I've done (and am doing) some work on this sort of thing, but mostly from the non-government standpoint (although I am working on one locally in relation to my ROTC work that may become formal at our Det at least).

Based on this experience as well as many years in the hobby, I tend to think that the best model is a free-play exercise with a control cell and a number of opposing teams. The control cell works as an encounter resolution system, information control point, and general game manager. We do a four team operational air warfare map exercise here every spring (it's too short, but it does give our cadets a taste of planning), and having free play and a control cell allows for many variations. I'm working on one now that will involve both Army and Air Force cadets and cover some aspects of small wars (mainly in a small theater conflict environment).

On the hobby level this is hard to model without computers, since one of the key aspects needs to be intelligence (or lack thereof) and political activity. Board games, IMO, don't model this very well because they ARE board games with unit counters and such. I'm not a huge fan of card-driven games, although they may possibly be able to simulate some aspects of small wars. The RPG framework could be very useful for small wars simulations, since most of their systems deal with interaction and influence on a personal level.

I can't speak much to the computer side of this, since my design experience has been in the paper realm. I do tend to see computers more useful as tools (information management and dissemination) than I do as actual gaming engines (mainly due to AI limitations...but again my experience here is limited).