I’ve got a seed for a blog post that I want to sort out. I’ve been working with a coach on artistic/creative stuff, and it’s been a blast. She’s an actress and dancer and knows nothing about games but is fascinated to understand game design and how I connect to it artistically. We’ve been doing a thing where outside our normal sessions I’ll play a game and she’ll watch and ask questions, and it’s an incredibly refreshing experience (she doesn’t know games but “gets” things in the sense of being an entertainer.

We were playing Nuclear Throne and she asked how did the enemies decide where to move, and I talked about decision making and random movement and behaviors. I told the anecdote about the Triple Town bears (completely random movement that players perceived as cruel AI) and that one Turok game (a simulation with so many specific conditions that the output just appeared random to the player) and how even Pacman had specific behaviors for the ghosts.

She asked me how do you know if randomness is enough, which gave me pause. I’ve developed an intuition for it, and I did explain how it’s often good to do it in prototyping phase to see if it’s enough off the bat, but I’d love to break it down more.

When you’re making enemy movement behavior in a game where the answer isn’t obvious, how do you decide when randomness will do the trick? What sort of criteria do you use when making those decisions and deciding what to try next?