Can Warhammer 40,000 actually be simulated?In this video, I break down one of the hardest problems I’ve been thinking about for a long time: whether it’s possible to simulate 40k to a meaningful standard, and where you would even begin.I cover a bit of my own background in machine learning and competitive 40k, some of the previous work I’ve done with Atlas, and why I think this is such a fascinating challenge. We look at earlier game-solving milestones like chess, Go, and StarCraft II, what made those problems solvable, and why 40k is awkward in a completely different way.In particular, I talk through:- why chess is tidy and discrete, while 40k lives in a messy continuous feature space- why movement, angles, footprints, terrain, and exact positioning make 40k so difficult to model- how randomness and hidden intent explode the complexity even further- why deployment is probably the best place to start if you want to build a useful simulation engineThis isn’t really a video about “AI magic.” It’s more about the real engineering and tactical problems underneath trying to model a game as complex as Warhammer 40k and why even getting part of the game right, like deployment, could already be incredibly useful for helping players improve.Would love to hear what you think