Electoral reform ideas

[Gabu]

Given that a presidential term lasts only four months, it seems to me that the more important thing is to try our hardest to ensure that this sort of thing doesn't happen again.

I've already introduced a bill dealing with the issue of whether or not listing one single candidate in a presidential vote should be valid, and I'll get around to fixing the dumb wording in the part that sparked the Boss Tweed controversy, as well.  Those are not the hard parts to fix, however.

If nothing else, the election has made me feel quite sure of the following things:

1. That there is a very serious problem with IRV in that the results can wildly vary simply by whether or not one candidate is knocked off at any given candidate, and
2. That there is a very serious problem with tactical voting in that it is currently in the voters' best interest not to simply make an honest vote based on who they truly want to win, but to make a vote at the late stages of the game for the sole purpose of getting a candidate to either lose or win.

Both issues seriously detract from the essence of what democracy is supposed to be about: for people to simply all give their opinion regarding what they want and have what most people want to occur actually come to pass.  I personally believe that the best solution is, first, to get rid of IRV in order not to have a voting system where voting for a candidate highly might make that candidate lose in actuality, and second, to establish a secret ballot, to eliminate the possibility of tactical voting, thereby restoring the essence of democracy by forcing each person to simply vote for who they wish to win.

There are, obviously, details to work out, however, which is the purpose of this topic.  I have ideas regarding what to do about both points, and thought that, before putting them into formal writing in the form of a bill introduced into the Senate, I would present them to the general public to get input.

Replacement for IRV

After considering all of our options, I personally feel that the best option is to install something simpler than IRV, since history has shown that figuring out the correct result of IRV races is a lot more difficult than one might expect.  While Condorcet might be a good system in theory, from what I've seen of it, it would be entirely too complicated to actually implemented in a practical setting, and we would, many times, be forced to simply manually insert every vote into an automated process found on the internet and hope that it gives us the right answer.

What I think is the best combination of applying simplicity while disallowing people to win with something like 28% of the vote (as would have happened were the last election run using FPTP) is a form of FPTP with a runoff included.  One suggestion is the Louisiana-style form of this, where the top two receivers of votes advance to a runoff in which the voters vote a second time, with their choices restricted to only those two.  The problem I see with this that has been partly voiced in the past is that, if we have the case where there are many candidates that amass a sizable amount of support (as was the case in the last election), then there will be many candidates knocked out who were perhaps only a vote away from being in second place.  Ideally, this should be avoided as best as possible without requiring a runoff per every number of candidate greater than 2.

John Ford has previously suggested that we simply make the runoff go to the top 3 vote-getters instead of the top 2, but the problem is that, again, this is an absolute number (and that, if we only have three candidates, this runoff will not exactly do a whole lot).

Here's my idea.  Suppose we have n candidates in an election.  After the initial vote, we first eliminate any candidate receiving at most 100/(3n) percent of the vote in order to remove those with limited support.  For example, in a 4-way race, any candidates receiving at most 100/(3 * 4) = 8.3% of the vote will be eliminated.  Then, left with m candidates, we will advance the top x candidates to a second and final runoff, where x = ceil(m/2) (the smallest integer greater than or equal to m/2 - basically, just round up).  Then, in the runoff, whoever gets the most votes wins.

For example, consider the election we just had.  There were 6 candidates, so the percentage at which which candidates are dropped is 100/(3 * 6) = 5.6%.  Jesus only had exactly 5.6% (this is a weird coincidence I did not intend to have occur), so he will be eliminated.  We have 5 candidates left, so we advance ceil(5/2) = 3 candidates to the final runoff - Emsworth, Supersoulty, and Joe Republic.  Then, the people vote on those three candidates, and whoever gets the most votes wins the election.

The number of advancing candidates for various values of m are as follows:

m = 3: 2 candidates advance
m = 4: 2 candidates advance
m = 5: 3 candidates advance
m = 6: 3 candidates advance
m = 7: 4 candidates advance

...and so on.

There is, quite simply, no perfect election system.  There are several mutually exclusive aims that one wants to achieve in any given election system.  However, I feel that this is a good compromise between simplicity (i.e., it does not take hours of data entry and/or calculation to determine the result) and the full accomodation of the voters' actual desires.  This system is designed such that voting for any given candidate can only improve that candidate's chances of winning, so we don't run into the weirdness in IRV of preferencing a candidate higher actually having the possibility of making that candidate do worse than if you had preferenced him lower.

The system is also designed such that voting for a candidate who is not likely to come in first in the initial election is not "throwing away your vote" and such that another less popular candidate entering is less likely to act as a "spoiler candidate", since the number of voters who progress to the final round increases linearly with the number of candidates in the initial round.  That less popular candidate will likely be simply eliminated, sending the remaining candidates to an unspoiled runoff.

Secret ballot implementation

The implementation of a secret ballot is deceptively difficult to get right.  The issues arise from the fact that there are many things that are required in any given system, and that increasing the prominence of one item generally decreases the prominence of another.  The main items that one would want to be present in any system of a secret ballot are as follows:

1. Simplicity - can voters vote via secret ballot without so much trouble that most voters would choose to simply not use it?
2. Transparency - is it clear that no one has tampered with the secret ballot results?
3. Confidentiality - can voters be sure that they will not be publicly linked to a specific vote that was cast?

My idea, incorporating the ideas of several other people, goes as follows:

Voters will make one single vote and then send it to the SoFA.  In the voting booth, the SoFA will provide a string containing his or her username that can then be directly copied and pasted into the "To" section of the new message interface, in order to make the voters' jobs as easy as possible.  If any ballot is found to be invalid for whatever reason in a rectifiable manner, the SoFA will discard the ballot and will then inform the voter of this fact and give the voter a chance to send a new ballot.  I don't consider this to be a problem because this is just equivalent to having the voter vote at a later time, which is perfectly legal and rightfully so.

On each ballot, the voter will randomly pick a 6-digit number to attach to the ballot as the ballot's ID number.  The ballots will be publicly displayed en masse once voting has closed with the ID numbers attached to them so all voters can ensure that their ballot has been counted.  In the unlikely event that two voters pick the same 6-digit number, letters of the alphabet will be appended to the end, starting with "a", going to "z", then going to "aa", "ab", and so on, in order to differentiate between the ballots, like so:

125322a
125322b
125322c
...
125322z
125322aa
(etc.)

When this occurs, the SoFA will inform those choosing that number that this appending of a letter has occurred, so the voter will still know which ballot is theirs.  Failure to attach a 6-digit number will render a ballot invalid and the measure described above will be taken.  Any votes discarded because they were invalid for whatever reason must still be produced by the SoFA here and a reason for its discarding must be given.  Since the voter whose vote was discarded will be able to identify the ballot as theirs through the ID number, the voter will then be able to challenge the discarding of the ballot if he or she so chooses.

I don't consider sending the ballot to two other people other than the SoFA to be necessary as long as the ID number idea is used, because the only point of sending it to more than one person would be to ensure that the SoFA has not tampered with the ballots, and having the SoFA publicly post all of the ballots will also achieve this goal.

Voting publicly, as we currently do, can still be an option for those who still don't trust the secret ballot method.

The following activities will be made criminal, to ensure that our three goals have been met:

1. Revealing the current results of the secret ballot at any time before voting is closed.
2. Publicly linking a voter to a ballot in any way at any time at all.
3. Tampering with the ballots in any way.

----

So, that about wraps it up for my proposal.  What do you all think?

[Gabu]

A note: Come to think of it, we could just have the SoFA do the assigning of the ID numbers.  That would further lessen the effort that the voter would have to go to and would clear up numerical conflicts.  We could just have the SoFA reply to each ballot with the ID number assigned to that ballot; that way, the voter would also have a record stored somewhere of their ID number.

I think I actually like this method more.

[TomC]

I'm with ya in the spirit of what you are trying to do, but the IRV replacement is too complicated. I'm sure some who want to analyze the votes will get it, but if we try explaining to potential new voters, we're gonna lose some when we start quoting formulas. I say we advance the top three, or we pick a percentage like 15 and say anyone who gets above 15 can advance. The formulas don't really solve the problem of having voters understand the results enough. {Ceil? What the hell is that??}

Your ballot proposal looks pretty good on first read. It makes it even more important that we have confidence the sofa can do the job.

[Gabu]

I'm with ya in the spirit of what you are trying to do, but the IRV replacement is too complicated. I'm sure some who want to analyze the votes will get it, but if we try explaining to potential new voters, we're gonna lose some when we start quoting formulas. I say we advance the top three, or we pick a percentage like 15 and say anyone who gets above 15 can advance. The formulas don't really solve the problem of having voters understand the results enough. {Ceil? What the hell is that??}

True enough.  This is only a rough draft, presented just so I can get comments like that on it.

It's not really a complicated concept, though; can you think of any way to word it so we don't have any scary-looking stuff?  Intuitively it feels like there should be, given that, for example, "ceil" is just saying "take this and round it up".  I'll think about it.  I simply feel that there should be a way to somehow incorporate the number of voters  in how many voters advance, given that I don't think it should be an absolute number for the reasons I stated in my post.

Your ballot proposal looks pretty good on first read. It makes it even more important that we have confidence the sofa can do the job.

Yes, it does make that important, but I think that we should make our proposals with the assumption that our SoFA can do the job.  If we dumb things down so much just because we're afraid that our SoFA might not be able to do his job, it seems to me that we might as well abolish the position, in that case.  If a SoFA isn't doing his job, he should be replaced, rather than having our regulations changed to accomodate that.

[Gabu]

Are there any other comments from anyone else?  Did I make my first post too long?

[The Duke]

One problem with the 100/n idea is that you could have a circumstance where the law mandates a runoff when only one candidate qualifies for said runoff.

If you have four candidates, A, B, C, & D, you'd obvioulsy need 25% to davance to a runoff.  Let's say A gets 21%, B gets 24%, C gets 23%, and D gets 33%.  D would be placed in a runoff against himself.  If you set up a rule where this would result in an automatic D victory, the logical fail safe in this situation, you've set up a system where someone could win without a runoff when their vote total is in the low 30s in a four way race.

Or you could have a situation where you have a huge number of candidates setting the bar very low for advancing to a runoff.  This time we had 6 candidates (Emsworth, Joe, Super, LiberalPA, Dubya, & Jesus), setting the bar at only 16% to advance to a runoff.  You could wind up with five candidates, all of them but Jesus, advancing to a runoff.

[Gabu]

One problem with the 100/n idea is that you could have a circumstance where the law mandates a runoff when only one candidate qualifies for said runoff.

If you have four candidates, A, B, C, & D, you'd obvioulsy need 25% to davance to a runoff.  Let's say A gets 21%, B gets 24%, C gets 23%, and D gets 33%.  D would be placed in a runoff against himself.  If you set up a rule where this would result in an automatic D victory, the logical fail safe in this situation, you've set up a system where someone could win without a runoff when their vote total is in the low 30s in a four way race.

Or you could have a situation where you have a huge number of candidates setting the bar very low for advancing to a runoff.  This time we had 6 candidates (Emsworth, Joe, Super, LiberalPA, Dubya, & Jesus), setting the bar at only 16% to advance to a runoff.  You could wind up with five candidates, all of them but Jesus, advancing to a runoff.

I think you read my post too quickly, as what I'm proposing for the replacement for IRV is not what I posted in that other topic in the Atlas Forum Government; it's a modified version.

[Bono]

I have a better idea.
Let's form a comission.

[WiseGuy]

I think the SoFA should PM the ballots to all the eligible voters, with the ID number on it.  From there the voter could decide whether to use the secret ballot, or post it directly to the voting booth.

And how is the SoFA going to post the votes, because I don't feel comfortable giving him an opportunity to type in the PM, even with the ID number.  Maybe he should take a screen capture and crop out the voter's name.

Also, I think we should say that  a third of the candidates advance.  Other than all that, your suggestions are great!

[minionofmidas]

Counting Condorcet is not more complicated than counting IRV. (It's not less complicated either, I'll grant you that.) I've found out that much last night.

That said, I could live with runoffs.

[TomC]

Your ballot proposal looks pretty good on first read. It makes it even more important that we have confidence the sofa can do the job.

Yes, it does make that important, but I think that we should make our proposals with the assumption that our SoFA can do the job.  If we dumb things down so much just because we're afraid that our SoFA might not be able to do his job, it seems to me that we might as well abolish the position, in that case.  If a SoFA isn't doing his job, he should be replaced, rather than having our regulations changed to accomodate that.

No, I wasn't saying dumb down the secret ballot policy to accommodate a less than competent sofa, I was saying make sure we pick a highly competent sofa so we can have faith in a secret ballot practice.

[The Duke]

Gabu, I re-read the post and I think that idea is brilliant.  Flexible but firm.

[Jake]

Gabu's proposal for a run-off style is very good. I still stand opposed to secret balloting because of how complicated we may have to make it to make it secure.

[Peter]

No, I wasn't saying dumb down the secret ballot policy to accommodate a less than competent sofa, I was saying make sure we pick a highly competent sofa so we can have faith in a secret ballot practice.

One of the events that lead to Verin being selected was the fact that many of the arguably most qualified persons to hold the office (myself included) declined to take the post.

[Gabu]

I think the SoFA should PM the ballots to all the eligible voters, with the ID number on it.  From there the voter could decide whether to use the secret ballot, or post it directly to the voting booth.

That sounds like quite a bit of work on the SoFA's part, given that he would need to determine the usernames of all eligible voters and then send all of them different messages (since the ID number would have to be different in each one).  It seems to me that it would be much simpler just to post a ballot that one can copy and paste into a PM, like this:

[ ] Apple
[ ] Orange
[ ] Peach
[ ] Write-in: ______

...and then the voter could PM that with an X in one of the boxes to the SoFA, at which point the SoFA assigns the ballot an ID.  Do you see a problem with this?  I don't immediately see why this wouldn't work.

And how is the SoFA going to post the votes, because I don't feel comfortable giving him an opportunity to type in the PM, even with the ID number.  Maybe he should take a screen capture and crop out the voter's name.

I was under the impression that each ballot would just be posted like this:

Ballot ID 235764

[ ] Apple
[X] Orange
[ ] Peach
[ ] Write-in: ______

It would be completely anonymous unless the user shared his ID with someone else.

Also, I think we should say that  a third of the candidates advance.  Other than all that, your suggestions are great!

Why a third?  That would make only two candidates advance unless we had at least six candidates with a sizable amount of support, which seems like it would kind of restrict voter choices.

[A18]

How do you mark a ballot under this system?

[ilikeverin]

This system sounds quite good

[Gabu]

How do you mark a ballot under this system?

You put your cursor between a [ and a ], hold shift, and then hit the X key.

[Gabu]

To whom it may concern: I have now introduced a bill implementing these ideas to the Senate.

[TomC]

In places it seems voters are casting one and only one vote for president, and elsewhere it talks about "highest preference votes," like here:

§10. Concession of Victory.

   1. If a candidate shall concede his or her victory of a Senate election after the certification of the election result, then the candidate with the next greatest number of highest preference votes in that election shall then be declared victor.
   2. If both members of a Presidential ticket shall concede their victory in the Presidential election after the certification of the election result, then the members of the ticket with the next greatest number of highest preference votes in that election shall then be declared victors.
   3. If a victor who has conceded shall wish to retract his or her concession, then he or she shall only be able to do so with the permission of the newly declared victor.
   4. Concessions made before the certification of election results, or on or after the date on which the newly elected official is due to be sworn in, are of no legal effect whatsoever.

So does this mean that if a Presidential candidate wins and concedes, that the next Presidential candidate, who has not offered a concession speech, is declared victor, and the VP candidate of the top ticket does not have any claim to the office?  What if in this chaos, everybody but Jesus, at some time, offered a concession speech, as do the top vote getters, as both Joe and Emsworth have at this point, doesn't take office due to disenchantment with the process- would Jesus be President or say, DeFarge?

There's something about tying concession speeches to someone else's victory that worries me. The ticket with the most votes wins, period. Allowing for all this quitting is bad law- we voters vote expecting a winner to take office. And if too many quit we get a President that few approved of.
And the record here points to the possibility.


Oh, and  they choose one ticket, right? The preference stuff was carried over by accident?

[Gabu]

In places it seems voters are casting one and only one vote for president, and elsewhere it talks about "highest preference votes," like here:

§10. Concession of Victory.

   1. If a candidate shall concede his or her victory of a Senate election after the certification of the election result, then the candidate with the next greatest number of highest preference votes in that election shall then be declared victor.
   2. If both members of a Presidential ticket shall concede their victory in the Presidential election after the certification of the election result, then the members of the ticket with the next greatest number of highest preference votes in that election shall then be declared victors.
   3. If a victor who has conceded shall wish to retract his or her concession, then he or she shall only be able to do so with the permission of the newly declared victor.
   4. Concessions made before the certification of election results, or on or after the date on which the newly elected official is due to be sworn in, are of no legal effect whatsoever.

So does this mean that if a Presidential candidate wins and concedes, that the next Presidential candidate, who has not offered a concession speech, is declared victor, and the VP candidate of the top ticket does not have any claim to the office?  What if in this chaos, everybody but Jesus, at some time, offered a concession speech, as do the top vote getters, as both Joe and Emsworth have at this point, doesn't take office due to disenchantment with the process- would Jesus be President or say, DeFarge?

There's something about tying concession speeches to someone else's victory that worries me. The ticket with the most votes wins, period. Allowing for all this quitting is bad law- we voters vote expecting a winner to take office. And if too many quit we get a President that few approved of.
And the record here points to the possibility.


Oh, and  they choose one ticket, right? The preference stuff was carried over by accident?

Oops.  I copied and pasted a few chunks of the Unified Electoral Code Act, and I guess I didn't check closely enough to make sure that everything I was posting was relevant.

I'll edit that; one sec.

[Hatman 🍁]

Gabu, you had me until you started with the math

[Gabu]

Okay, I fixed that.  If we want to change the stuff about concessions, I don't think that now is the time; if we change too many things with one single bill, it will increasingly have the chance of losing too many supporters.  That's a minor thing, anyway.

Are there any other spots talking about preferences that I missed?

Gabu, you had me until you started with the math

See clauses 6 and 7 of section 3 in the bill that I linked to above; that may clear things up, as I explain what's going on in plain language instead of symbols and whatnot.

[○∙◄☻¥tπ[╪AV┼cVê└]

What's wrong with Condorcet? You make one of those tables showing how many candidates prefer A over B. It's really simple if there's a Condorcet winner (Joe Republic was the Condorcet winner of this last election). A Condorcet winner is someone who beats everyone else in a 2 man race.

If there's not a Condorcet winner, some sort of tie breaking can be done. You use the results in the table for the tie breaking.

The Schultz method is a good one. It's a simple maximum flow problem. Assume you have pipes of given inches diameter. You are finding the path from A to B which has the thickest pipes (at its minimum).

http://en.wikipedia.org/wiki/Schulze_method

You can test out this last election with many voting systems here. I included the top 5 candidates. Add a "J>E" line if Boss Tweed's vote is valid.
Go here to test:
http://condorcet.ericgorr.net/

D>E
D>S>J>L
D>S>J
D>J>S>E
D>J>S>E
D>E>J>S>L
D>S>J
D>S
D>J>E
D>E>J>S
D>J>E>S
D>S>J>E>L
D>J>S>E
D>S>E>J
D>E>S>J>L

E>S>J>L
E>S>J
E>J>S>L
E
E>J>S
E>D>J>S>L
E>J>D
E>D>J>S>L
E>J>S>L>D
E
E>J>L
E
E>S>D>J>L
E>J>S>L>D
E>L
E>D
E>J>S>D>L
E>S>J
E>S>D>J>L
E>J>D>S>L
E>L>J
E>L>J
E>L>S

L>J>S
L>J>E>D
L>J>E
E>J>S>D>L

J>S>D>E
J>S>E>D>L
J>S>E>L>D
J>S>E>L>D
J>S>D>E
J>L>E>S>D
J>S>D>E>L
J>D>S>E>L
J>L>D
J>S>L>E
J>E>D
J>E

L>J>S>E
L>J>S>E
L>J>E>S
L>E>J>S
L>J>S>E>D
L>E>J>S>D
L>J>E>D>S
L
L>S>E
L>E
L>E>J
L>E>J>S>D

J>S>E>D>L
J>E>S>D>L
E>J>D>S>L

S>J>E
S>J>D>E
S>D>J>E>L
S>J>D>L>E
S>D>J>L
S>J>D>E>L
S>J>E>D>L
S>D>J>E>L
S>D>E>J
S>J>L>E>D
S>J>L>D
S
S>E>D>J>L
S>D>J>E>L
S

[Platypus]

preferential is not the probnlem. It IS the best method. The problem is the barstardised version used in atlasia.