@zachweinersmith.bsky.social I used to be skeptical of your UNCLOS-in-space approach from your book but then I see the tech right advocating for space supremacy by installing a constellation of rods from god to create a world safe for *checks article again* ancient warfare x.com/palladiummag...

1/At the beginning of the Pacific War the inexperience US Navy took over an hour to launch their flight deck while the elite Japanese Kido Butai took minutes. By 1944, this was reversed with the USN clearing flight decks in minutes and the IJN sending aircrew not even trusted to land their planes

1/ As a data pt, it was not uncommon in my grandparents' generation (b.1930-40s) to have as many as 7-8 children. In my parent's generation (b.1960s-70s) this reduced to 1-3. Partly because negative personal experience with crowded families and wanting to devote more attention to each child

1/ Germany & Japan saw unjust treaties (Versailles, racial equality clause, naval limits) & economic dependence as existential threats. The Great Depression confirmed their fears-when crisis hit, the West prioritized itself. Expansionists' goal was autarky, seeing global trade as a zero-sum game.

On intelligence: we don't even have a coherent idea what intelligence or consciousness really is, despite trying to measure its effects. Predictions of superintelligence then are based on shaky grounds, the speculations of quasi-religious sci-fi and philosophically dubious statistics.

First we had token predictors that imitated language - immediate thought; recall. Then we had it talk with itself; reasoned thought. I would guess the next step is to imitate an organization - many agents conversing. Diverse LLMs with unique specializations making connections from disparate fields.

Random guess: LLMs are inherently limited by their substrate: human language, which they imitate. They cannot therefore surpass the intelligence already encoded therein unless prompted, and may be limited to the constructs and abstractions already existing in the language its trained on.

When, long in the future, historians write about our era, they will probably say that the Long Twentieth Century (the American Century) beginning 1916 ended some time around 2025. The World-State has now been jerked from its local minimum and until it settles we will be in "interesting times".

So much of modern society is designed around optimization that we don't notice. Get the highest GPA, test scores. The lowest p-value. The best KPI. But in society any measure loses efficacy once it becomes a target. Hence, college admission scandals, reproducibility crises, failure of technocracy.