Things I don't know

Not a complete list, of course

Almost every article makes bold, confident statements and tries to persuade you that something is true. For a change, I’ll do the opposite, and write down some questions I don’t know the answer to.

I think it’s a good, interesting idea to map out the contours of one’s uncertainty. Admitting to ignorance forces one to be epistemologically modest, a valuable quality which is in short supply on the internet. It also clarifies what information one ought to be seeking out.

Prediction markets do something similar. It’s easy to sound certain about a topic, but putting money on the line encourages you to think hard about how much you actually know, and whether others might know more. This is another example of how markets have a civilizing effect. (Though actually, I’m amazed by how many MAGA types make confident, large and fairly silly bets. Think Keir Starmer has a 10% chance of losing office before July? Polymarket did last month.)

But many political prediction markets focus on trivial short-term outcomes, like how many times Donald Trump will use a given word in a speech. These outcomes have a definite answer, so you can bet on them; the most interesting questions aren’t like that. There’s no clearcut outcome which decides “will America or China dominate the twenty-first century”, but it is hard to think of a more important issue.

So my first criterion for questions is that they should be interesting and deep. On the other hand, they should be reasonably straightforward, preferably with a yes/no answer, even if they admit degrees of truth. “What is the meaning of democracy?” is too vague. I’m looking for something that could potentially be a research hypothesis.

That also rules out things that are intrinsically unpredictable. Will skirts be short or long in five seasons’ time? Who knows? It’s up to the whim of fashion. Or as a nerd would say, there are multiple equilibria: nothing about the world determines the answer.

Okay:

Is liberalism still necessary for economic growth?

Eighteenth-century Britain was politically liberal, encouraging free speech, individual freedom, and the rule of law. It was also the first country to take off into modern economic growth. That is probably not a coincidence: economic freedom encouraged prosperity, while tightly-constrained political rulers were able to borrow more cheaply. Since then other countries have sometimes tried autocracy as a short-cut to economic development. Results have been mixed, but at the moment, autocrats are on a high and liberal democracies look sclerotic.

One way to think about this: maybe you can separate economic liberalism from political liberalism. Economic liberalism means free markets, contract law, and competition between firms. Political liberalism means freedom of speech and thought: dissidents don’t get kidnapped, you can criticize the government, and so on. China seems to have the first but not the second. Is that stable? Or in the end, can you only preserve liberalism if it also extends into politics?

There are plenty of practical examples today where this question is playing out. Chinese markets are on a roll, but Hong Kong politicians have pulled the firm Hutchison in for questioning about its sale of international ports to the US Blackrock. In Turkey, Erdogan’s main rival was arrested on corruption charges; stock markets dropped like a rock, but they may recover. In the US, as Matt Yglesias puts it:

… business community elites don’t react to this kind of information [Trump’s threats to democracy, including pardoning the January 6th rioters] with “yikes, Trump is scary, we need to check him to preserve American liberty.” They tend to react with “yikes, Trump is scary, we need to bandwagon with him.”

Or as an anonymous Wallstreetbets commenter put it, “the US economy is now pay-to-play”.

So, can autocracy be growth-friendly in the long run? There are good theoretical arguments against it, most famously by Daron Acemoglu and James Robinson. Despite the numerous times I’ve written on the topic, I am not sure.

Will the new dirigisme succeed ?

Relatedly, within democracies there’s a new willingness to use industrial policy, and for the state to think big. This is one lens to see Trumpism through:

But there’s a similar agenda across the political spectrum — think of the CHIPS act passed by Biden, or Germany’s new infrastructure fund.

In the 1980s and 1990s, and especially in Britain, it became conventional wisdom that state intervention and dirigisme did not work. We certainly had plenty of examples of failure. Now many people are equally convinced that industrial policy is a necessity. But diehard neoliberals are not. Maybe underlying conditions have changed since the neoliberal era? I don’t know! A more precise question could be where and when dirigisme will succeed.

Low birth rates: permanent or transitory?

Birth rates have declined sharply almost everywhere in the world and are now below replacement level in surprising places. They are scarily low in East Asia, and have been below replacement for fifty years in the UK. This has generated a pro-natalist movement in response. I am a strong supporter: since most people want to have more children than they end up having, pro-natalism is both pro-life and pro-choice. Anti-natalists often seems to be rooted in weird, nihilistic pessimism of the “too many humans” kind.

Yet, birth rates do go up and down over generations, and it’s not obvious that there are deep reasons why they are low now. One theory is literally that campaigners convinced health elites that overpopulation was a looming disaster; these elites then launched top-down normative crusades to stop people having children, and the crusades worked. If so, perhaps that was a bad idea, but it doesn’t sound deep or unchangeable. A few campaigners changed the world for one generation! Maybe pro-natalists will change it back.

Or, maybe there are deep reasons. Another theory is called the “second demographic transition” — confusingly, because the second demographic transition both names a phenomenon, of societies dropping to below-replacement fertility, and provides a possible explanation for it. The possible explanation is basically “consumerism”! Modern humans are too busy having fun and consuming material goods to have children. Abolishing consumer capitalism seems hard, so if this is the cause of lower fertility, perhaps it is here to stay.

(One thing I’d love to know is whether smartphones have reduced fertility. There seem to be suggestive recent dips in fertility in many countries, and smartphones could plausibly have made people less likely to go out, meet each other and form families.)

Low birth rates have already been around too long to be truly just a transitory blip. But the question is whether they will stay long enough to cause many societies serious problems, or even threaten their viability, in the way that, for example, South Korea is threatened now. Obviously, it matters a lot whether that is true.

Can multicultural societies succeed?

Specifically, can the multicultural societies of the West succeed? There is a ton of high-quality academic work on ethnic heterogeneity, and on migration. Unfortunately, you can’t aggregate all this work and its insights into an overall prediction. Story 1 is that multicultural societies are simply more evolved than their rivals; they are more welcoming and friendly, they will continue to win the global competition to attract global human capital, their values are morally superior to narrow ethnic nationalism and will win out. Story 2 is that, as John Stuart Mill said, “among a people without fellow-feeling… the united public opinion necessary to the working of representative government cannot exist.” In multi-ethnic societies, democracy will be weakened, the public treasury will be plundered by competing groups, and the cultural bases of Western economic growth will disappear. I don’t think we know which is true. I am an immigration sceptic, but on the precautionary principle, not because I have a crystal ball.

Is meat unethical? How bad is it?

There are plenty of passionate well-argued cases against meat. I haven’t read as many defences. Is the debate one-sided simply because one side is in the right? Or have farmers and meat-eaters just not yet engaged?

Obviously this raises complicated ethical questions as well as factual ones. Here’s why it matters: a lot of us believe in progress, we think welfare has increased over time. But if you include non-human welfare, and factory farming is as bad as its opponents make out, then this may be the worst time ever to be alive (on average over all living beings!) and we may be the wickedest generation in human history.

Will we reach AGI? Will large language models do it? Will creating AGI require huge computing power?

From the torrent of writing on the subject, one thing comes out clearly: nobody really knows how large language models work. We are slowly finding out. The truth lies somewhere between Skynet and “a blurry JPEG of the web”, so there’s plenty of room for uncertainty. Meanwhile, the field itself is progressing so fast that Karl Popper’s famous argument about historicism applies: the only way to predict the innovations of the future is to invent them yourself. Trying to write down new versions of Moore’s law is just drawing a line based on a few years — or months — of data. We don’t know when diminishing returns will set in, or conversely when new techniques will explode.

If pushed, I would guess we will reach something like AGI, or at least a set of very useful and powerful capabilities, in ten years. I’d also be interested to know if getting there will require huge amounts of computing resources. Real humans get to human-level intelligence without having to read all of Wikipedia, so maybe future computers will be able to do the same… but then, maybe the correct analogy for training a model isn’t “teaching a human”, maybe it’s “evolving the human brain from scratch”?

Anyway, this question is important not just because it tells you how much to invest in Nvidia. If AGI requires vast investments, we may head for a future where the wealthy have a complete monopoly over frontier-level intelligence. If not, the opposite could happen: our world could become much more equal, because every individual will have the same effective, computer-aided IQ.

Will Bitcoin still be valuable in twenty years’ time?

Everyone’s got an opinion about this, haven’t they? And usually this opinion seems to make them angry.

I can’t see an obvious reason why people shouldn’t use Bitcoin as a store of value. For example, I’m not convinced by Nassim Taleb’s argument from theoretical finance that chance occurrences will one day inevitably make it worthless:

… if we expect that at any point in the future the value will be zero when miners are extinct, the technology becomes obsolete, or future generations get into other such “assets” and bitcoin loses its appeal for them, then the value must be zero now.

This seems too neat and mathsy. On the other hand, there’s no obvious reason that people should use Bitcoin, and not, say, Solana or some other close technological substitute. Perhaps this is perilously close to a “multiple equilibria” question. Okay, are there deep technological and/or social reasons pushing people to adopt some decentralized blockchain-based currency? Maybe there’s demand from all those countries where the government is not trusted not to debauch the currency? Maybe!

---

If you think this is interesting, why not become a paid subscriber? About half my posts are paid or partly paid. A subscription costs just £3.50/month, and yearly subscribers get a great big 40% discount, plus a free copy of my book.