2025 Letter

This year, I officially started working. The big idea that we’re working with is an understanding of humans in every possible way. To take positions in markets, you need an understanding of how economies (the coordinated group of productive activities) work, what other people competing with you think, and also an ability to extricate what you are doing as you reason through this, such that you can encode your thought process into a deterministic computer program.

With the advent of AI, it’s become a bit muddled on exactly what human intelligence is doing and whether we are different from others. You can take all sorts of shorthand - that LLMs can copy but not do anything new, that humans have creativity, etc - but the kernel of truth is not the whole cob.

The question of how humans think (in the wider definition encompassing how we collect information, process them to make decisions, and act, this is really “how humans live”) is important for three types of reasons. First, important in of itself. Humans need not know how humans work to live, but aren’t you curious, for no real reason? Second, it helps us practically. Superforecasters are less likely to fall into cognitive traps (the expanse one), and perhaps you can train mediocreforecasters to be super too. The baseline assumptions of therapy are that you can develop a working model of your emotional landscape and change it. These are the same assumptions, I see, of parts of a modern, Western Buddhism. Third, ???????

What is the nature of human intelligence? How do we think?

First, we are fundamentally animals. We have ancient RL built from the earliest vertebrates,

Intelligence - built from non-linguisic, but language helps us deal with a lot of these things. Buddhism and insight says that there are incommunicable things, and you must experience them to understand. Surely there are better ways than sitting for an hour a day for five years.

But that language is the greatest source of societal organization