One of my goals in Eggs vs Humans is to deemphasize the importance of speed and reflexes, making knowledge, smarts and strategy more important. A player that’s a brilliant strategist with mediocre reflexes should beat a middling strategist with great reflexes.
For a game such as Starcraft, APM puts a ceiling of what you can accomplish. (“APM” means “actions per minute”, a measure how fast you can mash your mouse and keyboard). It’s not that the game has no strategy, but it’s so fast-paced, many people barely have time to think. There’s always something to do; if you have more APM, you can use it. This isn’t necessarily a bad thing. Like soldiers and economy, your hand speed is a resource, and figuring out how to best use it is strategy in itself.
But it’s not what I wanted with Eggs vs Humans. There could be bursts where the player would feel lots of pressure, but a casual player shouldn’t feel constantly overwhelmed, even if he considers 30 APM to be taxing. I think this goal will be met, although some casuals will need to playtest to be sure!
A game that is accessible to slow players likely necessitates that it’s not useful to be superhumanly fast. Fortunately, this is also something I desired, wanting a useful APM of about 80. (As a reference, I’m about 50-60; a pro Starcraft player could be 200-300).
But at least when it comes to the AI, smarts may not be more important than speed. I have a feeling that as long as the AI simply builds stuff as quickly as possible, it will give casual players problems. It doesn’t have to be smart or good stuff, just lots of stuff.
Fortunately, humans invented this concept called “measurement”. I can measure player speed, and some testing of AI vs AI battles could determine how important smarts vs speed is.
So far, there are 3 AI’s, perhaps best called “Random”, “Simple”, and “Adequate”.
Random does what the package says; the AI essentially rolls a die to decide what to do.
Simple has some behavior to mimic a basic understanding of the game. It knows it needs at least one of each non-combat unit eventually. After that, it won’t train non-combat units if its army is too small. It will attack when its army is big enough, and retreat when its army is small. (What it considers a “big” army depends on its randomly generated personality.) Its choices of units, buildings and research is random. When its fort is almost destroyed, it will throw its army at you in a last-ditch attempt to save the game.
Adequate adds some more sophisticated ideas. It understands the difference between air and ground units, and will train anti-air units if needed. It has three personality traits, giving it tendencies and making its builds a bit more focused. (Each of these 3 traits have 4 values, so 64 different combinations!) For example, the “Air” personality will train more air units than normal, while leaving the cavern poorly defended. A “Defensive” personality prefers units with a higher health to DPS ratio, and will hesitate to attack until its army has reached maximum size. A “Rush” personality will try to attack you early, while “Develop” tries to get by with the smallest army possible.
It’s almost certainly true that if I pitted the AI’s against each other all able to build at the same speed, Adequate > Simple > Random. But how important is speed? What if Adequate had a 50% production penalty versus Simple? Can I measure how good the AI’s are?
This week, I’ll set up some tests and see what the results are!