He's being dead serious. He says papers on this are coming up. I don't think he means that this is a "mess", as you put it, but rather, a ripe opportunity to make papers more readable and their CSM code runnable in various ways. That last part is very important. A great deal of material is locked up in these non-machine-readable algorithm descriptions that could be made useful for things like: code generation, actual implementation, correctness proofs, and so on. This is fantastic. It's exciting. I'm jumping up and down with joy at the thought. I'm thanking my lucky stars I watched this video today. I can't overemphasize nor overestimate just what a wonderful innovation this is. Just wow!

Edit: have I mentioned that I'm wowed by this?!