I'm not very good at coming up with novel ideas, but I am pretty good at being objectively critical about ideas, so the best I can do is present to you what a good competition game will have:
-Simple: You should be able to explain the concept to a 5 year old in 2 minutes.
-Seemingly infinite solutions: The more possible approaches to winning the game, the better. For example with ants, you could be greedy and go for food, or be aggressive and go for hills, or defensive and kill off ants; you could use A* to route to locations, or diffusion maps to influence ants to locations, or random sampling.
-impossible to be optimal: You shouldn't be able to brute force, or brute force + cache an optimal solution.
-The game should be easy to track. You should be able to relate what is going on in the game, with the code you wrote.
-Being able to accept input and run should require very little code.
-modifying a starter kit to do the basics of playing the game should require no more than 5 minutes of development, given a tutorial.
I think that's all I got for now. Keep coming with the ideas. So far I'm not a fan of any of them
.