This post is from the now defunct website “A GameDev Plays…”, copied here for posterity
Gladiabots is a game of tactical bot programming. The player programs their battle bots in a graphical logic editor and then sets them against computer controlled enemies in an arena. Currently free and in alpha testing, it is already a fun game, and has a roadmap that suggests it will only improve.
When the call went out from one of my twitter followers to provide feedback on his in-development game, I decided to give it a try. And I’m very glad I did. Despite being an “alpha”, Gladiabots is already a fun and challenging game. The basic gameplay is programming your bots in a visual logic editor with a series of if-then logic gates based on predefined conditions and actions. The bots end up with a hierarchical set of conditions like “if shield is low retreat” and “if enemy in close range then shoot”. These bots are then placed into a sparse arena and compete against an opponent. Once battle has begun, the player no longer has any control of the bots. Their actions are completely decided by their AIs - they evaluate the conditions in order until they reach an action, perform it and then start the evaluation process again. The player simply watches the contest unfold - although they can slow-down, fast-forward or pause the level.
The design is quite clever to include two methods of winning - either destroying the enemy, or by scoring more points by collecting balls and placing them into bases. Thus the player has to decide their priorities. I have seen the same level solved multiple ways, one by a player focussed on delivering firepower on the enemy, another through collecting the balls. This means that despite the simple logic allowed in the AI (no loops or team messaging) there is still enough complexity in the scenarios to require some thought. The AI that wins one level is normally a complete loss on the next.
There are many tools to aid the player in creating their AIs. Firstly they can create a library of AI programs, so they can check back to previous designs or build from a duplicate of existing AIs. If there is more than one bot in a level each can be assigned a different AI, although I never won a level with a split-AI team. Also, when playing, selecting a bot will show its AI rules and how they are being evaluated. A vital debugging tool in the common situation of a bot not doing what is expected.
For an alpha this game feels surprisingly polished. There are a few minor annoyances (can’t change the sound level, and some of the text justification is a little off), but I’d be very surprised if they aren’t dealt with by the final release. I played for a total nearly 2 hours in order to complete the entire game. During an overnight break, much time was spent thinking about how to better program my bots - the sign of a compelling game. The roadmap suggests that further content is planned, including online play (coop, PvP and tournaments), custom maps and more levels. The developer appears quite responsive on his forums, so I have every expectation that this good game will only get better. Highly recommended.