It’s been going slow these days, with only a couple of hours of Mox programming per week. I had to rework my transaction framework to support AI. I had to implement choice resolving for mana payment (basically lots of combinatorics – How many different complete/final payments of cost A can you make using manapool B?).
But today is different. Today, the computer beat me for the first time. Now, it wasn’t exactly a blowout. In fact, I let it win. But after Shock-ing me 10 times (no, my Shock deck is not legal!), it won. But what is important is not its victory, but the fact that it chose to use these Shock against me, without me ever telling it to in the code for the Shock card. That means I can code other cards, and the AI should be able to use them correctly, as long as the AI heuristic is precise enough and that the look-ahead for the AI is far enough in the future.
I will try to post a video of this later on (not very exciting but the proof of concept is there).