Eden,

While you bring up sound objections, I disagree that they make gaming COIN impossibly difficult. Games have reached a point of creativity and depth that most things, given the proper focus by developers and interest by gamers, can be simulated.

1. The complexity of the COIN environment makes model-building extremely difficult. It is also very hard to filter out pre-conceived notions and false assumptions brought to the table by modelers or participants.
I think this can be addressed in one of two ways. The first is by using a sand-box mode for the game, which is an extremely popular model. Europa Universalis 2, Hearts of Iron, Victoria, and Geo-Political Simulator are a few examples of complex games played sand-box style. In EU2, the player guides the development of a one of several hundred countries over the course of 400 years, managing its political, economic, social, and military progress as well as balancing the ethnic, religious, resource conflicts within and without the state. GPS, in constrast, places the player in the executive seat of a modern state where he must manage every conceivable aspect of running the nation, including managing the different personalities leading political parties, the military, interest groups, religions, minorities, the media and so on. Both games are real time, with Eu2 running day-by-day and GPS hour-by-hour. Because there are so many options, the player can really take any approach he likes; so different models of politics, economics, etc can at least be simplisticly modelled by the course the player charts. The second option is to build a game that models a specific theory of insurgency. AGEOD's American Civil War, for example, is very detailed in its application and so many concepts and strategies which may be useful in generic strategy games can be severely punishing to the player. For example, in most basic strategy games, the player who attacks first usually wins. in AGEOD's ACW, the design emphasizes operational tempo as opposed to speed. I'm generally in favor of the sand-box design -- the more freedom of action given to players, the more ingeniunity you will see out of them.

2. The 'lessons' and 'principles' from previous COIN operations are rarely transferable - I believe that every COIN op is unique to a far greater extent than more conventional operations.
This would have to be inherent in the game's design. In a sand-box model, the player can apply his own principles and learn his own lessons. That could realistically change from session to session depending on what course the player is pursuing. In a specific model, the designers would have to working with a specific set of lessons already in mind.

3. For high-fidelity you would need a very large number of participants, which can be unmanageable, or some very sophisiticated software capable of simulating the COIN environment for a small number of players. The latter does not exist.
Not necessarily. EU2 and GPS have hundreds of autonomous simulataneously operating participants. EU2 can also run good-sized (16+ players) games. There are several online-browser based RPGs, such as Star Wars Combine, which are even larger. SWC not only drops the player as a single character in the middle of the entire Star Wars universe, it also runs in minute-by-minute real time. The game has been operating for 10+ years now with some of the original players and organizations still active.

4. COIN is slooooooooow. Things develop over months or years. The number one difficulty is figuring out whether you are being successful or not, and why. It requires patience, perseverence, and downright mulishness. You can't game that with people, and we don't have the software to realistically game it using computers.
See above. Most games, such as EU2 and GPS, also provide the option of controlling the speed, though not necessarily the pace, of the game.

Just my 2 cents as one from the gaming generation.