed HERE (provide link)
d) Scares method 1
e) Scares method 2
etc
etc

I think a laundry list is a bad idea because the only way we can currently vote (at least from my understanding) is by plurality (I think that's how we chose the HGAA name with only 41% of the vote). So that means that we'll end up fracturing the vote with some small minority winning. I think we should just list those options that are nominated by someone willing to provide a full and unambiguous description of what they do (including working code if possible).

Having said all that, I again assert my previous motion, and I welcome anyone willing to second it.

Re: Interim Voting Committee - Session #1
Postby bobk » Tue Jul 13, 2010 11:57 pm

While we're waiting for adult supervision (Scott, where are you?), I'd like to share a couple of flying stories related to "gaming" the system.

The first is a story told by Chris Wills on the movie "Big Blue Sky". He said that they studied the rules and they "knew the rules", and that helped them take the top positions in an early meet (maybe the first "Nationals"?). For anyone who hasn't seen "Big Blue Sky", I strongly endorse it. It tells the story of hang gliding in the early days and it's just fascinating. I bought 4 copies from USHPA and I must have watched it a dozen times. I've got one of the copies circulating through the Hawks membership list (although I think it's stalled right now). Check it out whenever you can.

The second story was told to me by Joe Greblo. I can't remember the details, but I think he (or someone else?) analyzed the rules of an early competition and figured out that the point system favored some particular manuver (maybe the number of 360 degree turns during the flight). So the pilot (maybe Joe?) used that information to just repeat that manuver over and over until they just about hit the ground. Because of the weighting that won the meet (come to think of it, this could be the same story told from a different perspective!!).

The point here is that people want to win. Whether it's picking their favorite logo or excluding PGs or including PGs or picking their buddy for leadership, people will look at the rules and analyze them for their best advantage. Some people will call that smart, others will call it dishonest. But regardless of what you call it, I guarantee that people will do it.

So the best thing that we can do is to pick a system where there are no obvious advantages to "misbehaving". The problem with the Range system is that anyone who chooses extreme values (100,99,98 or 2,1,0) will exert more voting power than someone who chooses more moderate values (70 through 30). People aren't stupid, and they're going to vote in a way that gives them what they think is the best result. It may be best for them personally (selfish best), or it may be what they think is best for our sport (altruistic best), but whatever their motivation, they're going to naturally vote in a way that produces what they think is best. We need a system that recognizes this aspect of human nature and handles it naturally without giving undue influence to those who will push its limits. I think the Range system fails in that regard, and I think the Condorcet system does not. If we set up a system that rewards people for voting 100,0,0,0 then people will figure that out, and that's how they'll vote on matters that matter. That's how I'll vote when I believe that it's the right thing to do. You can call me names if you like, and you can threaten to kick me out of the HGAA if you like, but if there's ever a Range vote choice between Rob McKenzie, Dave Jebb, Gabe Jebb, and Robin Marion, you can bet I'm voting 100,0,0,0 unless there's an explicit rule prohibiting me from doing so. I'll encourage everyone else to join me because you can bet the other side will be doing the same in reverse. Welcome to humanity.

Scott? Scott? Can we vote yet?

Jack's influence over the Transition Team was strong and a number of members felt that the HGAA would fail without Jack's technical efforts (Jack created and controlled the web site). So the Range voting system was selected. As noted earlier, the Range voting of the HGAA was used by a minority of the leaders (voting at the extremes) against the majority (fearful of Jack's threats for voting at the extremes). The HGAA collapsed shortly afterward.

However, a number of noteworthy organizations now use Condorcet methods. These include many of the high tech free software organizations including Debian, Wikimedia, Software in the Public Interest, Gentoo Foundation, as well as a number of political organizations (such as the Libertarian Party of Washington).