CEO, Harvard prof: “US politics rigged”

They tell how the “political industrial complex” hurts us; but I tell how we can fight back.

Recently, Katherine Gehl, a former CEO, and Michael Porter, a Harvard Business School professor, wrote a report about “why competition in the politics industry is failing America — a strategy for reinvigorating our democracy”.

Their diagnosis of the problem is based on a robust set of tools originally developed for analyzing business competition; and using those tools, they hit a bullseye. But when it comes to prescribing solutions, they seem unaware of the equally-robust set of tools that come from social choice theory and mechanism design. Thus, while their prescriptions point in generally the right direction, they can be improved upon. And, after I summarize their findings for you, that’s exactly what I’ll do: give a solution that’s better-grounded in both theory and practical considerations.

The problem (according to Gehl and Porter)

Their overall thesis is that US politics is failing voters not because it isn’t working as designed, but because it is. In this case, they’re not talking about how the founders designed it, but how the details of the modern system have been designed by its participants — politicians, lobbyists, and above all, parties. These groups have designed a system that serves them at the expense of voters. In a partisan duopoly, parties have more market power than voters, and they have used that power in a way harmful to voters.

To set the stage, they begin by pointing out all the ways in which U.S. government is working worse than ever. They show hard evidence of dismal public approval of government institutions; growing gridlock and partisanship; and deteriorating public infrastructure, education, and above all political system. These were each once the envy of much of the world, yet now far from it.

The root cause, they argue, is a political duopoly, a cartel that unlike almost any other industry actually sets its own rules. Viewed as an industry, U.S. politics has a scale that’s anywhere from $16 billion per congressional election at the extreme low end (counting just direct spending such as politicians, campaigns, and lobbyists; but surely leaving out significant hidden political spending) to $3.6 trillion per year (the entire government sector). [Note: the report has now been revised; the new figure for FY 2016 is $3.9 trillion total federal government spending.]

For Gehl and Porter, the best way to highlight the uniqueness of this industry is to analyze its competitive structure using the tools developed for other industries. They place the two major parties at the center, in a nominally competitive struggle. Unlike in other industries, they are competing not just for their customers’ money, but for both money and votes; and the value of a vote is not constant, but rather proportional to its chances of swinging an election. In this two-party world, it is often in their interest of both parties to reduce the scope of their competition. For instance, over time, gerrymandering (by either or both parties) has increased the number of “safe districts” where there’s no meaningful interparty competition.

Meanwhile, the factors that in a corporate setting might normally help preserve competition — “new entrants” (third parties) or “substitutes” (voter and ideological organizations without a partisan leaning)—are weak or nonexistent. This is largely due to rules set by the “industry” itself — such as gerrymandering, ballot and debate access rules, and campaign finance rules.

One anticompetitive rule which Gehl and Porter certainly discuss, but which I think deserves even more focus than they give it, is our choose-one plurality voting method. In their terms, this method acts to utterly devalue any ballots that aren’t cast for one of the two frontrunners, as such votes have no effect on the margin of victory. This one rule, far more than any rules about ballot or debate access, reinforces the unshakeable market power of the two major parties. And although choose-one plurality is far from the only possible voting method,¹ it’s the thing that created the two-party system, not vice versa.

That’s not to disparage Gehl and Porter’s broader analysis of how the political duopoly has entrenched itself ever deeper and how its increasing market power harms voters. Though choose-one plurality voting is at the root of the problem, their survey of how the whole tree has grown is still useful. They look beyond the limited power of “customers” (voters; though morally speaking, we’re more the government’s shareholders, I understand them classing us as customers from the point of view of power relations), and consider aspects such as a “channels” by which the parties reach the voters (that is, media, whose incentives are to hype the horse-race while glossing over the issues). They also give a well-argued account of the history of how the duopoly has grown more dysfunctional over time:

Ultimately, they find that aside from parties, power is mostly spread among three groups: partisan primary voters (usually extremists); donors; and special interests (lobbyists, PACs, and the like). No wonder normal voters are frustrated with the outcomes.

The solution (according to me)

Gehl and Porter devote considerable attention to how government is supposed to work and to how we can turn it around and point it in that direction. Government, they say, should deliver four things:

• Action — if government is completely paralyzed by partisan gridlock, none of the below will happen.
• Practical and effective solutions—solutions that are grounded in reality, not ideology.
• Reasonably broad-based buy-in by citizens over time—if solutions work, then people should tend to like them.
• Respect for the Constitution and the rights of all citizens—not tyranny of the majority.

In order to deliver these outcomes and break the entrenched monopoly power of the “political industrial complex”, they point out that solutions must be both powerful and achievable. Mere marginal adjustments won’t deliver the change we need; but pie-in-the-sky proposals won’t either if they have no chance of being realized. Though of course no reform will be easy, we should have realistic plans for making our proposals possible.

Up until this point, I’ve agreed with essentially everything they’ve said, and applauded how they’ve said it. Many of their points are not precisely new, but by using the lens of competitive analysis, they’ve brought them together in a compelling way. But when they come to making actual prescriptions, I have to quibble with some of what they say.

First, a point on tactics. Among the criteria they give for a viable reform is that it should be cross-partisan. Their reason, it seems, is that they think that a reform championed by only one of the two major parties will inevitably be corrupted by self-interest.

While it would certainly be nice if we could find a way to reform this problem with bipartisan support, it will be hard enough to find a reform that’s fair and effective, without also requiring that it maintains the current balance of power. If you could find a reform that would be win/win for ordinary voters of both parties and for incumbents of one of the two major parties, that’s pretty good. Such a reform could conceivably pass as soon as the party that it helped got power. Gehl and Porter’s (apparent) requirement that any good reform proposal should be able to get votes from politicians in both parties is unreasonably strict.

Second, about specific proposals. Gehl and Porter give many. Some of these are unrelated to choose-one plurality voting. For instance, they suggest allowing more candidates into presidential debates; creating a commission to change Congress’s rules of order to eliminate the hyper-partisan “Hastert rule”; making campaign funding more citizen-centric through vouchers or donation matching; and ensuring campaign funding transparency. All of these are good ideas, and important to taking back citizens’ control of politics. Even if we deal with the root of the problem, as discussed below, we’ll still need proposals like these to clear the branches.

But, as I argued above, that root of the problem is the voting method; so the proposals that relate to this are key.

Gehl and Porter’s first proposals in this regard are a step up, but are neither the best solutions nor the most politically-achievable ones. They suggest top-4 primaries; IRV (instant runoff voting) election methods; and nonpartisan redistricting. The first two of these would, through painstaking state-by-state reform organizing, create an overall voting method better than plurality; but one which would have lower voter satisfaction efficiency than almost any other proposal. The two reforms are interdependent, so that neither would be effective without the other; and both parties would be highly skeptical of 4-way primaries, while centrist voters would have reason to be so of IRV. And finally, nonpartisan redistricting would not fully solve the problem of wasted votes and disproportionate results in legislative elections.

I have better suggestions in all three regards. But before I share them, I want to make it clear that this should not be about the best being the enemy of the good. If Gehl and Porter can convince me that their proposals are more politically viable than mine, I’d happily embrace theirs. In order for my suggestions to be better, they must be both superior from a theoretical point of view, and more achievable in a practical sense.

So rather than go on at length about the flaws I see in their proposals, I’d rather explain my own history and analytical framework here.

I’ve been involved in activism around voting methods for over 20 years. It’s what brought me to apply to the doctoral program in statistics that I’m now pursuing here at Harvard, to help bring a set of insights of amateur voting theorists into the academic mainstream; it’s why I became a board member of the Center for Election Science (electology.org, a voting reform nonprofit; though in this article, I speak only for myself); and it’s how I was in a position to help the Hugo Awards develop and implement a new voting method to fend off an attempt by a minority of voters (the “rabid puppies”) to take over their nominations process.

In looking at legislative elections, I think that the best analytical framework is that of wasted votes. Technically speaking, legislative elections (apart from the US Senate) are “multi-winner” elections, because generally many seats in a legislature are being filled at once. This is still true even though the only multi-winner method used in most of the US is the most brain-dead one possible: to split the larger election into multiple single-winner elections and use choose-one plurality in each of them. This stupid method is also common in other English-speaking countries such as the UK and Canada, where, for reasons that have never really made sense to me, people call it “first past the post”, FPTP. In every election, it leads to a majority of ballots being wasted: either overvotes above and beyond what it would have taken to elect the winning candidate, or undervotes for some losing candidate. In fact, the art of gerrymandering is the necromancy of wasted votes; “packing and cracking” your opponents’ voters into districts that ensure most of their supporters cast over- or under-votes, respectively.

In terms of wasted votes, nonpartisan redistricting would help a little bit around the edges. But for a real solution, you need to look to proportional representation (PR). This is a blanket term for the voting methods, used in most democracies in the world, that ensure that a party that gets a given percentage of the votes will get approximately the same percentage of the seats. In a world of proportional representation, partisan gerrymandering becomes quite simply impossible; it’s much more effective than playing whack-a-mole with the loopholes in nonpartisan redistricting rules.

Is PR politically achievable in the US? I’d argue that the answer is yes. The key is to find a proposal that could unite various voter groups — independent and third-party voters, moderate Republicans, and Democrats—with Democratic politicians. Unfortunately, getting support from Republican incumbents is a lost cause; on average, Republicans are the beneficiaries of gerrymandering, except in a few overwhelmingly-blue states where there are almost no Republican incumbents to speak of. So this is something that, if it passes, will do so by being part of the Democratic platform, ready to pass whenever the national pendulum swings back to Democratic control of Congress.

The exact versions of PR that are used in other countries, though good, are probably ill-suited to getting support from a sufficient majority of Democratic politicians:

• Mixed-member proportional (MMP) methods as used in Germany or New Zealand involve two kinds of representatives — some elected in districts, and others added in at a national level to make the proportions work out. Switching to such a method would involve constitutional issues with equal representation across the states, and many incumbents would worry about losing their seats in the transition.
• Single transferable vote (STV) methods as used in Ireland or (in a few elections) Australia involve asking voters to rank their choices in order of preference. These methods are great for voter choice, but the complex ballots can be a pain, and as with MMP, Democratic incumbents would worry about losing their seats in a chaotic transition.

So we need a proposed PR method that keeps the advantages of FPTP — simple ballots; local representation; clear chain of accountability and ability for voters to “throw out” politicians who become unpopular, even if they’re entrenched party insiders; and incentives for parties to have comprehensive platforms and “big tent” appeal to various groups, rather than being single-issue splinter groups. Ideally, it should also give all the advantages of PR—fewer wasted votes; no gerrymandering; higher turnout; improved representation for minorities, be they ideological or ethnic; and less incentives for two-way zero-sum mudslinging campaigns.

As you might have guessed, I have such a proposal: PLACE voting. It stands for Proportional, Locally-Accountable, Candidate Endorsement voting. It works by keeping single-member districts, but letting voters write in candidates from other districts; then, counting the ballots as “optionally delegated”, that is, for those voters who don’t opt out, transforming their ballots into STV-like preference orders by combining the pre-election endorsements and party affiliation of their chosen candidate with the vote totals. Then you find the winners using STV, a well-understood voting method that involves transferring voting power away from losing candidates (and across districts) in order to ensure a proportional outcome where almost no votes are wasted. (If you want to know more about how this voting method works, follow the link above).

PLACE voting is a new proposal, so right now it probably seems a lot less politically realistic than better-known ideas like MMP or STV. But in the long run, it has key advantages over those. Not only does it do a better job preserving the FPTP advantages I listed above, it uses ballots that are very similar to FPTP—ballots where most voters will simply pick a local candidate. Because of that, it’s easy to simulate how PLACE would have worked in past elections. Democratic politicians could see that, while it would reduce their duopolistic power and make them more accountable to voters, in the short term the only incumbents it would directly threaten would be those who profit from gerrymandering — that is, mostly Republicans.

So PLACE voting — or, if you could find a way to make them politically viable, MMP or STV or other PR methods—would solve multi-winner elections and eliminate gerrymandering. But what about single-winner elections like mayor, governor, senator, or president? In those, as long as voters are ideologically divided, it’s impossible to get wasted votes much below 50%. So we need a different frame of analysis.

Such as Voter Satisfaction Efficiency (VSE). This is a way to measure the relative quality of single-winner voting methods by simulating millions of elections, using virtual candidates and electorates with realistic preference patterns. Because the voters exist only inside the computer, you can know their exact “utility” for each candidate — how satisfied they would be if that candidate won. So you can find the average total voter satisfaction of each voting method. To turn that into a VSE number, you scale it so that a voting method which magically read voters minds and picked the highest-total-utility candidate would get 100% VSE, and a voting method that just randomly chose a candidate regardless of voters’ preferences would get 0% VSE. (That means that it’s theoretically possible for a method to average to a negative VSE if it does worse than a random pick; but even choose-one plurality voting isn’t that bad.)

In order to measure VSE, you need to make various assumptions about preference patterns, what voters know about each each others’ preferences (through media or other means), and voting strategy. You can also do experiments and/or look at past election data to check how reasonable those assumptions are. I’ve done that work and I’m working on publishing a peer-reviewed paper on it. In the meantime, here are some of my results. (That link doesn’t discuss my assumption-checking, but that doesn’t mean I haven’t done it; I’ve run hundreds of elections on Mechanical Turk, with groups of 9 human voters at a time who earn money based on which virtual candidate wins).

The most important finding of my VSE work is the obvious: as voting experts already knew, plurality is just about the worst voting method, with a VSE that (depending on strategy assumptions) ranges from about 70% to 85%. The second finding is that IRV, the method proposed by Gehl and Porter, is only a bit better at around 80% to 90%. One method that beats that is approval voting, a method so simple you can sum it up in 5 words (“vote for one or more”), which gets 85% to 95%. And the best voting methods, such as 3–2–1 voting and Star voting, can consistently beat 90% VSE, and in favorable conditions border on 100%. (Note that impossibility results such as Arrow’s theorem and the Gibbard-Satterthwaite theorem strongly suggest that 100% VSE is impossible in practice.)

The fact that IRV doesn’t really solve the problems Gehl and Porter are talking about isn’t just a matter of VSE theory. It’s pretty clear in practice, too. For instance, IRV elections in Australia consistently lead to two-party dominance, so relying on IRV to break duopoly market power seems over-optimistic.

And in fact, IRV’s problems are highlighted by Gehl and Porter’s final set of proposals, which involve an across-the-board push for more centrist candidates. Yes, it’s true that our current voting method is unfair to centrists, but adding centrist candidates without a well-designed voting method would only lead winners to be more ideologically radical. In plurality voting, it’s obvious how that works: by drawing more votes from the most similar two-party candidate, centrists would actually help elect the more-radical alternative. IRV improves on that, but centrists would still tend to be prematurely eliminated, so the second-most-radical candidate would win. It would take a method like approval, 3–2–1, or Star (or any of several other proposals) to actually give centrists a real fair chance of winning.

Are approval, 3–2–1, or Star politically achievable? Unlike PLACE, they’re unlikely to be embraced by a specific major party; but as Gehl and Porter point out, bipartisan reforms can happen if they build enough grassroots energy. At electology.org, we’re trying to build that energy for approval voting, starting in Fargo, North Dakota.

So, to Katherine Gehl and Michael Porter: if you’re reading this, you know I agree with your analysis and your goals, but not with your prescription. I’d love to advise you or help you as you write a follow-up or addendum that includes some of the voting theory I’ve discussed here. And even if you disagree with my judgement of what’s politically viable, this kind of analysis can sharpen your suggestions and strengthen your case.

…

Footnotes:

¹Even as the founders were drafting the Constitution, Ben Franklin’s buddy in France, the Marquis de Condorcet, was publishing his work on voting theory, in which he invented the voting method that bears his name.