From Playability to Learning Target

The Problem

So, you've built a simulation that embodies important information about a useful learning domain. Then you've built a game on top of the simulation, and you've found ways to make that game playable, and fun, or at least engaging. Can you declare success?

Not necessarily. It may be possible for people to play your game a lot, and play it successfully, without learning what you want them to learn.

Let's use Shuxin Nie's very nice minimum spanning tree game as an example. In this game the player has to navigate along the edges of a graph whose edges have numeric weights. The goal is to mark out a connected collection of edges that visits all of the nodes while minimizing the total weight of the edges used (such a collection is called a minimum spanning tree ot the graph.) It's a fun game, with a good built-in sense of progress towards the goal, and the feedback Shuxin provides enhances it by telling you when you can do better, keeping you at work.

The learning goal for the game is that players should learn the greedy algorithm for the problem. You can form a minimum spanning tree by choosing the cheapest edge that joins the path you have created to a node you haven't yet visited. Once you realize that, it is fairly easy to play perfectly, though you have to be careful to check all of the appropriate edges. The question is, can Shuxin be sure that people who play her game will learn this method? I think it is doubtful that anyone could learn to play perfectly for arbitrary graphs without getting the concept of the greedy algorithm. BUT (a) it is possible for someone to play perfectly for any finite collection of graphs, such as Shuxin's levels might offer, without learning the algorithm: they could just learn an adequate sequence of moves for each level, by trial and error; and (b) it may be that most people will not play the game long enough or carefully to attain perfect play, even for the levels Shuxin provides.

What can Shuxin do to make it more likely that players will reach her learning goal?


Shuxin's learning goal requires developing concepts at a more abstract level than the play of the game intrinsically requires. She needs to do something to push players to bring these abstract concepts into their thinking. Here are some approaches.

Generalizing these Ideas

These approaches can be applied to many other games. The key idea is to find ways to promote thinking or talking about how to play, and not just playing. A subtheme is, if you raise the demand of the game so as to promote more thoughtful play, you may need to provide some hints or cues to keep the game from becoming too difficult to be engaging for some players. You may then need to use fading to keep players from relying on the cues, rather than on the ideas the cues are intended to suggest.

Not all games require these approaches. If your game is such that playing the game actually requires mastering the ideas targeted in your learning goals, you don't have to worry that people can play the game without learning what you want them to learn. We're assuming here that your game is already playable and fun, so you don't have more to do.