Ten Times

There's a specific kind of demo that doesn't ask you to imagine the future — it just shows it to you, slightly blurry, like something seen through glass that hasn't been cleaned yet.

I watched one today.

I'm not going to explain exactly what it is because I'm still processing it, and also because the what is almost beside the point. What matters is the feeling of watching it and doing silent arithmetic — of running through every prediction you've made about how and when 3D worlds become the default substrate for everything, and then multiplying each of those numbers by ten, and thinking: no, actually, maybe still too conservative.

The question that lodged in my brain while watching wasn't "is this impressive." It was: how much more would you actually need to close the gap to our world. As in the world outside, the one with light bouncing correctly off wet pavement and faces that read as faces rather than interpolations of faces. And the honest answer, watching this, is: less than I thought. A lot less than I thought last week.

---

The open source part is what converts this from "impressive research artifact" into "thing that reorganizes industries."

Closed, this is a demo. A blog post. A paper with a link to cherry-picked videos. Researchers cite it, a few companies replicate it quietly, the capability diffuses slowly through the usual channels over eighteen months.

Open source, this is infrastructure. Someone in Taipei remixes it by March. Someone in Berlin ships a product by May. A thousand people you've never heard of find the failure modes before the original authors even thought to look. The capability doesn't diffuse — it explodes, simultaneously, everywhere, and you can't put the number on what that's worth because the number is whatever the entire industry built on top of it turns out to be worth, which is not a small number.

The thing worth billions isn't the model. The model is the thing that makes someone else's thing worth billions. Open sourcing it is the act of distributing that leverage to everyone at once.

---

I've been wrong before about timelines — wrong in the direction of too slow, which is the embarrassing kind of wrong, the kind where you built a mental model of how these things compound and the model turned out to be the one part of the system that wasn't scaling.

So I'm updating. Publicly, while I still feel the weight of the update.

3D worlds — not as a gaming feature, not as a VR pitch, but as the invisible layer underneath everything from product design to film to robotics training to medicine — are arriving faster than I said they were, and whatever I said last, multiply it by ten, and then consider that I'm probably still being conservative because the part I can't model is what happens when several thousand people I've never heard of get their hands on this and start doing things the original authors didn't anticipate.

That's usually where the real number lives.