Slashdot Mirror

Alternative Vote can behave extremely oddly. Moving a candidate towards the top on your ballot can cause him to lose, and the Alternative Vote can also neglect to pick a candidate where a (different) majority prefers that candidate to each other candidate. The latter is what happened in the Burlington, Vermont election of 2009, and might have led the different majorities to unite against IRV (the Alternative Vote).

http://bolson.org/dist/

I think I've gotten pretty good results for CA, TX, IL, FL and PA

It tries to create impartial districts that keep people on average close to the center of their districts. It works pretty well, but is kinda computationally intense. It could almost become Redistricting@Home if there was interest in the approach.

Hand Counting is Cheaper than Voting Machines, even if you pay 3 people to redundantly count every ballot. No technology required beyond pen and paper.

I am a software engineer on emebedded systems. I see a lot of boards like this.

The ability to boot from different sources is a normal debugging feature, not in itself sinister. Should they have cleaned that up on the production model? Yeah, sure. But verifiability is ultimately a human concern anyway, not a tech one.

It all comes down to who you trust.

If you don't trust the polling place, make the voting machine tamper proof.
But then you have to trust the guy who built the voting machine.
You have to trust the guy who loaded the software on it at the factory or the elections office.
You have to trust the guy who wrote the code. Even if you inspected the code, you have to trust him to give you a binary based on that and not pull a fast one.
You have to trust his compiler to give him a binary without compiled in back doors.
I feel like I probably haven't listed all the points where this voting machine chain of trust can break down.

On top of all that, voting machines are not cost effective vs hand counted paper ballots. So, I advocate for no voting machines.

It's cheaper to count them by hand. A full county wide voting machine system costs a lot of money, a lot of money that could buy a lot of ballot counting labor hours.

I love a technofix as much as the next geek, but computerized voting machines are not the technology for now.

California has a similar law requiring the source code involved in voting machines. Diebold was cited or fined or something for having serviced election machines and changing out the software on them to something that was other than what had been submitted. So, an election was run on unverifiable source.

I still say the best solution is to ditch all voting machines and their software because it's cheaper to count elections by hand.

Easy. I argue that voting machines are a waste of money. It would be cheaper to count paper ballots by hand. When we can build super-cheap, super-reliable and super-trustworthy voting machines, then will be the time to move away from paper ballots. I'm a geek who loves the techno-fix for anything else, but it just doesn't make sense for voting.

It's the best format for posting my analysis of the Senate voting record:

http://bolson.org/gov/us/senate/2005/distgraph/sen ate20050428.svg

For fewer bytes I post better data than a comparable PNG.

http://bolson.org/ipodchart.html

Compare storage per dollar, storage per ounce, volume, density and other values.

I wrote a little calculator to help analyze the economies of voting machines. One of the problems of the recent election was that there weren't enough voting machines to go around, and cost may have played a factor. I heard from my country clerk that new HAVA complaiant DRE voting machines were often costing about $4000 each. Even if you estimate that they could be built for as little as $500, they're not economical. I combined the recent Ohio statistic of 100-200 votes per machine per election to arrive at the following conclusions.

A machine must last at least 7.5 elections to break even vs. hand counted ballots and possibly as many as 685.714285714286 .

Human cost per ballot counted $0.0583333333333333-$0.333333333333333.
Machine cost per ballot counted (over 10 elections) $0.25-$4.

http://bolson.org/cgi-bin/vote_tco

I wrote a little calculator to help analyze the economies of voting machines. One of the problems of the recent election was that there weren't enough voting machines to go around, and cost may have played a factor. I heard from my country clerk that new HAVA complaiant DRE voting machines were often costing about $4000 each. Even if you estimate that they could be built for as little as $500, they're not economical. I combined the recent Ohio statistic of 100-200 votes per machine per election to arrive at the following conclusions.

A machine must last at least 7.5 elections to break even vs. hand counted ballots and possibly as many as 685.714285714286 .

Human cost per ballot counted $0.0583333333333333-$0.333333333333333.
Machine cost per ballot counted (over 10 elections) $0.25-$4.

http://bolson.org/cgi-bin/vote_tco

No compromises, no strategies, no votes thrown away, no spoilers.

Vote your conscience about all of the choices. If you don't get your favorite you might still get your second or third choice.

You can vote the bums out, even your own bums, safely. If I don't like the crufty incumbent in my own party I can vote to prefer someone else, but still vote to keep my crufty incumbent over the alternative from the opposition.

(Yeah, this is kinda my holy cause.

Because I can write a better law than H.J.RES.109

http://bolson.org/voting/amendment.txt

Ah, IRV and other models have some assumptions. IRV assumes that the first choice is vastly preferred to the second, the second vastly preferred to the third, and so on. Condorcet, Borda, etc assume that it's a pretty even, linear scale and that one choice to the next is not such a large difference.

To get rid of both assumptions you have to vote on a ratings ballot. Rate each candidate on whatever scale and then you can explicitly record how much difference you feel there is between candidates. I suggest a system called Instant Runoff Normalized Ratings as a fair system for counting ratings ballots.

How to spin this when pitching it to Two Two Parties is that it lets you vote the bums out safely. When we have a distasteful incumbent within our own party, we can vote for someone else in our party over them, keep them as second or third choice and safely not elect some looser from the other party.

I have hope. I'm working with my prospective state assembly member to get a change to the California elections code introduced and hopefully passed. I write a bill for him, he gets some notoriety in the assembly, everybody wins. Ah, Democracy.

See also this poll which has some of the options other commenters want, and automatically tallies acording to many systems including Condorcet and IRV.

My take on this method is that the ballot should allow a voter to rate each choice (on a scale of 1..10, 1..100, -1.0 .. 1.0, whatever) and then behind the scenes the rating is normalized so that everyone has the same amount of vote (there's some debate over whether a L1 normalization or L2 normalization is more fair).

The next stage beyond that is a system I like to call Instant Runoff Normalized Ratings which I think really encourages people to vote honestly and put their true rating on each of the choices.

A is the first choice of 8 and second choice of 5.
B is the first choice of 5 and second choice of 4.

I think statements like that show that A has more support than B. Sometimes it does matter which voters rank A and B first and second, but this time I think that simplification is not invalid.

I think your fitting of real world names to this example is weak and effectively irrelevant. How many Bush voters do you know with Nader as their second choice? I find that R partisans are more likely to go Libertarian when they defect. The Nader,Kerry vote is reasonable, but I expect the Kerry,Bush crowd ought to be miniscule.

Here's a model that might more accurately represent this year's voting populace. Uninterestingly enough, everything but Approval agrees.

You might find this study interesting.

Simulation Of Various Voting Models for Close Elections by Brian Olsen

It's was simulation to see how well the overall happiness of the voting population was affected by voter error (i.e. mistakes, political lies) and number of choices. It found that One-vote and IRV had quite a lot less tolerance for errors than Rated, Approval, and Condorcet.

The order of best happiness was Rated, Condorcet, Approval, IRV, one-vote. However, he admits that he did not take into consideration strategic voting. He recommends Approval over Rated and Condorect due to this.

After poking around I came across this:

http://bolson.org/voting/cdata/contrived.html

It's full of examples showing that IRV gets the "wrong" answer. Could someone explain this to me? How can it choose wrongly? Isn't it just another way if getting an answer?

Looking at the first example, you can see IRV chooses differently, no doubt, but it still makes sense to me ...

17 voters, 3 candidates
8: A,C
5: B,A
4: C,B
Winner: B (the others choose A)

Alright, so lets put some names on this for fun, although it ends up not making much sense:

17 voters, 3 candidates
8: Bush,Nader
5: Kerry,Bush
4: Nader,Kerry
Winner: Kerry (I knew it!)

So IRV chooses Kerry, and the others choose Bush. I think I can see the argument that IRV is wrong - You've got 13 people that chose Bush as their first or second choice, 12 people that chose Nader as their first or second choice and 9 people that chose Kerry as their first or second choice, yet Kerry wins.

Not only that, but Bush almost had a majority to begin with and most of those that remained would pick him second. But everyone knows Bush shouldn't win so this must be wrong.

Seriously though, I think what IRV does is allow an election to be run where everyone can vote their conscience. Those 4 people that chose Nader (that's at least realistic, right?) are re-considered as if Nader hadn't even entered the race! And that's only done if no one manages to get a majority in round one.

Now we've got a new line up ... just Bush and Kerry. And in _that_ election, Kerry would win, given the numbers above.

That seems very fair to me, especially since Kerry ends up winning.

Kevin

That said, I think the positions they take are correct. There are some great theoretical results showing how poorly IRV performs in situations where Condorcet is a stable, rational system. Though there's not many elections that you can analyze to see this "in the wild", a recent Debian project leader election was a great example of where IRV would have been bad.

IRV isn't so bad, but it's also sadly inferior to many other better choices.

Rob

He has made a simulation that is open source.

So hack away. Look here and here.

He has made a simulation that is open source.

So hack away. Look here and here.

Take a look yourself on the graphs in Brian Olsons essay named Simulation Of Various Voting Models for Close Elections

And there is more info here

Take a look yourself on the graphs in Brian Olsons essay named Simulation Of Various Voting Models for Close Elections

And there is more info here

Simulation Of Various Voting Models for Close Elections

Here is the conclusion, I would like to have included the graphs but it don't seem to be possible.

Conclusions

Traditional One Vote was reliably the worst performer, and it should be sacked.

IRV is only barely better than present methods. I believe that the flaw is in that it is still based initially on the present method of only counting one vote from every person. Although it can solve some fractured electorate problems it is still very limited and unexpressive.

Acceptance Voting, while collecting less data then IRV, processes all that data at once to better match the will of the people. It is also simpler to implement and explain to a real electorate.

Condorcet voting, while not the top performer, had a uniquely better response to error in the voters. For this it deserves further study.

Ranked Voting collects the same data as IRV but uses it to better effect because the overall will of the voters is considered at once. The variant where no points are awarded to disliked candidates does not actually aid the voter. That variant is just throwing away data, and if a voter doesn't get a desired candidate they have no influence to get a lesser evil.

The various Rated Voting systems reliably got the highest happiness results. Unfortunately it would require relatively complicated or expensive, almost necessarily computerized, voting equipment. I'm imagining a GUI with a slider next to each candidate's name and image.

Fortunately, ``Rated 1..N'' and 1..10 don't have that problem and get results often identical to and always at least very close to pure Rated Voting.

Based on these findings Acceptance Voting or Rated Voting should be put into practice immediately. Acceptance Voting may be the easiest to implement with present equipment. The IRV advocacy sites I saw claimed that more and more voting equipment was capable of collecting IRV-type candidate rankings, but if such rankings can be collected, Rated Voting should be applied to the data.

An interesting trend in Max Happiness: it increases with more candidates. In a field of otherwise undifferentiated candidates, when there are more candidates there is a better chance that there will be one candidate that more people will agree on preferring. This may be the most important reason to break away from the One Vote system to a system that doesn't punish the voters for the existence of many choices.

The REAL reason is that because "throwing your vote away" is a very real phenomenon. It doesn't have to be. Ranked or Rated Ballots mean you can express your conscience on the ballot, no compromises, no throwing your vote away. Ballot access is a wink and a nod to say "good luck, chump" until the ballot enables better recording and enacting of the will of the people.

There's a long list of what's wrong with our democracy: campaign finance, voting machines, ballot access, the media, and so on. But my biggest issue is: One vote reinforces the Two Parties. Ranked or Rated ballots make voters matter again and make elections fair for any number of candidates. I think that'd revive democracy.

It took them over a month to connect my DSL. It's a long story. They are incompetant. They are probably breaking a handfull of FCC rules. I want very much to never do business with them again and I encourage others to avoid doing business with them.

Please use a better voting system

Like voting? More reasearch (mine) on what the "best" system is.

Condorcet doesn't always make the most people the happiest, but it's nicely impervious to several things that can go wrong with elections.

... there is a techno-fix for everything. Trustworthy, reliable voting machines record ballots for advanced election methods and we incontrovertibly know the will of the people within a minute of the polls closing.

On the other hand, I wonder if we could do better by paying minimum wage to two people per precinct to count each plain-old-paper-ballot twice, by hand. bc-of-envelope says $12/hr labor divided by 120ballots/hr processed = $.10 per ballot processing. About equal to the paper it's printed on. Sounds reasonable to me.

... there is a techno-fix for everything. Trustworthy, reliable voting machines record ballots for advanced election methods and we incontrovertibly know the will of the people within a minute of the polls closing.

On the other hand, I wonder if we could do better by paying minimum wage to two people per precinct to count each plain-old-paper-ballot twice, by hand. bc-of-envelope says $12/hr labor divided by 120ballots/hr processed = $.10 per ballot processing. About equal to the paper it's printed on. Sounds reasonable to me.

The Condorcet method of voting requires that each voter rank the candidates from best to worst. It's generally a good system, but has been criticised for being hard to understand (maybe not for those of us on /., but for the stupid voters). Another interesting voting method is range voting, which assigns a number value to each candidate based on that candidate's desirability.

Rated voting, which is a special case of range voting, was generally the best method (i.e., it maximised voter happiness) in a test of various voting systems. Also see ElectionMethods.org.

An improved voting system would certainly make lots of things better (though due to Arrow's paradox, a perfect system is impossible). I think we also need to improve the voters. The most heard criticism of Condorcet's method is that it's hard to understand, and it's really not all that complex at all.

I have to disagree. My own research (you can download the code and run the statistical simulation yourself) shows that IRV is only barely better than the current One Vote system. Any other of the alternate voting systems I tested did better.

See overview and some data.

I have to disagree. My own research (you can download the code and run the statistical simulation yourself) shows that IRV is only barely better than the current One Vote system. Any other of the alternate voting systems I tested did better.

See overview and some data.

I've done some simulation work on this topic. http://bolson.org/voting/

Compares current model against several other models, one of which is being implemented in a few places, Instant Runoff Voting, and turns out to be the worst thing better than the current way!