A linear system is when a response y (e.g. oil supply) can be modeled as a linear function of a set of independent predictors x (e.g. price, demand, GDP growth) using usually a generalized least square approach. Interactions between predictors and feedback from the response to predictors make it impossible to use that simple approach. When a system is stable, you can assume that some predictors are known and set them to a fixed value with a good confidence (e.g. demand growth humming at a constant 2% per year).

OK...Would it be fair to say that because of turmoil like we're experiencing now...supply disruptions from hurricanes and military conflicts (nigeria, iraq, georgia...ad nauseum), capital market chaos, changing patterns in consumer spending and other stuff that modeling of any kind made today would be highly complex and results suspect?