I'm a Rails/Python/Java developer working my way through the alphabet with different bot strategies. Once I run out of ideas, I will construct a superbot that implements every lesson I learned from the development of all the previous strategies. Surely, it will either fail embarrassingly or skyrocket directly to the top of the ladder.
- Alice (retired): Straightforward rule-based bot. Tries to gang up on enemies; suicides when low and attempts to occupy the center.
- Bob (1.0.2): First attempt at a heuristic spawncamping bot. Each bot attacks adjacent enemies if available, or else considers "all" possible combinations of moves its local bot cluster, and tries to pick the one with the best anticipated result. Bob is a very slow bot, with a timeout rate around 7%.
- Carol (1.0.1): Bob's cousin is a much more advanced heuristic spawncamper, with better prediction and a considerable broader scope in every way. Carol's so much better optimized than Bob is that she's a lot faster than Bob despite doing much more work, but she still has occasional timeout problems.
- Dave (1.2.4): By the time Carol 1.0.1 was finally completed, I really wanted to take a step back from the world of global heuristics, to try my hand again at a rule-based bot. Each Dave bot attacks and pursues enemies until it's low on health, then retreats to the back lines to play defense. Dave 1.0.0 was released the day I discovered that the build was capable of occasionally beating liquid 1.0.
- Eve (unreleased): Reserved for any overtly-exploitative cheats or hacks I can come up with.
- Frank(unreleased): My goal for Frank is to build a decent human-readable shortbot (no zlib). The result is a fairly reckless bot, based very loosely on Dave's behavior.