Talk:SODA voting (Simple Optionally-Delegated Approval)

From Electowiki
Jump to: navigation, search

Opinionated sales pitces (hard sell)

One of SODA's advantages is that, more than almost any other system, it is compatible with either a two-party or a multiparty system.

While many are convinced that, in the long run, a multiparty system is healthier, it is still important for a voting system to compatible with two parties. Current politicians, winners under a two-party system, are in many cases the gatekeepers of reform. Yes, that's largely a case of the fox guarding the henhouse, but it's also a fact of life.

Here are the pitches one could make. Both of those pitches are feeding on the sicknesses inherent in a two-party duopoly. But that doesn't mean that SODA would feed those sicknesses or make them worse; it's just using them to make an argument that is still ethically justified. Remember, SODA is fully compatible with a multiparty world; in fact, as explained above, it is in an important sense more Condorcet compliant than a Condorcet system.

For a politician (Democrat or Republican)

So, why is SODA compatible with two parties? Here's the kind of pitch you could make to a Republican or Democratic politician. This pitch is a "hard sell" aimed at the toughest audience: a dyed-in-the-wool partisan. As such, it includes some disingenuous trash talk about other good systems and about third parties or independents. If you were dealing with a more-reasonable person who could put the interests of democracy above their partisan interests, you would not pitch it this way.

"SODA encourages most voters to vote for a single candidate, just as they do today. So an average joe, who wants to put as little thought as possible into his ballot, will still be voting for one of the major parties. With the large majority of ballots in the same two-party split as today, the minor parties will have essentially no choice but to delegate their vote to one of the majors, or relegate themselves to irrelevance. So all this will do is make it so that the Libertarians (for a Rebublican)/Greens (for a Democrat) are your allies, not spoilers.

"Any other system is more of a danger to you. You ever heard of a Condorcet Winner? No? Well, most systems try to elect a Condorcet Winner, and lemme tell you something: H. Ross Perot, that's what a Condorcet winner is. Somebody who comes up in the center, in between the two parties, and it doesn't matter how incompetent or unexperienced he is, because the Democratic voters prefer him to a Republican, and the Republican voters prefer him to a Democrat, so it doesn't matter, he could be two wheels short of a tricycle, there's still no way to beat him. Well, look at how SODA handles that. The Democrat and the Republican, they don't plan to delegate their votes, so they don't announce a preference order. And then it's pointless for the centrist, the Perot, to ask them to delegate their votes to him - they can't. So if the Perot guy wants to be in the game and delegate to someone - whoever he pre-announced before the election, if anybody - he can do that; if he wants to be just a protest candidate, he doesn't announce a delegation order up front, so he either wins or loses on his own. In the first case, he's just a minor candidate, like a Green or a Libertarian, and you don't have to worry about him any more than about them. In the second case, he's irrelevant, at worst a spoiler, just as under plurality. So either way, you're at least as well-off as you are today."

For a third-party voter

OK, maybe that was a little overboard. So here's a pitch for a third-party supporter, to balance it out.

"What do you want, in the end? People like you are in a minority, and I'm sure you realize that you won't take over the world overnight. So you want a fair hearing, you want a seat at the table. Most voting systems are just selling you pie in the sky, unrealistic dreams. One day, they say, you're going to convince a majority to join your team, and on that glorious day your team's gonna be in charge. How well has that worked for Republican and Democratic voters? How much important change do you see when the pendulum swings back and forth between those two parties? Not enough. The truth is, if you ever do grow enough to sway a majority, your big ideas are going to be watered down - and, what's worse, you're going to have to be pretending that the watered-down version is the real thing.

"But there's another option. You can keep having big ideas, and just have a system that doesn't shut you out of the room. There are a lot, a lot of people who aren't fully satisfied with 'their' party, who are looking for another option. Take off their two-party shackles, let them safely vote for someone else, and they'll jump at the chance. And there your party will be, with serious candidates, real campaigns, and 10, 15, 25% of the vote. No, that won't be enough to win, but it will darn well be enough to get some respect, to get your ideas a fair hearing, get some of them tried, which is what you need in order to grow. And if the major party supposedly on 'your' side doesn't listen, you will have the power to take those votes and go home. You know and I know that major party politicians, they call themselves leaders, but a lot of them what they really are is cowards. When you're sitting on a double-digit-percentage pile of votes, they will listen to you, trust me."

obsolete definition of SODA-PR

SODA-PR is the proportional representation version of the above.

First, to simplify the ballots, the population is separated into a "district" for each seat, and "districts" are grouped into sets of 2 or 3 "co-districts". The ballot for each district lists the candidates from that district in a larger font, and the candidates from its co-districts in a smaller font. Write-ins may be used to vote for candidates from other districts not listed on the ballot, so the districts only matter for ballot simplicity (Voters do not want to have a ballot with many dozens of candidates on it, but write-ins allow full freedom for those voters who want it). Larger parties will usually run one candidate per district; smaller parties may just run one candidate per co-district set.

As with SODA, voters can either vote an approval-style vote or delegate by bullet voting. Instead of a pre-announced ranking which allows them to choose a cutoff after the election, candidates have a pre-announced list of approvals that will be added to any votes delegated to them (beyond the votes they need to be elected). Candidates may make their approval of other candidates conditional on that approval being mutual, but they may not make it conditional on anything else about the other candidate's approvals. Thus, parties cannot protect corrupt members by forcing other candidates in the party to approve of them.

The counting process is as follows:

  • While there are more uneliminated candidates than empty seats:
    • Divide each ballot by the number of uneliminated candidates it approves
    • If there are any candidates with more than a Droop quota:
      • Elect the one with the highest score (previously "unique ballots")
      • Reweight ballots for the elected candidate to reduce the total weight by a Droop quota, starting with the ones delegated to that candidate
      • Add the elected candidate's pre-declared approvals to any remaining delegated ballots for that candidate
    • Otherwise:
      • Eliminate the candidate with the lowest score
      • Add the eliminated candidate's pre-declared approvals to any delegated ballots for that candidate
  • Finally, elect all remaining candidates to fill the remaining seats.

Note: If all votes are for one of the two main-party candidates in the voter's district, and if all candidates approve everyone from their party, and if the districts are divided fairly so that plurality would give a proportional result, SODA-PR gives the same results as plurality. These assumptions will not generally be perfectly true, but they will generally be close to true, so SODA-PR will give results that are recognizably similar to those of single-member districts. It is hoped that this would make it a more acceptable system to politicians who have won under single-winner rules.

More stuff moved from main page: notes on rules

The "Chicken Dilemma": resolved

One of the toughest situations for any voting system is the "Chicken Dilemma" between two near-clones. Say there are 60% "Blue" voters and 40% "Orange" voters; but the Blue voters are split between two candidates, say 35% for Navy and 25% for Sky. Any good voting system will allow one of the Blue candidates to win, if all the Blue voters support both of their candidates. But under many systems, there is a temptation for the Sky voters not to support Navy, so as to make sure that Sky, rather than Navy, is the winner. But if the Navy voters respond in kind, then Orange could defeat them both.

Chickendilemma.png

This is called the "Chicken Dilemma" because it resembles a game of chicken, in that both blue factions benefit from being the only one to deny support, but both lose if both deny support. In fact, since no single blue voter can reliably know exactly how other voters in the two blue factions will behave, it resembles a blindfolded game of chicken.

This dilemma can be a problem for many voting systems. Approval voting, Range voting, Majority Judgment, and many Condorcet systems such as Schulze voting — all of them good systems — all can have some version of this problem. There are a few voting systems, such as IRV and some versions of Condorcet, which correctly elect Navy no matter what the Sky voters do. But these systems go too far in fixing the problem, and thus get bad results in some situations that superficially resemble the true "chicken" scenario in some way. For instance, if Orange voters strongly prefer Sky to Navy, IRV pushes them to dishonestly vote the "lesser evil" Sky over their true favorite Orange.

If candidate predeclarations are honest, SODA almost entirely resolves the chicken dilemma. The two blue candidates can each make sure that the other one includes them in the predeclared preference order. Then, if all votes are delegated, Navy, starting with more approvals, will add approvals to those delegated votes first (after Orange), and not approve Sky. Sky will be forced to approve Navy, or see the orange candidate win. Thus, Navy will win — the correct result.

If the Sky voters are especially strategically-inclined, they could give Sky non-delegable approvals instead of delegable votes. If Sky did not have enough delegable votes to help Navy defeat Orange, then Navy would be forced to approve Sky, and Sky would win. Of course, the Navy voters could do the same thing; either "defensively" or "in retaliation", it doesn't matter.

However, Orange could resolve this dilemma if they wished to. Remember, Orange was required to predeclare a preference between Navy and Sky to start out with. Say they preferred Navy. If Sky voters tried the trick of casting non-delegated votes, as explained in the previous, then Orange would simply give enough votes to Navy to prevent Navy being forced to give votes to Sky. This thus resolves the chicken dilemma (almost) completely, without causing other bad scenarios; something no other serious voting reform proposal can claim.

There is still technically a way this could fail: if Orange let it be known that they would not share votes despite their predeclarations, then the non-delegated trick could work for Sky voters, and so Navy voters could cast defensively non-delegated votes, and Orange could still win. And a similar process could happen if Orange voters cast non-delegated votes. However, this remnant of the Chicken dilemma involves third-order strategic thinking: Orange or their voters going out of their way to cause Navy voters to go out of their way to cause Sky voters not to go out of their way. It is rare enough to find evidence that voters are thinking in terms of even second-order strategy; the chances of third-order strategy being attempted seem negligible, and the chances of it working yet more so.

Prevent weak minor candidates from having kingmaker power

Step 3.b)ii. This rule helps make this system more attractive to major-party politicians. But it's a principled rule, not just a sop to the major parties. Consider the "kingmaker" case: in a basically 50/50 split, some tiny party has the balance of votes, and manages to extract concessions far bigger than their base of support justifies, just in order to delegate those votes or not. That's unjust, and this rule would prevent it.

5% is a good cutoff here; for instance, nationally in the US, that's tens of millions of voters, and enough to deserve a voice. It shouldn't be too high, because this rule is effectively taking power away from voters; that's only justified if the faction is so small that the power is not legitimate, and so it's better to err a bit on the small side if anything. But under 5% - that is, under 10% of the winning coalition - doesn't deserve kingmaker power.