Sora was the rehearsal and Walt Disney played along just fine
Hands with six fingers. People vanishing mid-scene. Disney licensed IP to this. Called it research.
Somewhere in a San Francisco conference room, a group of people decided the world needed a machine that made fake videos. Not convincing fake videos. Fake videos where hands had six fingers, people vanished mid-clip, and a cat walked through a wall. The system understood the physical world the way a bankrupt company understands its balance sheet: selectively, with omissions, and only as long as it serves the presentation.
They called it Sora. They said it would learn to understand the physical world in motion. They meant: we need something that lands well in demos. Fifteen million dollars a day. Not a mistake. A deliberate choice to keep going as long as there was capital to keep going with. You get it.
Disney signed. Of course Disney signed. A company that guards its intellectual property with the precision of a debt collector licensed that same IP to a system that produced usable disinformation eight times out of ten. Not a miscalculation. A trade-off: the potential upside outweighed the reputational risk, until it didnβt. Then Disney stopped. Not before. Right on time for itself.
The animators werenβt at that table. They were informed. That the industry was changing, that adaptation was necessary, that opportunities awaited those who kept up. When the plug was pulled, they were the ones who had to explain what their role still was. The people who signed the deal are now explaining why robotics is the next logical step.
The same architecture that couldnβt grasp that objects donβt pass through walls is now heading into the physical world. Systems that drive, grip, move. The limitations havenβt been solved. Relocated. To a context where the consequences are harder to dismiss than an app.
The mechanism is unchanged. Ship what doesnβt work yet, scale as long as capital follows, rebrand failure as research, leave the bill with whoever wasnβt at the table.