I'm at day 449 and Casey just wrote a layout system to create orphanages. It worked the first time, as expected by his little diagram. Until now, he coded countless complex systems (loading assets from a file, general memory allocator, a crazy debug system, immediate mode UI, depth peeling, global illumination, the fountain of heads, the software renderer, audio system, multithreading, SIMD, ...) and all of this, _off the top of his head_ while streaming. But how is that possible? Is this what 30+ years of experience look like or Casey is really special? I only have a few years of programming experience and I already suffered from "forgetting" things and not being able to code them off the top of my head anymore, while Casey seem to remember everything.

It's what decades of game engine programming experience looks like. And he's not doing everything off the top of his head, he consults API documentation on stream when he needs to, and I'm sure he does prep work for some of the episodes.

I'd add that some of (most of) those systems might seems complex at first but once you've work on them a few times, you have a better idea of what needs to happen to get those system running and what "tools" are available. So the next time is generally easier.