Theory: Test with a Random Robot

Here’s an easy test that can help determine whether your game has meaningful decisions. It’s boring and silly, but it can be very effective. All you have to do is play entirely randomly.

By “entirely randomly” I mean exactly that. Use a die or a deck of cards in place of a human player and then go from there. Let the random mechanic make every choice.

Your game passes if a person can get better results than random play. At that point you can be confident that the player’s decisions are relevant to the outcome. It might be that those decisions are boring, obvious, or otherwise not very satisfying, but the player at least needs to be involved.

Your game fails if random play beats a human who’s (a) reasonably capable of playing the game and (b) trying to win. Wins for random play mean that human players have no significant role in the game; their strategies, tactics, and choices are superfluous. They’re just there to carry out the physical steps required to move the game forward.

It’s worth noting that this test can be performed even on relatively complex games. Magic, for example, involves a lot of decisions, but you can still set up a random opponent. Does the opponent play a land? That’s a binary decision; flip a coin. OK, it does: which land? Assign each one a number and roll a die. Now it’s time to see if it casts spells, so assign each spell it could play a number, tack on an additional number for “no spell,” and roll again.

This sanity check for meaningful decisions might seem unnecessary. However, sound and fury at times signify nothing; it’s possible for the challenge of managing everything a game has going on to obscure the fact that none of it influences the outcome. (I’ve heard of at least one commercially published game where this turned out to be the case, such that a random player had the same chance to win as anybody else!) It’s worth taking an hour just to make sure you haven’t gotten lost in your own game’s complexities, and that the choices you think matter really do.

One thought on “Theory: Test with a Random Robot

  1. Good technique. When I made a simple puzzle game a few months ago I used a similar technique to randomly tap parts of the screen to see what type of performance I could get without thinking deeply.

Leave a comment