Compounding thinking, not data

Watch what's happening with AI right now: an explosion of artifacts. Tokens burned at real economic and environmental cost, decks and documents and code pouring out of every team. And something about it doesn't feel right to me. It doesn't feel like we're compounding thinking. We're compounding data.

Here's the part that bothers me most. When someone's artifact isn't good, it gets thrown away or hidden, and part of their thinking might have been really interesting. The internet's current rules say you need the fully fleshed-out version before you hit publish, or you're some loser. So the thinking stays trapped inside artifacts that never ship.

I'm not claiming a perfect articulation of some world-changing idea. What I'm trying to do is surface the thinking itself: the processes, the threads I'm pulling on, offered as ingredients. Because if we all started exploring each other's half-formed threads, something in mine might connect something in yours. The goal is to help unlock the thinking.

And there's a version of this I can't stop thinking about. If my system holds my texture graph and yours holds yours, we should be able to overlay them and co-think: let the membranes do the quiet work of sharing the shape of thinking without exposing what's private, and watch where the paths cross. Two graphs overlaid open pathways neither of us could see alone.

Underneath that is a bet about which game we're playing. The VC world, and honestly capitalism at large, keeps telling us this is zero-sum: build the empire, guard the moat. I think taking the zero-sum approach toward AI is a very quick way to risk humanity. Collaboration over empires isn't a nicety. It's a call to aggressively think about what we could do with this instead.