Are Republics Still Possible?

John David Pressman

My faith in humanism is being seriously tested right now.

I've been feeling a growing dread these past six months when I think about the future of the Internet, AI, and Western society at large. Part of the dread is brought on by the 2024 election, in which your candidates for president are the person that wants price controls on groceries and the guy that wants universal basic tariffs. Part of it is the way that discourse on SB 1047 has made it clear that AI X-Risk people are ready to retroactively confabulate whatever pattern is necessary for their AI policies to fit. Part of it is the socially repressive policy slate featured in Project 2025 and the implied popularity of that policy slate given the election is still a cointoss after its publication. But the real dread goes deeper than any of that, it's in the generating function I can infer in common between these things. In a sentence it might be summarized as:

If you were to delete the concept of a limited liability corporation from the Western legal tradition I don't feel like it would be reinvented.

This implies a few terrifying thesis about the United States:

  1. The consent of the governed has basically been lost. The whole legal, economic, administrative systems of government in the United States are running on pure momentum as their generating function has been lost. Whenever they degrade no force steps in to repair them, nor do I expect any such force to spontaneously arise as the decay starts taking down core parts of society like restaurants and grocery stores.

  2. The engine of wealth creation in the US is also running on momentum, the social conditions necessary for it to operate are in serious jeopardy.

What's brought this into clarity for me is the extent to which the things I feel dread about are popular policies. If you poll people in the West on AI they hate it and this extends to support for all kinds of measures to suppress it. Price controls on groceries is a popular policy. While the planks in Project 2025 might not poll well, I notice that Trump is still going to be president in November about half the time so they can't be that unpopular in practice. Josh Barro, who I linked on price controls above, summarizes the situation as follows:

The public demands action against “price gouging,” my objections to such laws are a political loser, and I should make peace with the fact that I won’t get my way on this issue. That is how democracy works.

He's entirely correct about this. That is how democracy works, which is why the the framers of the constitution made an explicit decision to avoid a democracy in favor of a republic. They restricted the franchise and implemented a representative congress not just because it was logistically intractable to have citizens vote on every issue, but because democracy is a poor system of government. Benjamin Franklin famously described the system of government laid out by the US constitution as "A republic, if you can keep it.". Unlike a democracy I think that a republic is generally speaking a good form of government. A healthy republic brings out the best traits of a democracy like sensitivity to the needs of the governed and dynamic public discourse while combining them with the best traits of an oligarchy like elite decisionmaking and a strong appetite for growth. To obtain these properties however a republic needs some insulation from public whims and opinion. This was the explicit purpose of the senate, which originally was not a popularly elected office at all. To be blunt the founders of the United States understood, correctly, that representative government could only work if the elite class elected to represent the citizenry had the necessary autonomy to overrule their constituency when their demands are idiotic and implement the machinery of society necessary to bring prosperity even if most of the public doesn't understand it.

Dan Jeffries has criticized the Pause AI coalition a lot for push polling, which is where you poll people with questions that are carefully framed to make one option seem better than another to claim you have popular support. I admittedly haven't read their polls very closely and I'm sure they in fact do a lot of push polling given their rhetoric, but I also doubt the negative AI sentiment is an artifact of push polling. It's not even clear to me that a lot of the 'push polling' is push polling, it's an entirely different kind of bad faith. Take for example the provision from a draft of SB 1047 that a model can't make it 'significantly easier' to cause $500M of damage. We could poll that more or less neutrally with something like:

Do you support liability for AI models that could be used to make it 'significantly easier' to cause $500m in damages through e.g. cyberattacks?

And I imagine this would poll fairly popular, even though in theory this description could apply to an ordinary home computer used to write ransomeware. This wouldn't be push polling, this is the literal text of the bill, I even put 'significantly easier' in scare quotes to make it clear that's what the bill says. I expect this would get 60-80% positive response rate, maybe more. It also sounds great in online discussion, you get to say "ha you think the probability of $500 million dollars in damages from open weights models is so high that they wouldn't be able to be released anymore, proving you know just how dangerous they are" and it's just kind of like...yeah the probability someone will write a ransomware with a strong open weights model rounds to 1, this isn't actually a good reason not to have open weights models.

I'm increasingly doubtful the magic of a republic is sustainable in the presence of the Internet. If you went up to the average person on the street and explained the concept of a limited liability corporation to them, I'm fairly sure their reaction would be something like "wtf how is this real" and an immediate demand they be abolished. I don't think people have become any less educated than they were in say, 1970. I'm pretty sure they didn't stop teaching this stuff in public schools, it was never taught in the first place. The modal person in the United States simply had no meaningful awareness or input into whether limited liability corporations are a good idea. They still don't, but in principle one could start polling:

Should limited liability for corporations exist? Limited liability shields the executives of a corporation and their investors from having their personal assets taken as the result of a lawsuit against the company.

And I'm pretty sure you would get a negative response rate over 50%. If representatives are expected to act on such preferences Western society is doomed. If they're not then it will become increasingly obvious to people in the concrete rather than the abstract which policies the elite class disagrees with them about and their consent to be governed by them will evaporate. Either they'll replace the current representatives with know-nothing crowd pleasers or do away with the government altogether. With a congressional approval rating of just 19.3% one could argue this has already happened and the other shoe just hasn't dropped yet. One way to look at the Internet from a governance perspective is a search process for whatever adversarial policies are both ruinous and unexpectedly popular. You simply didn't have the conditions for perverse policies like "the age of consent should be 25 because that's when your brain stops developing" to be heard by a lot of people until social media brought the marginal cost of publishing close to absolute zero.

If anything like what I've just laid out is true we'll have to change our governments from republics to something less sensitive to public opinion. I'm not sure how that would work, but do notice that for all the griping about how we need new forms of government there is almost no serious discourse about what they might look like. I don't consider the concept of a network state serious, though it and the charter cities movement are the closest we get to anyone even discussing the subject in public.

...It's gonna get worse before it gets better.