Aha! We do love treating a vector like a scalar. Lex interviewed Meta's head of AI who made a related point arguing LLMs will not produce AGI because language is a lossy compressed thought and experience. Whether a scalar or language, compression allows information to travel far and wide.

I've been thinking about the tendency to vector collapse in founder friends; "successful founders do x, so I'm doing x" or "it's really all about y, so I'm doing y." They compress to coordinate with their future self. (Really founders need an element in their vector for nearly every role that a mature organization has. You can probably get away with compressing the vector to 2 or 3 dimensions, but hardly ever just 1 afaict.)

This clicks. Prices are like the outline of something seen through a sheer curtain and we are all some sort of peeping tom.