I think the 5% threshold makes "targetted killing" a lot less viable, but it doesn't eliminate the possibility altogether. It may still be possible to use a slightly-more-mainstream party in such a way; if a party gets 5% of the vote nationwide but 15% in Trudeau's district, that's still a relative vote share of 3, which should be far more than Trudeau achieves. Still, I think the combination of a threshold with a "you don't actually need to be electing to Parliament to hold a leadership position" rule should be sufficient to reduce the "targetted killed" problem to a tolerable level at least.
The more I think about it though, the more I lean towards something closer to MMP over the 100% single-winner districts approaches. If a party's best showing in any one district is under 15%, I think it makes the most sense for that party to only have at-large seats; the remaining 85% of that district's voters will really feel like they drew the short straw otherwise. (Also, I imagine the possibility of losing one's seat to a third-party candidate with far less support within one's district would cause a lot of incumbents to oppose such a system.) Still, I think there is considerable middle ground between MMP and your system that allows for having substantially fewer at-large seats (as a fraction of all seats) than you get in Germany or New Zealand, and this middle group could easily be better than MMP.
I confess that I don't understand the detailed working of Fair Majority Voting either; it's clear enough in the two-party case, but I'd have to study it in a lot more detail to learn how the algorithm works in general.
What you're proposing is closely related to Michel Balinski's system of Fair Majority Voting. Both systems get proportional representation while having a representative from every district and without any sort of at-large seats. I think your system can be thought of as a greedy approximation of Balinski's, like how Sequential Proportional Approval Voting is a greedy approximation of Proportional Approval Voting. Fair Majority Voting should elect slightly better winners, but your algorithm is a lot easier to describe.
Warren Smith has criticized Fair Majority Voting while proposing something similar (which he also finds very problematic). Smith's main objection is that it's easy for tactical voters to unseat prominent MPs in what he calls "targetted killing":
With Balinski, the NaziLoon party (which gets 1% of the votes and hence deserves 1% of the seats) necessarily will enjoy a multiplicative scale-up factor of about 50, as compared with the Mainstream party (which gets 50% of the votes) with a multiplier of about 1. In other words, Balinski is going to force each NaziLoon vote to count about the same as 50 Mainstream votes.
This is a huge "leverage" or "amplification" factor subject to easy exploitation via "strategic voting." For example, suppose for the purposes of argument that Justin Trudeau (leader of Canada's Liberal party, recently elected prime minister in 2015) is the most popular MP in Canada. So the Tories might like it considerably if Trudeau's district were to un-elect him and thus remove him from power. Tory leadership can easily accomplish that goal. They simply tell the Tories in Trudeau's district "we suggest that instead of voting Tory, which is a lost cause, you vote NaziLoon." The 50× amplification factor then easily carries the day, and – voila – Trudeau is gone.
I call this "targeted killing." Now the Tories might now want to try that again, selecting as their target some other important Liberal MP. If, however, they kept re-using the NaziLoon party as their pawn in the targeted-killing game, it would no longer work as well. So instead, for their next "target" the Tories would switch to using the BeerLovers party as the (new) pawn. And so on.
This would cause all the important leaders in Canada to be "assassinated" – removed from political power – and instead a ragtag random collection of normally unelectable NaziLoons, BeerLovers, etc would be all over parliament. So perhaps the best two-word description of the situation that would be caused if Balinski's voting system were the design for Canada, would be "complete devastation."
Your system also looks vulnerable to this - if tactical voters cause the NaziLoon party to get 5% of the vote in Trudeau's district (compared to 1% nationwide) this would yield a relative vote share of 5, which should easily be enough to oust Trudeau. I think Smith overestimates the severity of this problem since it would be possible to enact a rule like "You don't actually need to get elected in order to be the prime minister so long as you got at least 40% of the vote in your district", but this is still a relevant consideration when there are other MPs that people want to remove. For example, if 10% of the electorate in each district was bigoted against a particular minority they could use this strategy to keep that minority out of parliament.
I fully agree with CES's decision not to commit any more resources in Denver or Broomfield. I also agree that "approval voting can thrive as a candidate and be implemented so that it may be tested alongside IRV", though I expect that most of this thriving will occur in places like Fargo where technical challenges prevent the implementation of IRV (or even STAR).
However, it's far too soon to dismiss the chances of STAR voting catching on. Getting it adopted in the first city will be the hardest since it doesn't have much of a track record. Portland didn't actually get to vote on it (they failed to get it onto the ballot due to a lack of funds), and the election where it was rejected occurred in Lane County, OR, where it still got 47.6% of the vote. Had they only tried to get STAR voting in the city of Eugene (the largest city in Lane County) they would have almost certainly succeeded; as you noted, "voting methods poll much, much better in cities", and it received over 70% of the vote where STAR advocates had done most of their campaigning (mainly in Eugene). Subsequent efforts to get STAR voting on the ballot in Eugene have been stymied by the pandemic and by some of the signatures they collected being wrongly rejected, but it's going to happen sooner or later and STAR voting is highly likely to pass when it does.
I don't blame you for dismissing Condorcet though; I'm in a very small minority in viewing it as politically viable and I don't know of anyone else who has written about why it can catch on.
(Some background: I have worked with CES on an unsuccessful effort to get approval voting on the ballot in Denver and wrote a grant request for a campaign in Broomfield. Nowadays I mainly work on voting methods with the League of Women Voters and the Equal Vote Coalition.)
I agree with CES on basically every point mentioned here. However, I believe that it’s possible to have a greater impact in the voting methods area by focusing on voting methods other than approval - specifically, STAR and Condorcet methods. Part of this is because I think these voting methods do a better job than approval at electing optimal winners and incentivizing politicians, but I won’t belabor this point since CES agrees with me about STAR yielding slightly better winners than approval. Also, the difference between approval and the best possible single-winner voting method is smaller than the difference between approval and plurality, or even between approval and IRV (probably). The main reason I think we can have more impact by advocating for STAR or Condorcet is that I believe they are more likely to outcompete IRV nationwide if they are given the level of exposure that approval voting has received.
While approval is a better voting method than IRV, I believe that IRV looks better superficially. Ranking is more expressive than up-or-down approval, and intuitively it seems like a voting method that uses more data will yield better outcomes. Also, the argument that approval voting is vulnerable to strategic bullet voting (voting for only one candidate) gets a lot of traction, and voters don’t like feeling that they need to consider candidate viability or that strategic voting will decide elections. These arguments are flawed and Aaron has linked to some excellent refutations of them, but my point is that they are subtly flawed. Unless you dig into the topic enough to learn about things like the Burlington election, IRV is going to seem like the better option.
Such arguments are a lot easier to rebut when talking about STAR or a Condoret method. Both options are far more expressive than approval voting. Bullet voting is clearly a bad idea in STAR, and it takes a rather convoluted argument to make a case that strategic voting matters at all under Condorcet methods.
CES’s main argument for approval voting over methods like STAR and Condorcet is approval voting’s simplicity. But the simplicity of approval voting has a flip side: it’s also the least sexy of the reasonable voting methods. It’s a whole lot easier to get excited about ranking than about (maybe) approving multiple candidates - approval voting doesn’t feel like as big of a change.
I think sexiness is more valuable than simplicity or ease of implementation when it comes to getting a new voting method adopted. Before Broomfield’s city council debated between approval and IRV, the elections director gave a presentation that portrayed approval as providing basically the same benefits as IRV but at a much lower cost. Not a single council member even mentioned this advantage, and they unanimously went for IRV over approval voting. I expect voters to care even less about implementation costs than city councils. Altogether, approval voting ends up sounding like a cheaper and inferior alternative.
Even if simplicity is important, STAR voting can’t be too complicated to get enacted because IRV isn’t too complicated to get enacted and STAR is simpler than IRV. As for Condorcet, some Condorcet methods are also simpler and easier to implement than IRV. Condorcet methods have a reputation of being complicated enough to be incapable of catching on more broadly, but this reputation is entirely undeserved. As I’ve written here, Condorcet methods are uniquely positioned to take advantage of RCV’s momentum and can be far more acceptable to RCV advocates than approval or STAR.
In the two cities where approval voting is used, Fargo and St. Louis, it wasn’t technically feasible to use IRV. Also, they had some very impressive campaigns; in Fargo, my recollection is that the “Approval Voting Army” was bigger than either the Republican or Democratic campaigns, and St. Louis also had a huge force of volunteers distributing yard signs, sending texts, and doing other kinds of outreach. I think such strong groups of volunteers could have gotten just about any voting method enacted if it was technically feasible and had they opted to pursue it; these ballot measures didn’t succeed because approval voting is uniquely appealing.
Another big point for focusing on other voting methods is that having momentum makes it a lot easier to get a voting method enacted, and this means that there is far greater leverage in advocating for as-yet unused voting methods. If Seattle were to adopt IRV, it would provide a rather minor boost for IRV nationwide. If Seattle adopts approval voting it will be a far bigger deal - but probably not as big a deal as the victory in St. Louis since the difference between “used in Fargo, St. Lous, and Seattle” and “used in Fargo and St. Louis” is smaller than the difference between “used in Fargo and St. Louis” and “used in Fargo”. If Seattle were to adopt STAR or a Condorcet method it would a really big deal for those methods since it would actually put them on the map. The less known a voting method is, the more good we can accomplish by giving it its first victory.
I am not overwhelmingly confident that STAR and Condorcet have a better chance than approval of outcompeting IRV. I believe that it’s quite difficult to predict how well a particular voting method will perform in a statewide ballot measure based on theory and (maybe) a few local results and opinion polls, and CES’s views on the importance of simplicity may be correct. Operating under this uncertainty, I think the best thing we can do is to support all of these better voting methods at once so we succeed both in universes in which approval (but not STAR) can outcompete IRV and in which STAR (but not approval) can outcompete IRV. In addition to funding CES, I think it also makes sense to fund the Equal Vote Coalition. They focus primarily on STAR, but also support approval and Condorcet. I think it’s also important for groups like CES to endorse initiatives to adopt STAR voting when they appear on the ballot and to provide visibility so that local supporters can get involved.
I want to emphasize that I agree with Aaron far more than I disagree with him. While IRV is a real improvement over plurality, it’s still severely flawed and can fall flat on its face when it comes to delivering improvements like decreased polarization and ending the two-party system. I very much hope that approval voting will outcompete IRV; I just don’t think it will.
How exactly would the money flow? I don't know how much of what I described is actually legal. Maybe it would be necessary to give the money to a candidate's super PAC instead of to the campaign. Maybe there's other issues I don't know about. Is there a lawyer, accountant, or other knowledgeable person in the community who could propose a legal means of handling the finances?
What kind of a team would it take to implement this? There would obviously need to be someone to make the website and there should presumably be someone in charge of publicizing it, but what would be required beyond this?
Where else should I send this proposal in hopes of finding people who will implement it?
How important would it be for the organization to be bipartisan? That is to say, how important is it for there to be both Democrats and Republicans running Altruistic Partisanship?
Would it be possible for the fraction of a donation that goes to charity to be tax deductible?
Has anyone heard of someone trying this before? I have conducted several Google searches and browsed Charity Navigator and have found no mention of anything like it. However, if someone attempted to implement this but was (for instance) stymied by an insurmountable legal issue it could be very difficult to find.
What should be the procedure for determining which charities are acceptable? My inclination is to have a standard that any charity which a substantial number of people want to donate to should be acceptable so long as not many people think it would be harmful for that charity to receive money. Thus, I think most environmentalist groups and any organization that deals with guns or abortion (regardless of their stance) should be ineligible for donations via Altruistic Partisanship. There are many different ways of doing this though.
What would be the effect of this on nonpolitical donations that aren't made through Altruistic Partisanship? Suppose someone donates $200 a year to to charity and $50 (in presidential election years) to the Democratic candidate. It seems plausible that this person would make the $50 donation through Altruistic Partisanship, subconsciously think of it as fulfilling his “giving quota”, and then forgo the $200 donation for the year. While I doubt this would be especially common, it's the most plausible reason I've thought of so far for why Altruistic Partisanship might end up having a negative impact.
How should this be advertised? This would be relatively easy if the sole effect of Altruistic Partisanship was to redirect money from politics to charity, but the existence of this service could encourage people to make political donations they wouldn't make otherwise. My first thought was to have ads saying things such as “Which would you prefer? Donating $20 to Trump, or sending $20 of Crooked Hillary's money to charity?” The problem is that this also functions as a fundraising ad for Trump. In addition to possible legal issues, if advertising is done like this then the counterfactual political impact of a donation could be muted. A donation for Trump would encourage Altruistic Partisanship to advertise in order to get donations from Clinton supporters, and if this happened then the basic promise to Trump donors (that a $100 donation for Trump would cause Clinton's campaign to have $100 less) would be broken. Even if the advertisement itself wasn't partisan there could still be issues; an ad placed on a liberal website would lead to more donations for Clinton than for Trump and bring the same concerns with counterfactual political impact.
How could this be generalized to work in elections with more than two candidates?
What would the effects of this be on politics? I expect them to be very minor unless this were to become extremely popular (with more than 10% of all political donations running through this). It could make fund-raising slightly more important for candidates since the marginal value of a campaign dollar increases when you have less money to begin with (due to some of it going to charity). It could also reduce the political will to make political reforms that could lead to a multiparty system such as instant runoff voting; Altruistic Partisanship is far less workable in an election with more than two major candidates.
What should it be called? “Altruistic Partisanship” is just a placeholder.
This would make people feel better about their donations than they would otherwise. When people read Givewell's recommendations and decides to donate to AMF instead of the local homeless shelter they don't always feel pure joy at the knowledge that their donation is doing more good; they may also feel slightly bad about not helping the local homeless. With Altruistic Partisanship, people would feel good about helping charity in addition to feeling good about helping their candidate. Thus, Altruistic Partisanship could elicit donations from people who aren't even remotely interested in effective altruism.
There are disadvantages to using Altruistic Partisanship as opposed to donating directly to a candidate. There would presumably be extra fees associated with moving the money around, or at least overhead associated with operating Altruistic Partisanship. There would be a delay in getting money to a candidate to provide time for the other candidate to raise a similar amount. Also, the assumption that depriving a campaign of $100 is just as good as donating $100 to your preferred candidate would not hold if your candidate has raised far less money than the opponent (or perhaps has lower name recognition, etc.). Finally, some donors may prefer to do whatever feels the most like supporting their candidate with little regard for the consequences, and donating directly to their candidate would probably be more appealing to them.
Politicians, even ones who are personally uninterested in philanthropy, would be incentivized to support Altruistic Partisanship. They could find it useful as a fundraising pitch; people who are on the fence about whether to donate to a candidate may be pushed over the edge by the prospect of simultaneously helping charities at no additional cost. Moreover, soliciting donations through Altruistic Partisanship could improve a politician's image since it signals that he or she is a caring person.
There is a small but relevant chance that the majority of political contributions would eventually go through Altruistic Partisanship, even if most donors aren't actually interested in helping charities. Here's an example of how it might happen. One candidate decides to establish Altruistic Partisanship as the primary means of making a donation to his campaign, either because he wants charities to have more money or because he wants favorable media coverage. His opponent declines to follow suit and is criticized for being more interested in her campaign's coffers than in helping people. Other politicians conclude from this that using Altruistic Partisanship is politically advantageous and decide to do so in future campaigns. If enough of them do so then the use of Altruistic Partisanship will become a political norm, and every time a politician sends an email asking for money it will double as a fundraiser for charity. (There could be an API, or perhaps human coordinators, to make it easy for campaigns to solicit donations via Altruistic Partisanship.)
I think Altruistic Partisanship's core values should include doing the most good possible, maintaining political neutrality, and having the counterfactual political impact of a donation through Altruistic Partisanship be as close as possible to that of donating directly to a candidate. That is to say, donating through Altruistic Partisanship should be as politically effective as donating directly to a candidate.
Not just any charity could be chosen since some would be objectionable. If a Republican were to divert money from Clinton's campaign to Planned Parenthood, he might not consider it to be an improvement and would greatly prefer to have donated directly to Trump's campaign instead. I think the best approach would be to have a list of acceptable charities with Givewell's recommended charities conveniently located at the top.
For the sake of transparency, each donor should receive an email telling them where their money went and where the donations marked for the opposing candidate were redirected due to their donation. The effects of a single donation could be determined by multiplying the effects of all donations associated with the same candidate by the amount of the one donation and dividing by the total amounts of all donations for that that candidate.
A. Basic proposal
Over two billion dollars were spent on the 2012 US presidential election. About half this total was spent to make Obama win and Mitt Romney lose; most of the other half was spent to make Romney win and Obama lose. In aggregate this was incredibly wasteful, and there should be a better way of influencing an election than throwing money into a zero-sum game. Instead of funding opposing advertisements, campaign money should support programs that everyone considers to be beneficial; that is to say, it should go to charity.
It should be possible to make a nonprofit that would implement this, and here's how it could work. The nonprofit, which I shall give the placeholder name of Altruistic Partisanship, would run a website on which people could make donations. For each donation, the donor would specify a political candidate (or party) she wishes to support as well as an apolitical charity. Altruistic Partisanship would hold onto the money until the end of the current month, at which point the total amount of money raised for each candidate would be tallied up. The candidate who has raised the most money this way will receive a donation equal to the amount of money that candidate has raised in excess of what the opposing candidate has raised. The remaining money (equal to twice as much as was associated with the less-preferred candidate) would go to the charities specified by the donors.
Here's how it could work. Suppose that Clinton raises $1,000,000 through Altruistic Partisanship and Trump raises $800,000. For each Clinton donor, 1/5 of what they donated would go to the Clinton campaign and 4/5 would go the charity or charities they specified. For each Trump donor, all of what they donated would go to their preferred charity. Clinton's campaign would receive $200,000, Trump's would receive nothing, and $1,600,000 would go to charity.
As for the marginal effects of a donation, suppose someone were to have donated $100 through Altruistic Partisanship and specified Hillary Clinton as her preferred candidate and AMF as her preferred charity. If Clinton raised more through Altruistic Partisanship, the marginal effect of this donation is that Clinton's campaign would have an additional $100 (just as if she donated directly to the Clinton campaign). If Trump had raised more, the marginal effects of this donation would be a $100 reduction in Trump's campaign funds, AMF receiving $100, and the (apolitical) charities preferred by Trump's supporters receiving $100.
If Trump's campaign lost a dollar, Clinton's campaign lost a dollar, and a charity got two dollars, would anyone be worse off?
I have drafted a proposal for a nonprofit based on this idea. I'm looking for two kinds of help: 1) people who would like to take the idea and run with it, and 2) feedback regarding the feasibility of this idea or how it might be improved.
My draft is in the subcomments, and includes a description of the plan as well as a list of questions and concerns. I lack both the skills and the health to move this idea forward, and I would welcome any of you who can.
Two points regarding the possible downsides of proportional representation and multiparty systems:
First, different ways of implementing PR lead to different numbers of parties. If you just do nationwide party list, without a threshold like requiring at least 5% support to win any seats, there will be tons of parties, and the wheeling and dealing between them will ultimately seem more impactful than the actual election. But if you have multi-winner districts with < 10 seats each, only a few parties will win representation (but probably more than two). There's an important sense in which politics is compromise. With a two-party system, voters have to compromise and support the more tolerable of two candidates. Then, whichever party wins a majority gets to govern uninhibited by the other party (aside from filibusters, the possibility of a split congress, etc.). Or with a 10+ party system, voters hardly need to compromise at all; with so many options, chances are they'll be able to find a party they absolutely love. But this doesn't remove the need for compromise, it just shifts it onto the elected officials. I think the optimal system splits the difference, putting some of the weight of compromising on voters and some of it on elected officials. Having more than two parties, but not more than six or so, should achieve this. To the extent you're worried about the effects of having significantly more than three parties, this is entirely manageable with multi-winner districts - but I think your points have a fair amount of validity even for a three-party system.
(An additional consideration is that the US has a directly elected executive branch, so you wouldn't necessarily have the dealmaking to determine who gets to form a government that you get in most places that use PR.)
Second, many of the harms of polarization are different from what you describe happening under PR systems. I don't think the risk of a civil war or democratic backsliding is driven by a handful of extremists who need to be excluded, it's driven by the widespread hatred that Democrats and Republicans have for one another and the perceived threat the other party poses to democracy. I think this has relatively little to do with the difficulties of dealmaking and controversial policy concessions to fringe parties, even though I lumped the risk of civil war and congressional deadlock together as "polarization". In my model, I assumed that electoral reform would affect all the consequences of polarization equally. This could be very wrong! Perhaps proportional representation eliminates most of the congressional gridlock caused by zero-sum partisan maneuvering, and replaces it with congressional gridlock caused by the difficulty of getting several different factions to work together. (I find Lee Drutman's argument from Breaking the Two-Party Doom Loop that having more parties will result in a more functional congress to be quite persuasive, however.)