Why this series exists

The intelligence curse is the argument that once intelligence can be bought rather than raised, the people who hold power stop needing populations. Historically, states and firms invested in people because they had to: you could not run an army, a factory or a bureaucracy without them. Remove that dependency and the incentive to educate, employ and listen to ordinary people goes with it.

I find the diagnosis persuasive and the response underdeveloped. Most writing on it stops at the warning. This series starts after the warning.

What is here

The problems take the curse apart. Treating it as one phenomenon hides the fact that it is several mechanisms pointing the same way, and they do not share a cure. Each gets its own page.

The solutions are a scoreboard of proposed fixes. Every one is scored on the same six criteria, so they can be compared rather than merely admired, and every one can be voted on by readers. I expect to be wrong about several and I would rather find out in public.

Who writes it

Charlie Wyatt. I work on machine learning research and I am finishing a thesis on learned compression in language models. Nothing here is a prediction market, a policy recommendation from an institution, or advice. It is an argument, scored, with its workings shown.

Corrections

If something here is wrong, the comments under the relevant page are the fastest route to fixing it. Substantive corrections get folded into the text and credited.