I’m working on some new ways for ANGELINA to consider maps and layouts in the arcade games it designs. Whenever I try to expand or adjust the parameters for designing these game components, I like to pluck out a few arcade/older game archetypes and think how the ideas would evaluate against these old games. As an example, today I mused about using heatmaps to evaluate powerup/enemy placements. Here’s some terrible notebook-scrawl by way of illustration.
The idea being, if ANGELINA records where score was gained or lost during playouts, when it comes to look at layouts it can see where the activity is mostly going on. That might come in handy, because if you think about PacMan such heatmaps show that rewards are pretty evenly spaced around the map, and that danger is normally mixed into areas that are dangerous.
I don’t know if heatmaps will help ANGELINA, hopefully I’ll find out later this week. What I noticed as I went through games testing this theory, though, was that it didn’t really fit all or even most templates I could think of. Asteroids heatmaps would look rather confusing indeed, and places where the player died don’t really correlate with the initial configuration of the map. For a game like Frogger it’s even less clear what’s going on, because a heatmap just looks like the game is a punishing mess of score loss (which, let’s face it, it is – I hate Frogger).
What impresses me is that no matter how many theories or ideas I come up with for evaluating these things, I find the huge variety of arcade games and games from older eras defeats me at some stage. This is to be expected, of course, and it means there is better research ahead and more things to learn. But it’s really great, really impressive, to see how varied and broad these games were. Whereas today, we often find them all in the same section – “Arcade” or “Retro”.
Fun is even more elusive a concept than I originally thought it to be when I started this research. It’s fragmented, confusing and subjective. The search continues!