I could be wrong but I think he's saying the opposite: people are missing that his code is about processes happening over time, whereas the focus of most games nowadays is on the objects. Or, what you can see in his static code isn't important, it's what happens during the running of the code that is.
Now I know almost nothing about game development so have no dataset myself or ability to understand the arguments either way.
I think we're seeing an increasing short term move to more complex, black box, DL trained models that will get behavior similar to the processes he seems to emphasize, and then we'll see a swing back to taking those gains but converting them into the type of readable algorithms like in his old games.
But I have no idea. Just complete 2 second conjecturing on my part.
Regardless, thoroughly enjoying reading his site and find his Object <=> Process gradient cycle to be crystal clear and fascinating and important idea.
Games like EU4 are more process oriented than object oriented, actually. There is no black-box DL model, just a great many low-medium complexity processes that chain into a large story.
These processes can almost always be reverse engineered, and they are actually quite understandable and readable.
Unless there is something I'm missing with his process/object distinction, admittedly I don't think I've parsed it with a great deal of clarity.
Now I know almost nothing about game development so have no dataset myself or ability to understand the arguments either way.
I think we're seeing an increasing short term move to more complex, black box, DL trained models that will get behavior similar to the processes he seems to emphasize, and then we'll see a swing back to taking those gains but converting them into the type of readable algorithms like in his old games.
But I have no idea. Just complete 2 second conjecturing on my part.
Regardless, thoroughly enjoying reading his site and find his Object <=> Process gradient cycle to be crystal clear and fascinating and important idea.