Free State Project Forum

Please login or register.

Login with username, password and session length
Pages: [1] 2   Go Down

Author Topic: Problems with all voting  (Read 12874 times)

Eddie_Bradford

  • FSP Participant
  • *****
  • Offline Offline
  • Posts: 567
Problems with all voting
« on: July 29, 2002, 12:15:27 pm »

This is an analysis of how almost all systems of voting can lead to a non optimal choice.
1. Highest # of votes wins
->Then you could win with 34% of the vote and 66% of people COULD hate that choice more than the other two.

2. Instant run-off
100 people voting
32 - NH   ----->    51 - DE  -----> DE!
33 - DE   ----->    49 - WY
35 - WY

In this case 35 people REALLY like WY, HATE DE and they like NH almost as much as WY.
33 people REALLY like DE and they like NH almost as much but HATE WY.
32 REALLY like NH of these 14 HATE DE and 18 disslike it but not as much as WY which they hate.
-> WE have DE chosen with
33- love it                           NH   32-love it
49- hate it                                  68- really like it
18- disslike it

3. Aproval voting (ie You just make a mark whether you approve or dissaprove of a particular candidate highest # of approvals win)
AKA Nader wins scenario
Like instant run off, it lets you vote 2 or more times so you don't have to 'sacrifice' your vote for the 3rd party.
So the liberals who also like Nader would vote for Nader and Gore and team up with the "anyone but Gore" faction to sweet Nader into office.  This system isn't as bad as some of them but it tends to award the victory to people who are neutral and who nobody really likes alot.

3. Point system, this is where you distribute points like in Jason's poll say 10 points.
Problem here is that let's say most people wanted NH and 60/100 people put 6 points into it and distributed 4 points to other random states but not Delaware.  NH now has 360 points.  The 40 people remaining realize that NH has popular support and they think it's probably going to win.  They don't hate NH they just happen to like 2 other state more than NH.  These guys all like different states as their #1 (like hawaii) but they know there really isn't alot of support out there for their particular state.  But they all favor Delaware as their #2 and they've heard a lot of talk about it and they think it's really the only other one they like to have a shot against NH.  So they vote all of their 10 points for DE and so DE wins with 400 points.  So the main problem with this is that your voting power is diminished by supporting more than one state, so there's really no incentive to put any points towards say hawaii or some other state that doesn't have much of a chance of winning because it takes points from a choice that has a better chance of winning.

4. Exhaustive combination voting
This is where you would answer a question like this
"Would you prefer DE or NH?"
for all comdinations of choices.
This too doesn't always work but the scenario I worked out was kind of complicated and now I forget exactly what it was.

5. Approval voting revisited
In this situation you would have 5 choices for each candidate from strongly dissaprove to strongly approve.  This way you could distinguish between choices you really like and choices you kind of like.  So "not offending anyone" will no longer sweep you into office.  Still people could abuse this by 'strongly dissaproving' of a front runner that they're actually neutral on so that their vote is more powerful on their other favority candidate.
Logged

JasonPSorens

  • Administrator
  • *****
  • Offline Offline
  • Posts: 5720
  • Neohantonum liberissimum erit.
    • My Homepage
Re:Problems with all voting
« Reply #1 on: July 29, 2002, 04:56:07 pm »

I'll just deal with #3 since that's the voting system the FSP uses. ;)  You've hit on a real potential problem with Cumulative Count: people can vote strategically if they know which state is the frontrunner.  So to avoid this we will conduct the vote concurrently (everyone has to vote around the same time, the result being revealed only after voting is over) and we'll have to remove the poll on the website a good while before we conduct the membership vote.  The key is to spread *uncertainty* over which state has the most support.  This may be one advantage of limiting the number of states, incidentally.  If we can get down to a choice among 6, all states will probably have their vocal supporters, and no one can be really sure which one is the likeliest choice.  So everyone will then vote "honestly" and we'll have an honest result.
Logged
"Educate your children, educate yourselves, in the love for the freedom of others, for only in this way will your own freedom not be a gratuitous gift from fate. You will be aware of its worth and will have the courage to defend it." --Joaquim Nabuco (1883), Abolitionism

Eddie_Bradford

  • FSP Participant
  • *****
  • Offline Offline
  • Posts: 567
Re:Problems with all voting
« Reply #2 on: July 29, 2002, 06:24:49 pm »

The other question I have is do we go with a state that has a 'majority'?  Meaning how many rounds do we do through?  Personally I think the final choice should have a 'majority' otherwise we go into a run off.  Also I think we should have a vote when there are 3 or 4 choices left so people have a good chance of deciding between the states that are most likely instead of 'wasting' it on one that doesn't have a chance.  Back to the 'majority' if we have 5,000 people so 50,000 vote points total should we consider 25,000 points (half of the people giving 10 points) a majority or 12,500 (half of the people giving at least 5 points)
Do you think the state needs to have a 'majority'? or just the most points?
-Eddie
Logged

Elizabeth

  • Former FSP Vice President
  • *****
  • Offline Offline
  • Posts: 1650
  • Someone has to ask the tough questions...
    • Free State Project
Re:Problems with all voting
« Reply #3 on: July 29, 2002, 08:45:20 pm »

Eddie, no offense, but this is in the FAQ:

20Q. How does the Cumulative Count vote mechanism that you'll be using to decide on the state work?
20A. Cumulative Count gives each voter the same number of votes and allows them to distribute these votes among the available candidates however they choose. Let's say, just for example, that the vote is between Alaska, New Hampshire, Delaware, and Montana, and each voter is given 10 points. You can give 6 points to Alaska, 3 to New Hampshire, 1 to Delaware, and none to Montana if you choose. This method is superior to majority rule because it takes account of strength of preference (distances between candidates) and because it allows compromise candidates (which may be no one's first choice but a good second or third choice for everyone) to gain votes they wouldn't get under majority rule. It is also better than sequential elimination voting, in which one candidate is eliminated each round, because it can be done all at once and with less expense.

Also, here's a good resource:

http://www.fairvote.org/cumulative/index.html
Logged

Eddie_Bradford

  • FSP Participant
  • *****
  • Offline Offline
  • Posts: 567
Re:Problems with all voting
« Reply #4 on: July 29, 2002, 08:55:45 pm »

I know it's mentioned there but that doesn't mean we shouldn't disscuss the advantages and dissadvantages of this system.
The only issue really I have with this method is that if you happen to like more than one state then your first choice is less powerful than other people's first choice who put all their points into one state.  I may be wrong but it seems to me that a 1-5 ranking system could be more fair where you could mark two states you really like with a 5 one state that okay with a 3 and the rest with a 1.  That way liking more than one state doesn' penalize you, you can still give it a '5'.
-Eddie
Logged

Elizabeth

  • Former FSP Vice President
  • *****
  • Offline Offline
  • Posts: 1650
  • Someone has to ask the tough questions...
    • Free State Project
Re:Problems with all voting
« Reply #5 on: July 29, 2002, 09:07:01 pm »

But Eddie, the whole point is to measure preferences.  If you like two states -- let's say a strong preference for A and a minor preference for B over the rest of the states -- and someone else only wants one state (A), and doesn't give the least hoot about any other state, CC measures those relative preferences.  You only get 10 units of preference, so that no one gets to express more desire than anyone else.  If you want state A 7, and state B 3, then you can't want state A as much as 10, which the other votore does by giving state A all his votes.  See?
Logged

Elizabeth

  • Former FSP Vice President
  • *****
  • Offline Offline
  • Posts: 1650
  • Someone has to ask the tough questions...
    • Free State Project
Re:Problems with all voting
« Reply #6 on: July 29, 2002, 09:09:20 pm »


I may be wrong but it seems to me that a 1-5 ranking system could be more fair where you could mark two states you really like with a 5 one state that okay with a 3 and the rest with a 1.  That way liking more than one state doesn' penalize you, you can still give it a '5'.
-Eddie


Then what's to keep everyone from giving every state a 5?  The only way CC works is if there's a limit on points aka ranking aka preference.
Logged

Eddie_Bradford

  • FSP Participant
  • *****
  • Offline Offline
  • Posts: 567
Re:Problems with all voting
« Reply #7 on: July 29, 2002, 09:28:46 pm »

Quote

Then what's to keep everyone from giving every state a 5? 


Nothing but if you give every state a 5 then you basically haven't voted.  If you give every state a 5 except one which you give a 1 then in effect you've given an 'anti-vote' which you should be able to do.  If you really hate one choice now you would have to evenly spread out your points to all the other states instead of giving one state a -10.  This is much less powerfull than voting a 5 for all of them except for 1.  The current system assumes that if you really like two state you don't REALLY like them as much as the guy who only likes one state.  So let's say most (90%) people are pretty easy going and really really like any state except Wyoming so they spread their votes out on all (let's say) ten other choices, and then there is 10% hard core Wyoming people.  Well then if there is 100 people Wyoming gets 100 points and all the other ones get 90 points so Wyoming wins even though 90% of the people really really hate it!  I know it's a minor point and I'm sure we'll come up with the utility maximizing state and all it's just that some of these issues I do think are worth disscussing.    ;) ;) :D :P :D ;) ;)
-Eddie
Logged

Elizabeth

  • Former FSP Vice President
  • *****
  • Offline Offline
  • Posts: 1650
  • Someone has to ask the tough questions...
    • Free State Project
Re:Problems with all voting
« Reply #8 on: July 29, 2002, 09:51:25 pm »



Nothing but if you give every state a 5 then you basically haven't voted.  
-Eddie


Yes you have, and you've been allowed to indicate more "desire ponts" than someone who gives a 5, a 4, a 3, a 2, and a 1.
Logged

Eddie_Bradford

  • FSP Participant
  • *****
  • Offline Offline
  • Posts: 567
Re:Problems with all voting
« Reply #9 on: July 30, 2002, 12:38:27 am »

No that not true because say someone voted, as you said, and gave a 5, 4, 3, 2, and 1.

So the score would be

NH 5
DE 4
WY 3
AK 2
VT 1

Now assume someone voted all 5s, the vote would be
NH 10
DE 9
WY 8
AK 7
VT 6

So in other words nothing has changed the states are all in the same order and no one state improved or been diminished because you haven't indicated a preference of one over another.  So giving all the state a 5 is the same as giving all the states a 4 is the same as giving all states a  3... 2... 1.  Likewise if you give all states a 4 except NH which you give a 5 then that is the same as giving all states a 1 and NH a 2.  Think about it and it will make sense.
-Eddie :) ;) :D ;D >:( :( :o 8) ??? ::) :P :-[ :-X :-\ :-* :'(
« Last Edit: July 30, 2002, 12:40:25 am by Eddie_Bradford »
Logged

Mike R.

  • ****
  • Offline Offline
  • Posts: 19
  • I'm a llama!
Re:Problems with all voting
« Reply #10 on: July 30, 2002, 03:24:57 am »

Let us say we are voting between 5 states by using the Cumulative system, 10 points each to 5000 voters. If 1000 voters were to give all 10 of their points to Montana, they can still lose if the remaining 4000 each give so much as 3 points or some other differential combination (one gives a 5 and another a 1 to average 3) to, say, Delaware, which may merely be a popular second preference to those more interested in New Hampshire or Vermont and giving 7 points to them as a first choice.  You can play with the variables. 1500, 10-point voters for a single state, 3500 other. There is always a threat to the strategy vote.  It would require 2500 voters giving all 10 of their points, and 1 more voter giving at least 1 additional point to a single state to strategically ensure a fool-proof victory, which is a majority count (50%+)regardless of the voting technique used. I dont see that as likely to happen.

The beauty of the cumulative system is that it is possible for a state to be a more popular second choice than a first choice, giving the victory to the state of highest compromise between the voters.  
Logged

amyday

  • Guest
Re:Problems with all voting
« Reply #11 on: July 30, 2002, 10:40:11 am »

I like Eddie's 5-3-1 idea because it is closer to the method I think would work the best. I think you should be able to give your opinion of every choice. If there are 5 choices a,b,c,d,and e, then you should give a preferance number to each choice. Either 1 thru 100, or 1 thru 10, maybe even 1 thru 5. The higher you allow, the more fine tuning you get. So when you vote you say how you feel about each choice. A gets 90(strong approval), b gets 80(slightly strong approval), c gets 50(don't care either way), d gets 30(don't support, but not totaly against), and e gets a 0(total disapproval).  

Say you use the 10 point distribution method. And there are 1000 voters. And there are 3 choices, a,b, and c. 800 people like a and b equally well and like c only a litte so they vote a 4, b 4, and c 2. A group of 200 are big supporters of c, even though they are not against the others, they just feel it would be best to throw all their 10 points at their favorite and put all 10 points to c. This will give totals of a  3200, b 3200, and c 3600.  The majority is evenly split between a and b, and gave a small support to c, but with that small support and the support of a minority, c get chosen.  

With the ability to give your preference for all choices, people wouldn't give all their points to their favorite and hope it got chosen. They would give their preference to all. So the same 1000 and a,b, and c. 800 people like both a and b so give both 80%. They also aren't totally against c and give it a 20. The 200 who support c give it 100 but also slightly like a so give it a 50 and aren't totally against b and give it a 20.  This would give a total of a 74,000, b 68,000, and c 36,000.  This scoring shows a much more accurate portrayal of the voters preference.

Logged

Mike R.

  • ****
  • Offline Offline
  • Posts: 19
  • I'm a llama!
Re:Problems with all voting
« Reply #12 on: July 30, 2002, 01:20:06 pm »


Say you use the 10 point distribution method. And there are 1000 voters. And there are 3 choices, a,b, and c. 800 people like a and b equally well and like c only a litte so they vote a 4, b 4, and c 2.

A group of 200 are big supporters of c, even though they are not against the others, they just feel it would be best to throw all their 10 points at their favorite and put all 10 points to c. This will give totals of a  3200, b 3200, and c 3600.  The majority is evenly split between a and b, and gave a small support to c, but with that small support and the support of a minority, c get chosen.  

With the ability to give your preference for all choices, people wouldn't give all their points to their favorite and hope it got chosen. They would give their preference to all. So the same 1000 and a,b, and c. 800 people like both a and b so give both 80%. They also aren't totally against c and give it a 20. The 200 who support c give it 100 but also slightly like a so give it a 50 and aren't totally against b and give it a 20.  This would give a total of a 74,000, b 68,000, and c 36,000.  This scoring shows a much more accurate portrayal of the voters preference.




What if I opted out of choice C when I joined? Should I have to grant points to it? I can do the same in cumulative voting by granting a 6 to A, 3 to B and 1 to C. I also have the option of granting 7 to A and 3 to B, giving 0 to C because I dont like C as a choice in the least bit.
Logged

Eddie_Bradford

  • FSP Participant
  • *****
  • Offline Offline
  • Posts: 567
Re:Problems with all voting
« Reply #13 on: July 30, 2002, 02:25:33 pm »

"The beauty of the cumulative system is that it is possible for a state to be a more popular second choice than a first choice, giving the victory to the state of highest compromise between the voters."

This still happens with the system I'm talking about.  You like one a whole lot you give it a 5, you like another one alot and give it a 4.  If people give out one 5 each to a bunch of different states but they mostly all give a 4 to say Delaware then Delaware wins.  In fact I'd even go as far to say that people's second choice will get little to no credit under the current system because if you think there's a chance you favorite will win then you don't have any incentive to give your second choice any points because it will deminish the 'power' of your first choice.

"What if I opted out of choice C when I joined? Should I have to grant points to it? "

AHHHHH!  No no no you are missing the whole idea.  if you give c a '1' on a scale of 1-5 then you are NOT giving it any points!
We could just say the scale is 0-4 then you could give it a '0' and be happy, or we could have the scale be
-10 -9 -8 -7 -6 with -6 as the 'best' choice and who ever comes out with this smallest negative number wins.  OR we could have the scale be
6.02*10^23 + 1, 6.02*10^23 + 2, 6.02*10^23 + 3, 6.02*10^23 + 4, 6.02*10^23 + 5
and the results would be the same.  If you give a state the lowest score then you are NOT "helping" it just think of it like this
1- Strongly dissapprove
2- dissapprove
3- neutral
4- approve
5- strongly approve

-Eddie
« Last Edit: July 30, 2002, 03:05:29 pm by Eddie_Bradford »
Logged

JasonPSorens

  • Administrator
  • *****
  • Offline Offline
  • Posts: 5720
  • Neohantonum liberissimum erit.
    • My Homepage
Re:Problems with all voting
« Reply #14 on: July 30, 2002, 05:04:12 pm »

Under Cumulative Count, if you don't know which state is the frontrunner, it would be silly to give all your votes to one state because you would be effectively risking everything on that 1 state.  If your 2nd and 3rd preferences, say, are even mildly important, you should assign some points to them and hope that if your 1st choice doesn't win, your 2nd or 3rd preference votes would make a difference.  So I think under Cumulative Count people will vote honestly (they will if they think about it at all).  Sure, it's possible for a strongly partisan minority to overrule an apathetic majority - but that's a good thing!  Cumulative Count is the only system that really treats *strength* of preference seriously.  People with strong preferences should be allowed to express that fully.  That's why we chose CC after considering it along with the Approval Voting system (similar to what Eddie has proposed) along with other systems.
Logged
"Educate your children, educate yourselves, in the love for the freedom of others, for only in this way will your own freedom not be a gratuitous gift from fate. You will be aware of its worth and will have the courage to defend it." --Joaquim Nabuco (1883), Abolitionism
Pages: [1] 2   Go Up