The fact that everyone here is saying his ideas have continued, but keep naming the same two or three games made by only a couple gaming companies, in a niche genre proves how right he is. If you want any form of deep interactivity play a deep strategy game, anything else besides that in game design is still about as interactive as pacman.
His point is that after 35 years and literal orders of magnitude improvement in processing power, the average game is still modeled using incredibly simplistic logic and makes for empty interactions. And take a look at most any top selling game and it's true.
Your average squirrel in the park has more interesting behavior or interaction than most game characters. It's essentially just been "better graphics, bigger explosions" for three decades now. None of that precludes dynamic interactions.
He's essentially arguing for Probabilistic Programming without using the term. And saying it should be at the heart of all games, no just deep strategy games. Actually a step further he says it should be at the heart of any interactive software
The second link below gives a pretty good overview of his thoughts on interactivity and how to model human and system behavior. Lesson 7 (how to express ideas mathily) while simple, was mildly interesting. Some of the others are better, but are harder to read standalone, build on prior items.
Essentially it's a summarized course of how to model the essentials of human behavior in a system. Not ground breaking, but at the same time, way more advanced than anything you see in the vast majority of games.
If I were writing a simple game, I'd skim link 2 for ideas I could steal and easily apply to up he interest factor. You definitely could build some interesting mechanics with these ideas.
[1] http://www.erasmatazz.com/library/game-design/crawfords-laws...
[2] http://www.erasmatazz.com/library/course-description-2018/in...
[3] http://www.erasmatazz.com/library/course-description-2018/ho...