This is the latest post in my constitutional design series. You don’t have to have read any of the others to read this one, but it is helpful.
Modern democratic countries are representative democracies (“republics”), where people rule through intermediaries who set public policy. I’ll define direct democracy to include any other means besides elections for ordinary people (ie, not interest groups) to have their voice heard in politics. Should we adopt more elements of direct democracy? That’s what this piece is about.
Why Representative Democracy?
Logistics
Most of the justification for representative democracy is merely logistical. Even if you worked out a way to pass laws that doesn’t involve a legislature, you still need someone to execute laws. Unless you dramatically change what we think of as a government, you’re going to need an executive, and the most direct you can do is have its head(s) chosen by sortition. You’re also going to need professionals to staff the judiciary and bureaucracy - you can’t just pick any joe shmo to be a judge or tax collector - which would have to be chosen by election or appointment by the head of the executive. So even in this hypothetical system without a legislature you’re relying on institutions as intermediaries between public will and real policy.
Replacing the legislature is probably doable but still logistically difficult. The natural option is the referendum, but elections are big complicated procedures, and having to organize one every time you want to pass a law is hard. People are also busy and don’t want to sign petitions or go to the polls too often, much less go yell about politics in a town hall as done in other forms of direct democracy. The most direct instance of national democracy is found in Switzerland, and even there referenda are held at most four times a year alongside a legislature that can pass laws.
A sortition-selected legislature is also possible logistically, but still faces a number of logistical challenges. For one thing, the people selected would hate it even more than everyone hates jury duty. For another, you’d have to inform selectees of their responsibilities and the political issues at hand, which would be especially difficult if the responsibilities are vast and the selection term is short. Even in representative democracy, where newly elected leaders are expected to be informed about political issues and the responsibilities of the office, there is usually some transition period for the future-leader to get acquainted with their position and hear about national security issues.
Substance
Setting aside logistical issues, there are a number of substantive reasons for preferring representative democracy.
Legislators are better policymakers
The main one is that politicians are just better at setting policy than ordinary people.
For starters, they’re generally smart, well educated, and often trained in law. All these are helpful in setting policy. It’s particularly helpful for drafting legislation, but it’s also helpful in assessing legislation, proposing amendments, etc.
One thing I think politicians are particularly skilled at is balancing different interests. Ordinary people mostly have their own views and want them to be implemented in government. Politicians, because they have to appeal to a wide range of people to win elections, are better at weighing different interests and coming up with compromise solutions.
Politicians are also notable in having particularly coherent political views. If you run polls, people will say that they want smaller government but then respond that they want more government spending in any area you will ask them about. Obviously, this makes it very difficult to actually set national policy in a sensible way. Politicians have real views and so can actually run a country.
Legislating is a full-time job
Writing law is hard. There are a lot of details to work out, some of which are technical issues where there’s an ~objectively better way to do it and some of which (not necessarily mutually exclusive) represent ideological choices on behalf of the writer. It’s an involved process that involves a lot of negotiation, amendments, interviewing experts, etc. Since people can’t be full time legislators, they won’t be able to do nearly as good a job in crafting good legislation that optimizes for their own views as legislators can. You can say the same thing about bureaucrats making executive law and judges making judicial law.
Accountability Democracy
As I’ll talk about in a bit, direct democracy is a tool of consensus democracy, a way to ensure that public policy reflects the public view on an issue. And as always, this means it hurts accountability. You want politicians to be able to pursue actions that are initially unpopular but which, after enactment, the public likes the result of.
Why Direct Democracy?
So why adopt direct democracy? A number of reasons I can see:
Not all issues decide elections
This is I think the #1 rationale for direct democracy. Elections are a rather blunt instrument for having people’s political opinions accurately represented in government. The people in government are very likely not to share many of the views of the public at large. This is particularly true on low salience issues. Direct democracy is a tool for counteracting this; in that sense it’s part of consensus democracy.
Politicians may not know how the public feels
The only alternative to direct democracy for a politician to know how the public feels about an issue is issue polling. Issue polling has its fair share of problems. It suffers from poll error, which is especially bad for issue polling as there is no ground truth to measure against. It suffers from acquiescence bias, where respondents are more likely to respond “yes” to an issue polling question than they would at the ballot box. And it suffers heavily from framing effects, where how exactly you ask a question has a big effect on the results you get. Politicians might also, for whatever reason, not be paying attention to issue polling.
Direct democracy may be a way to get around this.
Politicians may collude to freeze out the popular view
Politicians might also know how the public feels about an issue, and not care. There could be bipartisan consensus among politicians on a certain unpopular view, leaving voters with no choice but to accept it. This becomes less likely in a multi-party system, and with primaries, but it’s still possible.
This was arguably was the politics of immigration for a long time - the public was staunchly nativist, but immigration rates stayed pretty high because politicians from the left and right both wanted immigration (the left for moral reasons, and the right because of business interests)1.
You might prefer the elite view in general because elites have better political views perhaps because they’re smarter or better informed. I am of the opinion, for instance, that immigration is good, and politicians were right to ignore the public. But it’s always have to find a balance between the elite view and the public view, and direct democracy is a tool to give the public view more power if you prefer that.
Politicians may follow their personal interest
No matter who is elected, politicians are ultimately people who will sometimes follow their own interest over the public’s. The two things I like to point to here are corruption - their desire to enrich themselves - and antidemocratic behavior - their desire to stay in office.
A good example of this is congressional term limits, an example that has been floated by politicians left and right but which will never ever happen because congresspeople want to keep their jobs. I happen to think that congressional term limits are bad, but the public definitely disagrees. Direct democracy can be a way to bypass these personal interests.
(Although funnily enough, there’s an argument that before 2015, politicians actually cared more about democratic norms than voters. Trump came in and violated a ton of democratic norms and voters didn’t really care - it’s possible that voters started caring less about democracy recently, but it’s also very possible that voters never really cared and democracy was a self-imposed constraint by politicians)
Tyrannical Government
Finally, there’s the classic tale of a tyrannical state. So the story goes, the people in government (politicians) are always going to have certain bad inclinations, and you need some mechanism for the public to check them on it. The classic example is that whoever is in government will want to be able to arbitrarily imprison their enemies, and so you have this jury system that protects against this. Other people have made this argument about things like private property infringement, or violating the right to privacy. Direct democracy can be a way to counter these inclinations.
I think this take is a very 18th century one, and I don’t find it particularly coherent - who decides what is tyrannical? why would we expect elected people to behave tyranically? Then again, I also didn’t expect Trump to be randomly deporting innocent people to a Salvadoran prison, so maybe it has more force than I thought.
Elements of Direct Democracy
Yelling at the Town Hall
Ahh, local government.
The fundamental problem is that the people who show up to town hall are not representative of the local community; most people don’t have the desire or ability to show up a local town hall. This is a recurring theme in direct democracy, but maybe nowhere is it more true than in the town hall. In practice, town hall attendants tend to more resistant to new development than the district as a whole, which is currently causing big problems.
So you should only use a town hall when you are pretty confident that the people who show up are reasonably representative, or when you have reason to actually prefer attendants’ views over the district as a whole. A good case might be if there’s a certain project that will really only affect a certain neighborhood (and won’t have externalities or affect future residents who cannot attend the town hall!), and you want input on how to build it rather than if to build it. In these types of cases, town halls may be a particularly efficient way for the politician to hear from people. (Or they may not be - I don’t know how well that works in practice).
Call Your Legislator
Calling, mailing or emailing your legislator is much easier than attending a town hall, so you get a lot less of this effect where people aren’t able to do it. You do however also get this effect where only the most motivated people will actually do it. This might actually be good (maybe more motivated people should have more of a voice) or might be bad (everyone should have equal political power!).
I think it’s overall an okay practice for local politicians to note the input of the constituents calling their office, as long as they keep in mind that these people may not be representative and seek out other perspectives.
Petitions
This has a fairly similar barrier to entry as calling a legislator - everyone has the time to do it, but only the most motivated will likely actually do it.
The main difference is that you can count the number of people who have signed a petition, and so you can create a mechanism around what happens if a petition gets enough signatures. This threshold should be set proportionally to an area’s population. I would also like to see someone explore the idea of a “counterpetition”, a petition you can sign if you really don’t like the primary petition, and then the number you put the threshold in in is (signatures for petition - signatures for counterpetition). The main flaw of petitions is that you have to verify the signatures somehow, which as with mail voting is probably pretty difficult to do (you can just sign on behalf of someone else). Unless you have a really good way of preventing this you probably shouldn’t use petitions for something where you really care about the precise number of people signing it - I think signature requirements when you want a rough indication that something is popular can be ok, but I wouldn’t hold elections by petition.
Petitions have a few common uses. One basic one that’s a no-brainer is to use signature requirements for getting on the ballot to begin with. Another is if a petition for a law gets enough signatures, that petition gets put as a ballot measure in the next election - I’ll talk about that one in a sec.
The mechanism I want to discuss here: if a petition gets a threshold number of signatures, the legislature must vote on it, up-or-down. (Or perhaps with some amendments if you have some neutral body assessing that these amendments aren’t too drastic). What this mechanism targets is not that politicians will do things that voters don’t like, but that politicians will obscure what they’re doing by not even putting things to a vote. This mechanism combined with primaries or a multiparty system then allows people to vote out whoever voted against their petitioned bill. I like this mechanism because it complements representative democracy rather than replaces it, which avoids many of the problems I’ll discuss in a bit.
Referenda
Referendum is the element of direct democracy with the smallest unrepresentativeness. But for referenda held not coinciding with national elections, the effect is still not zero: the people who show up to vote for these are going to be disproportionately engaged. I worry a lot that even symbolic unrepresentative referenda are worse than no referenda because they give the strong illusion of democratic process; you have to be a pretty sophisticated observer to realize off-year referenda are not necessarily representative. When possible, you should aim to have referenda coincide with national elections.
If you do decide to incorporate referenda into a democracy, there are a number of decisions to make.
The first is how they get on the ballot. The options are:
if a petition gets a certain number of signatures, as discussed earlier
like ordinary law, passed by the legislature
through some avenue that’s easier than passing a law:
The leader of a country can put it on the ballot, bypassing the legislature
If the leader has a veto power, it can be the other way around: the legislature can put it on the ballot, bypassing the leader
The second is what the threshold for passage is. The most natural option is majority vote, but you can make it higher if you think that’s too easy.
The third is what happens when the referendum passes. The options are:
Symbolic: nothing happens
As we saw with brexit, symbolic referenda can be quite powerful. I think this is for two reasons:
They inform politicians on what the public is thinking
They raise the salience of an issue. People will debate the issue and then be more likely to vote based on it in the future.
The legislature must vote on it
This is similar to symbolic referenda, except the referendum must be on a specific law (rather than a vague statement), and the legislature must publicly override the public will if that’s what they want to do.
It goes into law, but the legislature can override it by passing a law
If there’s no veto power or other limitation on the laws that the legislature can pass, this functions very similarly to the previous one. If there is, this adds an extra barrier to overriding the referendum
It goes into the constitution, so the legislature needs to meet supermajority or other stricter requirements for overriding it
Non-symbolic referenda have the big problem that the public are not lawyers and do not have the time to read a lot and so have to rely on intentionally biased descriptions of a law. I mean look at this referendum (a “yes” vote removes a provision from the Kansas constitution that ensures the right to an abortion):
The best way to get around this is probably to allow both proponents and opponents to write a description of the referendum, and have a neutral party (eg the courts) approve the description as at least reasonably accurate.
The fourth decision is which laws referenda apply to. The options are:
All laws must be approved by referendum before being enacted
Don’t do this. This suffers from the logistical issues discussed previously and furthermore destroys accountability. Switzerland has a version of this on the national level, but the legislature can arbitrarily override the requirement by declaring the law “urgent” so it has no real force. Some Swiss regional governments (“cantons”) have this without the loophole, and I think it makes more sense there but is still harmful.
Certain types of laws must be approved by referendum before being enacted
No laws have to be approved by petition, but any law may be put to a referendum.
Only certain types of laws may be put to referendum
Let me discuss the choices I think are most promising:
If a petition meets a signature requirement, it is put to a symbolic referendum
This could be useful purely as an informational tool. If there’s an idea that people are really clamoring for, and which they claim is popular, they can petition for a referendum and prove it. As someone who loves data, it’s kind of a policy without downside: it’s like a better issue poll. As a nonbinding referendum it may inherit some of the flaws of issue polling (acquiesence bias, framing effects), but hopefully these are somewhat lower and also there’s no poll error. Also unlike an issue poll, a lot more people will know about and it’s a lot less disputable (you can usually just cite a different poll whose result you like better).
The legislature can put any law to a referendum by majority, overriding the leader’s veto, and if approved it goes into law - up to one per year
I talked in my post on the executive and the leader that I thought the best system is that mostly the leader gets to control the agenda of the legislature, but occasionally the legislature should be able to override the leader. Well, having a veto and then using referenda to get around the veto could be a good mechanism for that. The key point here is that because you need a majority of the legislature to approve the law before referendum, the legislature is not going to turn around and override the results of the referendum.
Usually, particularly in parliamentary democracy, a majority of the legislature will agree with the leader, so this mechanism won’t be necessary. Sometimes, however, a majority in the legislature will take a position against the leader. This happens in one of three cases:
It is a presidential democracy, and the leader does not come from the majority bloc in the legislature. As I talked about in my post on leaders, this is just an overall bad state to be in, and I’ll ignore it in this piece.
The leader comes from the majority bloc in the legislature, but holds an opinion that is very unpopular within this bloc.
A minority of the majority-bloc in the legislature breaks off and supports legislation that the minority-bloc is pushing for and which the leader (who comes from the majority bloc) does not support.
I think in both of the latter two cases, you want some mechanism by which, if a majority of the legislature is really clamoring for something, they should be able to get it. This could be a good system for ensuring that. Because you have to call a referendum on it, and because voters (due to status quo bias) are probably less likely to approve “omnibus” bills, this does a good job of limiting the amount of legislation that can go through this route, while still allowing some popular legislation to make it through.
A constitutional amendment can go into effect if it is approved by a majority of the legislature and then a supermajority of the public
I think it’s important that this is not the only constitutional amendment process. The legislature should, by a 2/3 vote, be able to pass any constitutional amendment it wants. This allows the legislature to move quickly when necessary (eg to prevent a president convicted of crimes from taking office) or for a consensus of politicians to do things even when there is no public consensus.
But this could be a good secondary process, for when the minority party is never going to agree to demands that a consensus of the public supports. I’m thinking specifically about democracy-relevant provision: like, a party is able to overcome gerrymandered districts for the first time in 20 years due to a huge wave, and they then want to put in place an antigerrymandering constitutional amendment so the system can start being fair again. (In fact, I would make this process exlusive to democracy-relevant legislation).
I think 60% is a good supermajority threshold. Unlike for legislature constitutional amendment votes, the target we’re aiming for is something like “nonpartisan support” rather than “bipartisan support”. We want to pass laws that are just so commonsense that even when the minority party campaigns against it they can’t stop it, rather than this image of bipartisan support which is more that partisans from both sides are in favor. Now maybe a supermajority of the public is unlikely to support a system that the minority party in the legislature opposes, but it is possible: for instance, under the 60% threshold, this anti-gerrymandering amendment would have passed.
If a petition meets a signature requirement, it is put to a referendum, and if it passes it it goes into the constitution
Some important details here:
Don’t allow it to pertain to budgetary matters. Referenda are just a really bad tool for dealing with the budget - the public will always vote for lower taxes and higher spending, and the constraints they enact make the job of politicians really difficult
I think there’s a good case for making it a supermajority requirement. A supermajority requirement makes it harder to enact laws, which might be preferable for a mechanism that circumvents the legislature, but then also harder to repeal those same laws. The one thing is if you do make it a simple majority, don’t allow the public to amend/repeal laws that are passed with 2/3 of the legislature - I don’t want this process being used for regular constitutional amendment
This is the most forceful approach to petitions that I think is defensible. I don’t love it. In general, I’m not a fan of constitutional law - it tends to become outdated and in practice courts have a lot of discretion in how they enforce it. Having constitutional law be a routinely used part of the political system sounds like a nightmare.
Sortition
I don’t think it’s a good idea to use sortition as a replacement for some part of representative democracy. Like, I don’t think it’s a good idea to set out a portion of law that a citizen commission is responsible for enacting. We basically know how this would work in the U.S., because many state governments are part-time with short term limits and thus are already essentially run by joe shmos. What we have discovered is that lawmaking is a skill. You need a good understanding of current policy and law to know how to write good legislation. When lawmakers don’t have that, policymaking ends up dominated by lobbyists. While lobbyists aren’t all bad, they do have the fundamental issue of incentive, so they really shouldn’t be the only actors with power.
I do however think sortition can be an attractive option when you have a very specific task you want selectees to accomplish2. Juries are a good example - they have the job of determining if someone is guilty or not, not setting criminal justice law. They’re there to protect against arbitrary imprisonment discussed earlier3.
Overall, while I think some elements of direct democracy are worth adopting, I would tread with caution. Representative democracy is mostly an ok system on its own; the downsides of too much direct democracy are bigger than the downsides of too little.
I’m a little skeptical of this argument just due to the lack of really any meaningfully nativist politicians until 2015 - you’d think that if the public really felt so strongly about it, anti-immigration candidates would have broken through eventually. But it’s definitely an argument you can make
You might think of using sortition as a replacement for elections or referenda to save energy - instead of holding an election, you could pick 1000 random people from the country and ask them the question. Maybe you could run a “trial” where both sides make their case and then the selectees deliberate. Unfortunately, if you do anything other than ask selectees a question, a lot of selectees won’t have the time or ability to do it, so your sample won’t be representative. Even if you just ask selectees a question, the fact that only 1000 people in the country do it means that it won’t succeed at making the election salient - all the dynamics of public opinion fall apart
I’m not actually a fan of the jury system - I think the real thing ensuring that the government doesn’t arbitrarily imprison people is that there is procedure and transparency to ensure that this procedure is actually followed. Still, I think that there is at least a plausible case for juries.