06-07-2008, 01:09 PM
|
#141 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi! Quote:
Originally Posted by Capt. Slo-mo  Their choice of candidates by personal preference is Ro, Jerry and Bruno, in that order. But they want to keep the NomCom cabal from a clean sweep...so they're telling everyone to give Bruno the #1 slot, as a way to help block Mark's "pseudo-slate" package voting advantage. | Well, that depends on how popular Bruno is among those who vote for a total Hurley slate on the VP slots. If he is unpopular among them, and the NC slate in general holds a strong lead, it would in most cases be best to vote for the most popular Hurley candidate, if one wants to avoid a clean sweep for the NC group. Quote:
Originally Posted by Capt. Slo-mo If they get 5,000 USFA voters to swing their way, wouldn't that game the results? | Well, that depends on how many voters there are! If USFA has 20,000 members, yes, then that would be a significant voter group. Quote:
Originally Posted by Capt. Slo-mo And wouldn't it be beneficial to know if there was a substantial 1 vs 3 penalty factor? | Of course. If the USFA were to use the constrained approval voting system (X X X votes), then there would not be any 1 vs. 3 difference, but the 3 vs. 4 difference would be huge. If the USFA were to used conventional Borda count, then the difference would be equally spread over the voter preference spectrum. If the USFA were to use veto voting (only well-known use in Survivor reality show), then there would be a huge 5 vs. 6 difference. If the USFA were to use its complete counterpart, FPTP voting (used in almost all US. political elections), then there would be a huge 1 vs. 2 difference.
All of those systems can be fairly easily gamed by a substantial voter contingent which has some knowledge about how the rest of the voters will vote. The gameability, or lack of resistance to insincere voting, varies among them, with veto and FPTP voting being the two worst-performing systems.
However, the resistance can be drastically improved by a fairly simple measure, called hybridization of voting systems. IRV (as recommended by RRoO) can be regarded as a self-hybridization of FPTP, and the former has much less drawbacks than the latter. Coombs voting is a self-hybridization of veto vote, and the Nanson and Baldwin systems are different self-hybridizations of Borda count voting. Nanson and Baldwin have much better performance than Borda, while not being much more computationally intensive.
More in following posts.
Have a nice time!
Peter Gustafsson
(Unfunny extraneous stuff typical of Slo-mo removed for clarity. Whew. I can not imagine that there in the whole world exists any other person with such a unrequited love for comedy.    ) | 
|  | | And now for this message... |
| |
06-07-2008, 02:11 PM
|
#142 | | Moderator
Join Date: Feb 2005 Location: Austin, TX
Posts: 10,151
| Just to ask, is there software where you can type in the results of a ballot and then the software will calculate the results of the election based on your selected system? Because if we don't use that, and have say 5000 votes, and it matters what who placed a person #2 vs #1, that's a LOT of paper shuffling. | |
| | |
06-07-2008, 02:38 PM
|
#143 | | Senior Member
Join Date: Oct 2002
Posts: 491
| Quote:
Originally Posted by KD5MDK Just to ask, is there software where you can type in the results of a ballot and then the software will calculate the results of the election based on your selected system? | No, there's no software.
Why none of the associations, organizations, companies, localities or governments using voting systems have ever thought of using software; the idea has never come up before. | |
| | |
06-07-2008, 04:19 PM
|
#144 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi!
[quote=telkanuru]
Who cares? Regardless of the technicalities of the system, you would still put (1) down for your first choice (2) for your second and (3) for your third. They provided you will all the information needed to fill out the ballot correctly. Since you don't actually know everyone else's votes, there's no way to vote where you could plan certain outcomes, regardless of process.
To which Dberke responded: Quote:
Originally Posted by dberke I think this is a valid point, and potentially gives a more "pure" vote. By not disclosing the counting method, it makes it harder for people to "game the system" by voting in an order other than their true preference. Is it possible that the election committee has decided exactly what counting procedure they plan on using, but they intentionally have not disclosed it for that reason?
Dan | Bold emphasis mine.
If that is the reason for the present non-disclosure, then that would be (To the best of my knowledge) a first in the history of voting systems. And, there is a much better way to achieve that end.
The problem of how to discourage insincere voting - people voting in an order which is not their true preference - has been extensively studied. However, the so far only countermeasure which has been studied in detail is to design voting systems which are more resistant to strategic voting. Such systems exist, and it is not very hard to show that some of them are considerably more resistant than other.
Furthermore, it is possible to prove that if one combines more than one voting system - hybridization of voting algorithms - then it is possible to achieve a voting system which is extremely resistant to strategic voting, even if the simple voting systems which are the components of the hybrid in themselves are easily manipulated.
Such a hybrid voting system works by:
1. Use one simple voting system (FPTP, Borda, Veto, Binary Cup, etc.) to remove one candidate from the running.
2. Take all votes, and strike that candidate from the preferential lists. Those candidates which were ahead of him in a particular vote keep their placement on that particular vote, those that were behind him move up one step.
3. Use all the abridged votes from step #2, and run them through a simple voting system to remove one candidate from the running. This voting system may be the same as in step #1 (in which case we have a self-hybridized voting system) or another one. Not all combinations of simple voting systems will give good results, but a surprisingly large proportion of the possible combinations do.
4. Repeat steps #2-3 until only one candidate is left in the field, that candidate is then the winner of the election.
It can be proved that even fairly simple combinations of simple voting systems will give hybridizations which are exceedingly hard to manipulate. In many cases where there is a close election it is so hard so that even if a presumptive manipulator has access to good information on how the other voters will cast their votes and a strong computer, he can not find the best insincere vote ballot in the time available from announcement of candidates to actual voting date. Also, if he has access to good but not perfect information on how the other voters will cast their votes and a strong computer, it is quite likely that the computer will spit out a strategic vote ballot which will hurt his interests.
Those characteristics of the hybrid voting systems make them very good countermeasures to strategic voting.
However, if a voting committee would act as Dberke hypothesizes, then it would open itself up to accusations that they are searching for a voting system which gives the best results according to their preferences. Also, the committee would have no way to disprove that accusation.
Furthermore, such a behavior would suggest a certain canniness with regard to voting theory, while at the same time showing ignorance of the theory of hybrid voting systems. Personally, I can not square the former with my perception of the USFA.
Thus, I assume that the present non-disclosure has other reasons, probably major ignorance with voting theory.
Have a nice time!
Peter Gustafsson | |
| | |
06-07-2008, 04:24 PM
|
#145 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi! Quote:
Originally Posted by KD5MDK Just to ask, is there software where you can type in the results of a ballot and then the software will calculate the results of the election based on your selected system? Because if we don't use that, and have say 5000 votes, and it matters what who placed a person #2 vs #1, that's a LOT of paper shuffling. | Here you are: http://userfs.cec.wustl.edu/~rhl1/rbvote/calc.html A word of warning: the site which this comes from is rabidly pro-approval voting, and just as rabidly anti-IRV. If you think that ROW discussions can be heated - you have NOT seen discussions among voting system theorists discussing their preferred systems. They are even more hot-heated that the pro-choice/pro-life discussions!
Have a nice time!
Peter Gustafsson | |
| | |
06-07-2008, 04:28 PM
|
#146 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi! Quote:
Originally Posted by dberke I don't think so - I would expect that votes for the first choice candidate weigh more than votes for the second and third choice. And the second weighs more than than the third choice.
Are there any preferential voting systems where that isn't the case? Unless there are, I don't see how which counting method is being used would cause someone to change their vote. | Simple answer to question in bold:
Yes, there is. One-round veto voting
See, that was not so hard!
Have a nice time!
Peter Gustafsson | |
| | |
06-07-2008, 04:35 PM
|
#147 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Silver bullet
Hi!
There is a simple, sane, fair solution to this conundrum. It only requires a relatively minor change in the bylaws, and would make elections - and their tabulation - much easier next time around.
In order to further discussion and the general education on this topic, I will not reveal it before anyone else has outlined it. Spoon-fed students will not learn much, and forget quickly what they might have grasped.
Have a nice time!
Peter Gustafsson | |
| | |
06-07-2008, 04:38 PM
|
#148 | | Senior Member
Join Date: Oct 2003 Location: North Carolina
Posts: 1,322
| The problem with IRV in the case of the USFA VP election (which is the only race where this is an issue) is that you are actually trying to elect 3 candidates not one. IRV is meant to achieve a majority vote for one candidate.
In the USFA VP case its entirely likely (in fact I'd say probably) that you have no preference between your top 3, but there's a big drop-off between 3 and 4. Using IRV you really one get one vote, you just get one vote in each subsequent run-off of the election. That works best if you're trying to elect one person not three.
For a three person election, most votes counted works relatively well, or you could do it proportional voting, where you vote for a slate. | 
| | |
06-07-2008, 04:51 PM
|
#149 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi! Quote:
Originally Posted by Philistine AIUI, there are cases where the weights are different--under different types of preferential voting for instance, the same ballots would produce different results. | "Weights" is usually used in connection with the voting system called Borda count. In the countries which use it (Slovenia - partly, Kiribati, Nauru) all have differing weights allotment systems, which obviously give different results. Quote:
Originally Posted by Philistine It's not so much "gaming the system" as being able to determine the optimum voting strategy. It may be possible, for instance, depending on the method used and the # of supporters out there, for one "slate's" supporters to insure that one member of their slate is voted in if a certain voting strategy is pursued by its supporters, even if there is no hope of all 3 VP candidate's being voted in. | Yes, that is the case if the voting system is not really bad, and the sides are reasonably evenly matched. Quote:
Originally Posted by Philistine Personally, my vote (  ) would be for a modified form of the illustration in Robert's Rules of Order: If, by counting each 1, 2 and 3 vote as equally valid, those who receive a majority of votes are elected. If there aren't 3 elected by majority, you go to the "instant runoff" election style described in RRO. That way, XXX votes get counted (at least in the first instance). | I am not sure that I understand you, but if I do then you are suggesting something similar to Bucklin voting. That voting system has the distinction that it is the only one that I know of that actually has been found unconstitutional in an US. court. That can not be a good thing.
See: Brown v. Smallwood, 130 Minn. 492, 153 N. W. 953
Dove v. Oglesby, 114 Okla. 144, 244 P. 798
Have a nice time!
Peter Gustafsson | |
| | |
06-07-2008, 05:03 PM
|
#150 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi! Quote:
Originally Posted by seak The problem with IRV in the case of the USFA VP election (which is the only race where this is an issue) is that you are actually trying to elect 3 candidates not one. IRV is meant to achieve a majority vote for one candidate. | Actually, there is an extended version of IRV which is intended to elect a multi-seat chamber - Single Transferable Vote. See: http://en.wikipedia.org/wiki/Single_transferable_vote Quote:
Originally Posted by seak In the USFA VP case its entirely likely (in fact I'd say probably) that you have no preference between your top 3, but there's a big drop-off between 3 and 4. Using IRV you really one get one vote, you just get one vote in each subsequent run-off of the election. That works best if you're trying to elect one person not three.
For a three person election, most votes counted works relatively well, or you could do it proportional voting, where you vote for a slate. | Can you please rephrase the sentence marked in bold? I can not get it to match with the rest of your statements.
Have a nice time!
Peter Gustafsson | |
| | |
06-07-2008, 08:39 PM
|
#151 | | Senior Member
Join Date: Oct 2003 Location: North Carolina
Posts: 1,322
| Essentially (and this is just going from RRO since that is what is prescribed in the USFA bylaws and why I didn't discuss the voting system Peter mentions) everyone has a first choice marked 1 votes are counted such that your vote is counted for the candidate you marked 1, your vote is not counted for the people you marked 2 and 3, so one vote.
Person with fewest votes is cut, and if that is your person 1 then they look at who you voted for number 2 and your vote is counted for that 1 person. So your vote gets counted through every round, but it only gets counted for 1 person, even though their are three spots.
I think for choosing three people for three spots, this is a horrible system, and would be happy to learn I missed something in Roberts (I didn't read it that carefully), or the election committee isn't following it.
Last edited by seak; 06-07-2008 at 08:41 PM.
| |
| | |
06-07-2008, 09:31 PM
|
#152 | | Senior Member
Join Date: Oct 2002 Location: West Coast
Posts: 2,402
| Quote:
Originally Posted by PeterGustafsson In order to further discussion and the general education on this topic, I will not reveal it before anyone else has outlined it. Spoon-fed students will not learn much, and forget quickly what they might have grasped. | ::waves hand wildly:: Oh, me! Me!
I want to be spoon-fed the mystical trappings of the Masonic Universal System of Electoral Results! (MUSER). It sounds so much more betterer than that outdated and dysfunctional Australian Borda Universal System of Electoral Results (ABUSER), or the long-discredited D'Hondt Ogliarchy System of Electoral Results (D'HOSER).
Please tell me the initiation process doesn't involve a yak, inclined plane computations using Swedish meatballs and waterboarding in a barrel of lutefisk oil. I had enough of that trying to master the rites of the Ljungby Occupational System of Ergonomic Rastafarianism (LOSER).
__________________
"Fraud is the creation of trust. And then: its betrayal."
William Black, Ph.D.
| |
| | |
06-08-2008, 12:10 PM
|
#154 | | Senior Member
Join Date: Nov 2002 Location: Sweden
Posts: 3,046
| Hi!
Look at my first example: Quote:
Originally Posted by PeterGustafsson Let us assume this vote tally:
49%: A1, A2, A3
17%: B1, B2, B3
17%: B3, B1, B2
17%: B2, B3, B1
<Explanations and calculations omitted>
A1 is elected first, followed by B1 and B2. | Let us assume that the voters really have preferences below place #3, and that they look like this:
49%: A1, A2, A3, B1, B2, B3
17%: B1, B2, B3, A3, A2, A1
17%: B3, B1, B2, A3, A2, A1
17%: B2, B3, B1, A3, A2, A1
The preferences noted on the ballot are in bold. In this example, some things should be noted:
1. Candidate A1 is a highly divisive person. He is either best or worst according to the voters, but never anything in between.
2. In pairwise comparisons, A1 loses (49-51) against every other candidate. Yet, he is elected first, and the only one from his side! This example shows that the election system, as previously described in this thread, fails the Condorcet loser criterion.
3. A1 is elected, despite being considered the worst possible candidate by an absolute majority of the voters. Thus this system also fails the anti-plurality criterion.
Have a nice time!
Peter Gustafsson | |
| | |
06-08-2008, 05:29 PM
|
#155 | | Curmudgeon-in-Chief
Join Date: Jul 2001 Location: Somewhere in your nightmares!
Posts: 23,475
| Quote:
Originally Posted by PeterGustafsson Hi!
Look at my first example:
| No.
NO.
I am averting my eyes, in fact!
Aughhhh!
__________________
Use the Shift key, people! Keyboard manufacturers everywhere are ineffably saddened when you ignore what they made just for you!
| |
| | | Thread Tools | | | | Display Modes |  Linear Mode | 
Posting Rules
| You may not post new threads You may not post replies You may not post attachments You may not edit your posts
HTML code is Off | | | All times are GMT -4. The time now is 07:23 PM. |
