Accountability vs. Consensus

The biggest tradeoff

This is part of my constitutional design series.

There are two different models of how democracy works that are in constant conflict.

The first is the consensus model, where all government action is based on public approval. You want to make sure policy is good - or rather, as the best approximation we have, popular - before it is implemented. More radical shifts should require greater consensus. In this model, people’s preferences are fairly constant, so they don’t change their mind much after seeing policy take effect.

The second is the accountability model, where people vote for a party (or person) who gets all the power, and then vote them out if they don’t do a good job. In this model, voters can’t accurately form an opinion about policy until it is adopted, so it’s all about trying things and seeing if they work.

I’d like to further divide consensus democracy. In cases where there is no overwhelming consensus on an issue, there are two options. One option is to take the median position on this issue - eg, for abortion, legal abortion for a certain period at the beginning of the term, with exceptions for late-term abortion in the cases of rape, incest or life of the mother. Call this median consensus democracy. The other option is to go with the status quo - eg, pre-2022, this was Roe v. Wade. Some of this is the left-winger in me talking, but I usually prefer median consensus democracy to status quo consensus democracy. In issues of an emerging technology, status quo consensus democracy means no regulation. In issues of changing social norms, status quo means the old social norms, which are ~always worse. Status quo consensus democracy also leads to arcane bureaucracy that everyone hates and inadequate law that must be interpreted by courts. I just don’t see the status quo meaningfully as the “neutral” option.

Assumptions

The consensus model of democracy is dependent on people voting based on their political views and this notion of “collective intelligence”. How exactly collective intelligence works is somewhat of a mystery, but I’ll give the version of it that I buy the most. Broadly speaking, people value many different things - liberty, equality, justice, security, faith, etc. Whether voter A agrees with policy choice X is a function of the values that this issue touches on (abortion might be faith vs. liberty, for example) and how much the voter values these different issues (a religious voter prioritizes faith over liberty and thus supports abortion restrictions; a secular voter the opposite). A more complex version of this model could include the role of elites and the media.

The accountability model, on the other hand, works best when people are voting based on their own personal interest. It also works pretty well for issues where everyone has common values and there are clear statistics on which to judge the government by, like growing the economy or preventing crime. Even still it is nowhere near perfect - people hated the economy for much of 2023 and 24 despite all statistics being really positive, and right-wing voters incorrectly think NYC is a crime-ridden hellhole. Voters also need to compare how things are going to their expectations, which can introduce a lot of personal bias that undermines objective assessment.

Debatable metrics like how well a government protects personal liberties are generally not the strong suit of accountability democracy - people are going to approve of a party in power matches their own political views and disapprove otherwise, in that case you might as well just opt for consensus democracy so we can more fairly decide what the best policy is. Even so, accountability democracy can still be useful because when a party is in power, voters can more clearly assess their political views. Otherwise, people have to rely on campaign rhetoric, which can be difficult since politicians love to lie.

I think the general rule of thumb is that the more things are polarized, the more consensus democracy becomes compelling over accountability democracy. High polarization mean that people aren’t able to objectively assess conditions very well, and mostly just vote based on their political views. Given high polarization on the, I think it’s really important to have these political views represented in government accurately. When polarization is low however, I think it’s more ok to give those in power more leeway to do things so that the public can judge them as a result.

It is for this reason that I think consensus democracy generally works better on the national level, while accountability democracy generally works better on the local level. I can see merit in both visions however, and throughout this series my primary objective will be to discuss how this tradeoff manifests in design choices rather than to state which design choices are necessarily optimal. Speaking of which…

How this affects design choices

Every single constitutional mechanism that restrains the party or person currently in power pushes democracy away from the accountability model and toward the consensus model. These restraints make unpopular actions less likely at the cost of limiting the ability of those in power to experiment and limiting voters’ ability to pinpoint who is responsible for outcomes.

Decisions over constraints are made even harder by the many other concerns hidden in the choice between accountability and consensus. Constraints lead to more stability in policy, which might be good (it allows people to work around policy) or might be bad (it allows people to discover and abuse loopholes; also less data informing policymakers on what good policy looks like). Constraints limit the ability of the people in power to destroy democracy. Constraints give the opposition party the incentive to intentionally sabotage the country so that people vote out the party in power. Lastly, there is a very blunt way of looking at this - if you think the bad party does more bad than the good party does good (relative to whatever the middle point is) you should lean toward consensus (because neither side will ever truly be in charge), while if you think the opposite you should lean toward accountability (because power will alternate between the two sides).

The exact constraints affect all of these concerns, and I think the details can be quite important. I hope throughout this series to highlight cases where accountability and consensus would seem to work particularly well. Join me!