Yes, that's part of making a good model, what are the negatives, vs the interests for parties (It's also what the Bush administration specifically said for their reasoning too, which was in the paper I cited). For weapons, for example (Of all times), limiting production has the negative of reducing research into delivery mechanisms, and potential countermeasures to those new developments subsequently. That's something that needs to be taken into account.
You act like modelling never attempts to discern the precise method of action, in fact its been one of your biggest arguments. But that's completely wrong--in fact, getting the precise methods of action, so better models can be made, is one of the most important parts of data collection. Once you start having multiple points of risk and multiple rewards--models become extremely complex, and less reliable. Some still have enormous value, and can give you a pretty good idea of possible outcomes. But purposely getting exact methods of action is important (This is why I brought up plane overbooking yesterday--understanding that some of the behavior comes from flight delays, allows you to create a much better model).
But that all illustrates why GW, in regards to the current attempts at cooperation, is a lot easier to 'predict'--because it falls into a very specific common problem in modeling (Tragedy of the commons). Where self interest is very high for those willing to use the resource, and the direct punishment doesn't exist. That self interest is compounded because its driven by basic needs, billions of people are trying to just feed themselves and heat their homes--expecting altruistic cooperation that makes that more difficult is naive. That kind of thing disrupts cooperation at *every level* studied. All models are wrong, but some have value--this one has a truck load of value, so much so that I'd almost say its common sense.