America's Zero-Sum Electoral Systems
Or, "Why We Always Get Shitty Outcomes From Our Politics"
I’m currently reading Heather McGhee’s “The Sum Of Us: What Racism Costs Everyone and How We Can Prosper Together”, which is a really great book that you should go read. Heather is a very insightful writer who has managed to cut through a lot of the back-and-forth narrative wars that pass for reasoned debate in our society, and actually understands the underlying problems.
That said, while she writes quite thoroughly about how much of our politics has been infected by appealing to zero-sum impulses, she doesn’t really explain why those appeals have been so consistently successful. After all, our political mythology purports us to be a nation based on compromise and progress; these are positive-sum values, so zero-sum appeals ought to have difficulty finding purchase, right?
But this isn’t what we see. Something else is going on, something I think a lot of Americans are deeply uneasy with, but don’t really have the words to describe.
The motivation for writing this piece is a recognition that although I bring up “zero-sum” a lot, I probably haven’t ever laid out the specifics all at once. And it’s a very complicated, detailed story! One which I’ve been shorthanding for way too long. So I want to actually take the time to explain what exactly is zero-sum about each aspect of our system. But first, we have to talk about what even makes something “zero sum” to begin with.
What is “zero-sum”?
The term originates from game theory. It basically means that the sum of all possible outcomes is zero from the perspective of each actor. In a two-party election, for instance, if there can only be one winner, then from each party’s perspective, they either win (let’s code this as “+1”), or lose (let’s code this as “-1”). The sum of both outcomes is thus 0.
It’s not really all that helpful to obsess about the math of it all, so let’s translate this into something simpler and more recognizable - a moral principle. We can state that something is zero-sum if the victory condition is wholly excludable: one person’s win necessarily means that someone else - or everyone else - loses. Right off the bat, we can recognize that while individual political decision-making is ultimately zero-sum (either a bill passes, or it doesn’t), in the aggregate we can consider multiple bills and options, and passing any one of them does not inherently exclude all other options, although some will naturally exclude each other.
Conversely, while I’ll save deeper discussions for later articles, my general position is that the project of humanity - of progress - is finding ways to transcend the zero-sum world around us, by finding positive-sum solutions to our problems. There’s nothing more purely expressive of the notion of positive-sum than the distinctly American phrase “win-win scenario”. But positive-sum is hard work! It’s not all idealism and puppy dogs and unicorns. We can’t, for instance, declare that the Two-State-Solution is positive-sum, and then magically expect the Israelis and Palestinians to cease fire and agree to a partition. Rather, we must nurture as many positive-sum dynamics as we can, in order to overcome the zero-sum dynamics currently dominating each polity.
Finally, I just want us to keep in mind as we go along, that zero-sum dynamics are not necessarily all bad. Sometimes they’re the least bad option for accomplishing a particular goal we have for a given branch of government. Some decisions truly are binary and exclusive, and we do not have the luxury of not making them, or taking the time to attempt to think beyond them. My core objection is simply that there’s no historical indication we’ve ever exhaustively examined our political system and identified which aspects are zero-sum, let alone considered the pros, cons, or positive-sum alternatives1: they’re mostly a product of historical happenstance, ideas that either no one questioned because it was simply always done that way (as we’ll see with our first example), or were just popular whims enacted for other reasons (like direct election of Senators). And as a result, today we suffer at the mercy of a set of dynamics most of us don’t understand, and yet desperately want to rid ourselves of (more on this in the conclusion).
Which American institutions are zero-sum?
The best place to start is with First-Past-The-Post (FPTP). Almost every election in the country uses this method: the House, Senate, Electoral College (both the statewide votes and the actual EC ballot governed by the 12th Amendment2), and up and down state and municipal governments. It’s an import from our British colonial origins, and one of the starkest reminders that our Revolution didn’t repudiate nearly as much of British political culture as we like to tell ourselves. Also known as “plurality voting”, FPTP basically says that the candidate with the most votes wins, and voters can only vote for one candidate. It’s the single worst form of voting, because it fails almost all sensible criteria. At the end of the day, FPTP is quite obviously zero-sum because a vote for one excludes a vote for all others, and only one can win.
The next electoral dynamic is our single-member races. Again, nearly every state and federal race is for a single seat: the House, Senate, Presidency, governorships, and state legislatures. And again, it should also be quite obvious that single-member races are zero-sum: only one candidate can win, and all others are excluded. Now, as I alluded earlier, this is not necessarily all bad. For singular offices, like the Presidency, governorships, and arguably Senate seats, it’s kind of unavoidable, and even if it weren’t, constitutional concerns take pretty much all radical options for reform off the table. But there’s also no definitively good reason for the House to be solely made up of single-member districts. That’s not to say single-member constituencies don’t have their merits, especially considering that Americans’ intellectual love affair with parochialism means they’d be impossible to completely eradicate as a practical matter. But they’re so obviously easy to gerrymander, that most modern democracies combine them with a “mixed-member proportional” or “parallel voting” scheme: constituents vote first for a direct representative, and then for a party, and the party vote is used to top-up the legislature with at-large members so that the parties’ seat shares remain proportional to their national vote. Japan and Germany notably do this — and it’s notable because both countries’ legislatures were designed with a lot of help from… us.
Now is a good time to stop and bring up Duverger’s Law. It basically says that FPTP tends to lead to there only being two parties: since only a plurality is required, losers are incentivized to form coalitions to win the next election, which over repeated iterations results in two stable party coalitions whose only two options are to nibble at each other’s edges with wedge politics, or activate/persuade nonvoters. Duverger was originally only applied to FPTP, but I think the principle actually expands to pretty much all zero-sum systems3: the losers of any zero-sum struggle will always recruit allies. Regardless, it’s also helpful to think of it more like a rule of thumb, not an iron law, because there are some glaring exceptions to it like Canada and the UK, both of which have strong regional third parties.
One quick note about the Senate and direct election. What we had before the 17th Amendment wasn’t exactly positive-sum, but I think we can describe it as attenuating the zero-sum nature of having singular Senate seats. In practice, this probably wouldn’t make much of a difference today, since Senate election outcomes tend to match the partisan makeup of their state legislatures. But direct election definitely cemented Senate races as zero-sum, representing a very real step backward.
The last major zero-sum national institution is something people deeply hate, but misunderstand so much that they misdirect their hatred into arguing past each other about something entirely else: Winner-Take-All (WTA), with the “something entirely else” being the Electoral College, obviously. The EC’s WTA aspects actually work on two different levels: (1) as previously mentioned, the 12th Amendment requires a bare majority threshold when Congress counts the EC’s votes, and (2) except for Maine and Nebraska, the state EC delegations themselves are won on a statewide WTA basis - the entire slate of a state’s electors is given to the party that wins the statewide popular vote. I really want to call attention to how quixotic this is: We wouldn’t accept this kind of rule at any other level of our government! Imagine if your state gave its entire House delegation to the party who wins the statewide popular vote. Or if it gave the entire legislature to either party in a similar manner4. People would lose their shit! And it wasn’t always like this. The first 13 colonies were originally split into about even 1/3’s between WTA, proportionalized EC delegations, and a mixed bag of various other schemes.
Finally, we can’t truly wrap up the national institutions without talking about primaries. The conversation is kind of awkward, because they’re more “acknowledged” by federal/state law than “governed” by them; American primaries are this wierd hybrid private-public Rube-Goldberg machine. In terms of zero-sum, they’re also kind of a mixed bag, because they vary so wildly from state to state5. But overall, what we can say is that primaries themselves (and their attendant ballot access laws6) are mostly conducted according to zero-sum rules (FPTP, single-winner, WTA, etc.). This is intended to achieve two main goals: (1) to conform with the single-seat offices they aim to nominate candidates for, and (2) to resolve intra-party disputes definitively without leading to party splits. The flaw that was not apparent when primaries were instituted, but which is now widely noted, is that when parties are ideologically polarized, this empowers the very extremists that a “compromise-based” system needs to exclude.
Wrap-Up
Now, for the deviations. St. Louis, my hometown, recently adopted Approval Voting for the “jungle” mayoral primary. New York City likewise is implementing Ranked Choice Voting in its local elections, and Maine uses RCV for its federal elections. Somewhat longer in use are the Iowa caucuses, which use what’s essentially an RCV formula done through live headcounts.
And… that’s it. That’s why we’re stuck in such an ossified system: we never reformed any of these things during any of our previous periods of systemic reform (like Jacksonian democracy or direct election of Senators), so the whole thing just proceeded along the polarizing path it was always headed on. My hypothesis is that this is because while various coalitions have at turns won seemingly-durable majorities, no one region of the US has ever truly been beaten into submission. The postbellum South is the prime example: even badly beaten, their elites were able to reassert control within a generation. By contrast, almost every other functioning democracy in the world has been beaten at some point in the past 100 years to the point of collapse, which has given them opportunities to wipe the slate clean and apply lessons-learned from their catastrophic political failures (anocracy in Japan, too many parties in Germany). Without any catastrophic internal failures to push America to reform, our system instead is stuck on a path of long-term collapse as its contradictions play themselves to their logical conclusions.
In fact, the only such occurrence I can think of where anyone ever even tried to game these things out (although they still fell victim to their own blind spots), was indeed the American Constitutional Convention and Federalist papers. All that talk of “we must do this, or the colonies will end up in perpetual wars with each other” is quite obviously the Founders calling out a pretty dire zero-sum dynamic in order to argue for a positive-sum solution.
I should note that 12A specifically requires a majority of Electoral Votes as its threshold, rather than a mere plurality, but it is counted just the same: one vote for one candidate per voter.
Call it “Dave-rger’s Law”.
Hey, look, until the paid subscriptions start, the joke quality’s not improving.
Some of the original colonies actually did have this system in their early assemblies, and a few even kept it well into the early 19th century.
The stark contrast is actually quite an interesting historical demonstration of the psychological principle of anchoring. Primaries were developed during the postbellum party consolidation period: Democrats and Republicans abhorred the spoiler-party-ridden electoral chaos that led up to the Civil War, and lacked the national consensus nor imagination to experiment with making a multiparty system work (they would have had to adopt proportionalism, which was still not an exact science at the time and wouldn’t be for another century), so instead they created primaries and their attending ballot access laws as an institution to prevent disgruntled factions from splitting off the two main parties after losing nominating conventions. Freed from the twin shackles of the Constitution and the risk of spoilers, the states got rather creative, leading to the incoherent diversity that persists today in the primary process.
See footnote 3. Someone always bites my head off and screams about how “bAlLoT aCcEsS iS tHe ReAl ObStAcLe To A tHiRd PaRtY!”, but the history pretty clearly shows that they were a function of reluctance to amend the Constitution, lack of reasonable alternative multiparty schemes, and an antipathy among contemporary Americans towards allowing party establishments the sort of official role that party lists typically played in proportional systems at the time. Americans were, and still are, far more satisfied with ballot access laws that superficially seem not to be biased against any one party, but in practice privilege the two established ones with the same sort of power Americans detest in party lists.
In other words, ballot access laws weren’t the genesis of the two parties (that was Duverger/FPTP), they were a product of those parties trying to solve the spoiler problem without actually fixing the underlying voting math that creates those spoilers (FPTP).
It’s an utter analytical error to view ballot access as the sole obstacle to third parties, and watching nascent third party efforts focus the majority of their energy on loosening those laws is like watching Sideshow Bob stepping on rakes. Ballot access is worthless against the backdrop of spoiler-averse voters.