In the first post in this series, I talked about versions of tyranny of the majority to do with dealing with the minority. Here, I’ll talk about more mechanistic ways of limiting the majority’s power.
A Higher Threshold
The other version of tyranny of the majority I can think of is dependent the idea that there are majorities we don’t think of as “solid” enough to be making important societal decisions. I think there are two coherent variants of non-solid majority.
The first variant is what I will call transient majority, contrasted with enduring majority. The image here is that you have an initially popular movement or figure that uses newfound power to radically alter things in bad ways, and by the time people realize the policy is bad there is already immense damage. The intuition is that ideas without enduring popularity are less likely to be good.
The second variant is what I will call narrow majority, contrasted with broad or super-majority. The image here is that you have a very divided country like the modern-day US, and you don’t want whichever party that gets 50.01% of the vote to be able to make whatever partisan changes they want. Ideas without broad popularity are less likely to be good, and in fact the downsides are disproportionate, so the benefit of letting the good guys govern unrestrictedly when they’re in power is outweighed by the harm of letting the bad guys do the same.
These fears are in many cases one and the same. Popular enough ideas have both supermajority and enduring majority approval. Ideas that gradually become more popular (eg gay marriage) will get supermajority approval when they’ve had majority approval long enough. Partisan ideas oscillate from slightly below to slightly above the 50% mark forever, giving them neither enduring nor supermajority approval.
The constitutional solution for these fears is to have supermajority or enduring majority requirements for stuff. An explicit supermajority requirement would be a threshold greater than 50% in the legislature to pass a law. An explicit enduring majority requirement would be a requirement that x consecutive legislatures (ie with elections in between) pass a law before it take effect. One way to think about these requirements is that in lack of broad or enduring majority for any one view, you default to the status quo, which is either no law or whatever law was passed before this issue became controversial. Note that this biases legislation toward libertarianism and toward older laws; whether you think this is good or bad depends on your political ideology.
I would argue that most existing constitutional mechanisms are actually just complicated forms of supermajority or enduring majority requirements. The Senate filibuster is a clear supermajority requirement. The Supreme Court is an enduring majority requirement because it takes a long time to replace older justices with ones that reflect more current ideology, and before that the court will limit your power. Midterms act as enduring majority requirements, as the amount of legislation you are able to pass with just 2 years of power is limited. The constitutional amendment process has supermajority requirements. You get the point.
Recommendations
I don’t think you should have supermajority or enduring majority requirements on most legislation like the US does with the filibuster1. Few other countries have them and they turn out fine or even better for not having them. The fears of non-solid majority are just overblown, I think for a number of reasons.
One is that on any given issue the median legislator is pretty moderate. Sometimes this is because you have a multi-party system where the median legislator on an issue matches the median view of voters pretty well, and on any given issue the median view of voters is pretty moderate. Other times it’s because of the natural diversity that exists within a political party’s elite. Either way, the median legislator is usually skeptical of radical change.
Another is that there are many quasi-supermajority/enduring majority requirements inherent in democracy even without adding extra. Periodic elections act as durable majority requirements because the amount of legislation you can pass is limited by a bunch of natural factors - the speed at which legislators read legislation, the amount of negotiation required, different bureaucratic traditions that slow things down. Even once a law is passed, there are all sorts of enforcement issues that can slow a law’s actual effect if it’s unpopular enough, and thus act as quasi-supermajority/enduring majority requirement. I think it’s right to be skeptical of these limits as laws of nature you should be relying on in a constitution - although you can probably enforce some of them explicitly - but they seem to be pretty universal.
The third and I think most important reason is the concept of accountability. Politicians shy away from unpopular actions because they don’t want to get voted out next election. Since voters tend to have a status quo bias, this especially disincentivizes radical change. Supermajority and enduring majority requirements on most legislation, however, all but completely break these incentives. If healthcare premiums spike but the prime minister is not allowed to pass healthcare reform due to lack of supermajority support, voters are going to punish them for things they had no control over. The tradeoff between accountability and guardrails is an ever present one in constitutional design, and I think added supermajority/enduring majority mechanisms are some of the most restrictive. Political arguments have become so disassociated from actual policy in the US because of how many guardrails (what Matthew Yglesias calls “veto points”) there are. Only when someone with actual broad authority comes along like the Supreme Court did in Dobbs v. Jackson do the political arguments actually become meaningful again.
Another thing I want to warn against is being too clever and creating a big hierarchy of levels of legislation, with higher level laws having more stringent passage requirements and overriding lower level laws. The typical size and endurance of majorities is not a law of nature - in the US for example, majorities used to be bigger, and there were periods of single-party dominance in the late 1800s/early 1900s (Republican dominance) and mid-to-late 1900s (Democratic dominance) that you don’t see nowadays. You want to try to avoid a situation where changing laws is too hard simply because the typical majority was more solid back then. I also think people just overestimate how sure we can be about a principle - you don’t want to pass legislation that requires 90% approval to change only to realize 200 years later that oops you have to amend it. I’m looking at you, requirement that all 50 states approve any change to Senate apportionment (although you can probably get around that one).
The Constitution
I think you should have one higher level of laws which has supermajority/enduring majority requirements for changes and which overrides all other laws. This set of laws would be the constitution. In fact, I would argue, that’s just what a constitution is - a higher level of laws with stricter requirements for change. The one way in which constitutional provisions differ from laws in substance is that, on account of overriding other law, constitutional provisions often speak directly to what laws can and can’t do (“Congress shall make no law that…”), which is something you wouldn’t see in a law. In every other way, constitutional provisions are just laws.
There are some other differences in what you typically see in a constitution and in law, but they are not consistent enough to be defining. One difference is that a constitution contains more abstract principles while laws are more concrete. While this tends to be true, there are some quite concrete constitutional provisions and some really abstract laws. Another difference is that the constitution describes governmental structure and process while laws describe what these structures and processes are used for. But the entirety of the US police and jail systems, which I would definitely call structure and process, are laid out in law, while constitutional protections like free speech and right to bear arms are neither structure nor process.
I think there should be multiple processes for amending the constitution. There should, at minimum, be one with a supermajority requirement and one with an enduring majority requirement. The supermajority process should be used for bipartisan legislation, especially that which is time-dependent, like if someone is running for president from jail and you want to disallow that before they win the election. I think 2/3 is a good supermajority threshold - it’s high enough that it basically always necessitates support from the left and right, an empirical observation that holds across countries and throughout history, especially with a proportional electoral system. It’s also a very common existing number for constitutional amendment2. I’m less confident about the enduring majority threshold, but I think the approval of 3 consecutive legislatures (ie with elections in between) could be good34. It is hard but not impossible for a party or coalition to do well enough in 3 consecutive elections that they (maybe alongside a few members of the opposition party/coalition) make up a majority. I imagine enduring majority to be the easier route, used for constitutional provisions which the minority party is never going to approve because it limits their power but which is nevertheless necessary.
Supermajority and enduring majority aren’t the only forms of thresholds more strict than simple majority, and many existing constitutional amendment processes require different forms. Two other types in use are requiring that constitutional amendments be approved by regional governments (like the U.S. constitution does with its three-quarters-of-states requirement, although that one is definitely too stringent) and requiring that constitutional amendments be approved in referendum (used in e.g. Australia and Spain). These two forms rely on different constitutional design principles that I hope to think more about and cover in future posts, so I won’t comment on them too much here. You can also mix and match different requirements. I think enduring majority and referendum approval might go well together, for example, because they both allow you to override the minority party in Congress in cases where their position is really unpopular (eg Republicans in Congress opposing gerrymandering reform because it would hurt them).
The more I think about supermajority and enduring majority, the more I believe in them as the basis for much of how democracies function, in and outside of government. Consensus mechanisms are the glue that holds society together - they always have been.
Though I think the filibuster in particular may be a net positive anyway because the Senate is such an antimajoritarian institution. Unless, that is, you can kill the filibuster and add new states to make it less antimajoritarian
One other nice feature of 2/3 is that for any majority short of 2/3, it’s possible to define the second-biggest bloc as the “minority” and give them explicit rights, as often done. If the majority party holds more than 2/3 of seats, they can split in two as a hack to control the “minority” bloc
This is a bit weird with midterms in the US. You could either do 3 consecutive congresses or do 3 consecutive presidential terms, honestly I think either could work.
One other note - in a democracy with snap elections I would actually make this a temporal requirement (eg constitutional reforms only take effect in 12 years) rather than a requirement on the number of legislatures. Constitutional amendment in Greece has a ‘number of legislatures’ requirement but the majority party can call elections in quick succession like they did this year, which sort of defeats the point