What are representatives for, anyway?

I’m returning from my self-imposed exile now that I have some time to write over Thanksgiving break. I’m not going to talk about the election or what we can expect from the upcoming administration (I do enough of that over on the Discourse Machine podcast). Instead I’ll keep the promise I made when I started the blog, to use this space for bigger-picture analysis that isn’t necessarily tied to the most recent news cycle. Today I want to make the argument that they way we think about the purpose of representative government, and what we expect from our elected officials, is all muddled up. More than that, it often contains logically inconsistent (and therefore impossible) demands on our politicians. This inconsistency, and our collective failure to recognize it, can explain a fair amount of our collective frustration with electoral politics. This also helps to create real vulnerabilities for democracy.

So to start with I’m going to suggest that you decide — right now, before reading my argument — what you think the point of electing politicians is. You don’t have to have a really firm and unshakeable belief about this, just make a quick judgment about what your priorities and inclinations are. What do you expect these people to do for you when they go to Washington? What do you want from them? Then I want you to ask yourself whether or not you’ve ever made any of the following complaints or statements (or heard other people make them):

• Why is this politician flip-flopping about this issue?

• Politicians are all weathervanes, they just do what will get them elected and they don’t have any principles!

• These Washington elites are so out of touch! They don’t care about my priorities or values and just follow their own agenda!

• I just want my elected officials to do what’s right about this issue that I care about, even if it isn’t popular.

• We can’t sacrifice our ethical principles in electoral politics!

Done? Great. My guess is that you easily recognized the complaints in the second question, and also that you answered the first question more or less along one of two modes of thinking. (If not, come find me on Bluesky and tell me! It’s useful information!) The first mode is something like “I want them to go to Washington and represent the will of the people that put them there.” Let’s call this the responsiveness model of electoral politics. The second mode is something like “I want them to go to Washington and make good policy. I want them to make the country better and to do the right thing.” Let’s call this the conscientiousness model of electoral politics. You might also have answered using a combination of the two modes of thinking. Let’s call this the mixed model of electoral politics. I think that a lot of people end up doing something like the mixed model, and that’s the thing that I think contains logical inconsistencies, but let’s come back to that a bit later. For now I want to explain what each of the first two models looks like in practice.

The responsiveness model says that politicians are like avatars for the popular will. The most important thing is for politicians to stay in touch with the needs, desires, and values of their constituents, and to put them into practice to the best of their ability. Politicians fail, on this model, when they stop being responsive to their voters’ preferences. When that happens, the voters should elect somebody else, to remind the politicians that their job is to represent the voters’ preferences. This model has some costs and some benefits.

• Benefits: 

  • Politicians can have lots of bad incentives. They might simply enjoy having power and access. They might get ‘bought out’ by monied interests and become corrupt. They might adopt unpopular, extreme beliefs and try to force them on others. The responsiveness model can help prevent that by evaluating politicians by a simple, easy-to-understand metric: do their actions, words, and votes align with the majority preferences in their constituency? If so, things are working. If not, things aren’t working and it’s time for change.

  • This is an easy model to understand how to effect change. If politicians respond to majority desires, the way to get your preferred policy enacted is to persuade your fellow citizens and form majority consensus. It’s not the simplest or quickest way, but it should be pretty reliable on the whole.

  • We like feeling like our government is attentive and responsive to us, and will do what the majority of the country wants. Isn’t that what democracy is for? Isn’t anything else undemocratic?

• Costs:

  • This model strongly incentivizes flip-flopping, weathervane behavior. After all, one of the basic ideas of the model is that politicians are supposed to do what the majority wants!

  • Because of this, the model leaves little room for ethics or principles. At best, it suggests that a politician who wants to go against the majority preference for principled reasons has to either persuade their constituents to endorse their position before the next election, or lose the election. Human nature being what it is, few people would be willing to take that risk.

The conscientiousness model says that politicians are people that we bet on to make good policy decisions on our behalf. The most important thing is for a politician to keep the actual good of their district/state/country in mind, and do their best to learn what policies will help achieve good outcomes. Politicians fail, on this model, when they have bad character (they don’t actually care about doing what’s best) or when they have bad ideas (they try to have good policies but they don’t know what good policies are, and so enact harmful policies). When this happens, the voters should elect somebody else, to remind politicians that their job is to have good character and good ideas. This model, too, has some costs and some benefits.

• Benefits:

  • Most of us aren’t experts in policy. Even if we’re pretty knowledgeable about some policy areas (the economy, perhaps) we’re probably not very knowledgeable at all about others (foreign policy, maybe). This is to be expected — we have lots of demands on our time, and can’t spend all day learning about policy. If we select some people to do that, as their job, they have a better chance of becoming more knowledgeable than we would be. So they have a better chance of making good policy than we would.

  • All of us probably have beliefs that we think are right but that we know don’t enjoy majority support. For people on the right (broadly speaking) it might be a pro-life position, or a belief that we need to be more fiscally responsible. For people on the left (broadly speaking) it might be support for Palestine or immigration. Most of us get upset when politicians do things that we think are wrong, just because they’re popular. The responsiveness model tells us that we just have to accept that, at least until we can gain genuine majority support. But the conscientiousness model doesn’t require that.

• Costs:

  • It’s much harder to evaluate the quality of politicians’ character or ideas than it is to evaluate whether they are being responsive to majority preference. So we have a much higher risk of having politicians say they have principled stances that are supported by expertise when, in reality, they are corrupt, evil, or whatever.

  • The conscientiousness model implies that, at least some of the time, the right thing for a politician to say is “I know that many of you want me to do X. However, I am more knowledgeable than most of you, and I know that X isn’t a good idea. So you have to trust that I have good character and good ideas, and let me do the right thing even if you don’t like it very much.” That doesn’t sit well with a lot of people. It’s perfectly consistent with a democratic political structure and institutions, but it goes against some people’s democratic ethos.

So we’ve looked at two ways of thinking about the job of elected officials, and we’ve seen that there are benefits and costs of both models. There isn’t a perfect answer. But now think back to the questions I asked at the start of this piece. If you picked something like one of the two models, you might still like that model best (even given its costs), but you also might be a little less certain. Either way, you’re probably pretty familiar with the list of complaints. And hopefully now you see how the two models can lead to some of those complaints — they’re the ‘costs’ of the models.

In reality, I think a lot of people try to have their cake and eat it too. They mix features of the two models, meaning that what they want ends up being something like this: “Politicians should be generally responsive to the will of the people (as in the responsiveness model) but they shouldn’t follow majority preferences into the worst, most unethical positions (as in the conscientiousness model).” Now, as a statement of individual preference, that probably looks like a pretty reasonable position. And internally it’s perfectly consistent. There’s no principled reason why politicians shouldn’t act as avatars of the popular will about all sorts of decisions, but stand up against really unjust things even if they’re popular. So it seems like a mixed model might work out pretty well.

I don’t think it does, though. In fact, I think it works out worse than either of the other two models. The reason is because it’s not possible to apply it consistently at scale, even if it’s possible to think it, internally, in a logically consistent way. The source of this problem is the way in which the mixed model creates a massively confusing epistemic environment for the politician. This is because each voter’s private understanding of which policies fall into which category in the mixed model can be different. Let’s take a simplified model to see how this happens.

Even in this very simplified example, the mixed model turns out to be a losing situation for almost everyone. By multiplying the criteria, the mixed model makes it almost impossible to make everybody happy. The politician doesn’t have a way to make everyone (or even a majority of people) happy. And the voters don’t have a clear theory of change, either, since gaining majority support won’t reliably lead to getting their desired outcome. Instead of getting the best of both models, the epistemic features of the mixed model end up looking more like the worst of both models. And, I think it’s clear to see that adding variables — in this case more policy issues — dials up the complexity a lot.

I’ve already gone on for too long, so thanks for sticking around if you’re still here. But I definitely won’t test your patience by going into how we might resolve this issue. Instead, I’ll close by pointing out a few bad consequences of the problem with the mixed model: besides the fact that it’s just dumb and doesn’t work very well, I think it creates massive vulnerabilities for democracy. Here’s four.

1. The mixed model plus the speed and scope of political communication makes issue polarization and voter dissatisfaction more likely: more voters care more deeply about more issues (without necessarily having rational or justified beliefs about those issues), which makes the basic model described above much more complex. Fewer voters will feel like democracy is ‘working,’ in part because their expectations are unrealizable.

2. Issue polarization and voter dissatisfaction help give rise to anti-incumbency bias: Because the mixed model creates unattainable (or nearly so) “win conditions” for politicians, it’s extremely likely that, given enough time in office, they’ll become sufficiently unpopular to lose. And if I’m right then this is a necessary consequence of the epistemic structure of the system, not really something they could do a lot about. Politicians and parties might get punished by voters because they’ve actually done something wrong, but they’re also likely to get punished because of the features of the system.

3. Anti-incumbency bias makes party capture much more threatening: if a party is captured by an extreme faction, and if the anti-incumbency bias created by our collective confusion about the purpose of representative democracy makes voters likely to vote that (now extreme) party into power at some point, then that is a plausible path to victory for extreme factions/parties.

4. Party capture makes democratic backsliding more likely: of course this is a pure hypothetical and not at all like anything we are observing in countries on several continents right now. Just a fun little bit of armchair philosophy.