Quote Originally Posted by nichols View Post
Here's what's going on now with JSAF down at JFCOM. We are also looking at incorporating aspects of PMFserv into a first person shooter that already links up with JSAF through a HLA link.

http://www.seas.upenn.edu/~npelecha/...V_CROWDS05.pdf
I hate to talk down our own products, but I am deeply skeptical of the utility of the "non-kinetic" models being developed here (=JFCOM) and elsewhere, like SEAS and JNEM. Creating a complex, social environment that reacts to a training audience's actions in simulation in realistic ways, is not just beyond the technology as it exists today, I don't think we understand that sort of thing mathematically well enough to model it to the degree where it could be a useful interactive training tool. Maybe it is not even possible?

"Simulation-driven exercise" (as opposed to MSEL-driven exercises) is the latest craze and is supposed to be the fruit of increasingly sophisticated behavioral modeling, but I don't think we'll see the day. Automated red/white/gray forces might work at a very tactical level (like maybe in VBS2), but at a larger scale it gets too complicated, too quickly. I am also concerned we're barking up the wrong tree if we think we can reduce complex social, cultural and political contexts and interactions to something calculable. I know the adage "all simulations are wrong, but some are useful," but improperly used simulation can also be counterproductive, because it can teach the wrong lessons and/ or breed false confidence in the assumptions underlying the models.

Simulation is excellent for creating simulated physical reality that allows training where it would otherwise be impossible or impractical in a completely live setting, but falls far short of simulating human reality.

EDIT: it is nice to see this discussed on SWC