A Surface You Bounce Off

●●○○○sketch

At the scale being built right now, the other scholars aren't simulated people. They're reactive functions: things your actions bounce off, that respond in believable ways without ever wanting anything of their own between one interaction and the next. We're not hiding that. The question is how much more of a person they'd need to become before it started to matter, and we don't have a settled answer.

What's actually there today

A named reviewer's positions, a corpus author's biases, a student's one well-timed question: all of these already come from something with a fixed shape, generated once and read back consistently. What isn't there yet is anything like an NPC noticing a pattern across several interactions with you, holding a grudge that develops rather than one that's simply present or absent, or changing their mind about you over time the way the game already expects you to change your mind about the world.

The expansion path, roughly sketched

What's been discussed, loosely, is something closer to simulated agents: scholars with their own shifting positions, actual rivalries that develop rather than static ones, something like the internal politics of a department playing out whether or not you're currently looking at it. None of that is scoped in detail yet. It's a direction, not a design.

What we don't know

The honest sketch is that we don't know where the useful line sits. A scholar who's genuinely simulated is expensive to build and might not even read as more alive to a player than a well-tuned reactive function does; a lot of what makes another person feel real in a system like this is probably closer to good writing than it is to deep simulation. But we also don't want to assume that before actually trying it, because a shallow reactive corpus might quietly cap how far the parts of the game that depend on other people, review, reputation, rivalry, can ever really go.