Some thoughts on Serious/Simulation gaming.
So, serious games are games intended solely for learning/training purposes. This raises a question; why are they used and why do they work? Looking at games in general, they all involve learning in some way. At the base level, you need to learn the various interfaces and control systems used in the game and (preferably) how to use these effectively. Then you need to learn the strategy level of a game. This is a broad level and depends on the game you are playing. In a game like Tetris, you simply need to learn how to get the most points and how to stay playing for as long as possible. In a more complex game, such as a real-time strategy or city management game, these strategies would be more complex and may seem counter-intuitive at times as complex games will not always offer a clear line of sight to your "end goal".
The question of why these games work can be answered by this look at games (meaning being able to play multiple games well, rather than just a specific game or a specific genre of games). The most effective way to be good at games is not to understand the "how" of playing a game, but to understand the "why" of playing the game. This means to understand the fundamental aspects of why what you do in a game works. This means you develop skills that can be applied elsewhere, rather than simply to one game.
This leads on to answering the question of why the games are used. While there are other concerns, such as cost or the fact that "on the job" training would be impossible (or irresponsible), the main reason should be this disambiguation of skills. You don't need to train people to push a button (or if you do, you don't need an expensive simulation to do it), you need to train people to understand why they are pushing a button. This is also useful as otherwise you'd end up with people trained to be good at simulations rather than real life situations.
This reminds me a lot of the training games in Ender's game, where the recruits play games in order to develop strategic skills. These games are basically simulations of the battles they would ultimately end up fighting, but they are still very unlike the battles themselves. In the end, the students end up playing a "game" that turns out to not be a simulation. (Interestingly, the Wikipedia article on America's Army compares it to these games).
This brings me to simulation games. Mostly it brings me to various games made by Will Wright (such as SimCity and The Sims). It makes me think of how, often the most effective games as learning tools, are ones where you don't realise you're meant to be learning. This took me back to a talk by Will Wright at TED, where he talks about designing games that are "learning toys" and his upcoming game, "Spore". In these games, the object is not to "win", as the games are open-ended (much like Tetris), but instead focuses on "not losing" for as long as possible.