It's been a month since my last dev log. Does that mean I gave up? No dear reader, it means I spent a lot of time and procrastination on a large refactor into state machines. This in turn allowed me to get my head around an AI that makes "better" decisions.

AI improvements

Previously the AI would completely ignore cities. Sometimes it would take them "by accident". Now it will intentionally move toward your cities and neutral cities. Since the AI now captures cities, it now builds more units. This makes a more formidable opponent. 

Previously the AI would always chase your first unit. Once it chose "the first", it would never change focus. That allowed you to bait the AI and lead its tanks around. Now the AI chooses your closest unit. That means any of your units might be the target of AI aggression.

Too difficult

I was afraid that making these improvements would make the AI too challenging to defeat and lead to a tedious slog. However I found it's still easy to win if you use the "right" technique. Can you guess what that technique is? Here's a hint in the code.

I also decided to update the second training scenario. In this second scenario you have one city, three neutral cities, and one AI unit. Previously the AI would ignore the cities and head towards your unit. This was more beginner friendly. However now that the AI targets cities, it makes it harder. To counteract that I moved the cities close to the human player.