2.2) Experiments, Prediction And Simulation

A common source of confusion, particularly where socio-economic systems are involved, is in what it means for a model or simulation to make predictions. Dynamic models project values into the future, giving rise to the perception that they provide predictions. Two considerations are important to answer this question:

• The systems we study are complex; the models we build are simplified. No model provides a complete description of the world and any results are only accurate up to the errors introduced by the omission of variables
• Complex systems are often stochastic and sensitive to initial conditions. A particular starting point may lead to a variety of outcomes and slight disturbances can radically alter outcomes

Given these conditions, the term prediction is very rarely applicable to the outputs of simulation runs. For example, the Limits of Growth report [Meadows 1971] showed the consequences of continued economic development and resource dependence and was heavily (and falsely [Turner 2008]) criticized for inaccurate predictions. In fact, the report's conclusions intended to warn of the limits that resource dependence placed on growth with scenarios as illustrations, not predictions.

A more accurate description is that simulations provide scenarios; given a set of assumptions the system will behave according to certain patterns.