What does it take for a game to be good and also centered around an abstract dynamic?
For instance, can we have a game that "explores" or is centered around the idea of allometric scaling? (This is where relative strengths and weaknesses change disproportionately with changes in something's size.) Such an idea seems promising, but how can a game be made from it?
First, it seems there must be complications of various kinds, including randomness and human opponents. If the principles are layed out too directly, too obviously, then the game becomes a fairly simple optimization problem, or is merely a didactic lesson. The principle must be woven into other game elements, and they must all interact. The trick is to keep the principle of major importance -- not to smother it in so much other stuff that it becomes invisible -- while keeping things interesting.
Second, the players must have repeated choices that directly relate to the principle: they must provide an input, and the principle guides the output dictated by the rules. In this way, the player will come to understand the principle at work, and thus generally as well. The key is that these decisions must be repeated, and the response must be fairly direct, otherwise the players will not be able to test the game elements. Of course, repeated playings always allow more attempts, but for that the game must be very good and fairly short.