There's still a translation problem, even when all of the underlying access mechanisms are open. And maybe there's the opportunity for a kind of meta schema self-description mechanism that can sort of ease the translation. I think it's going to be very, very, very troublesome in the next few years. We could do distributed computing by just injecting a lot of data into a cosmic canvas and have something run into the edge of all of those images to train a new model in a decentralized way. That was awesome, thank you. Alex, John, anywhere you'd like to go?