You have a point there. But going with your example, the ideal gas law (PV = nRT) still models a system where all inputs are known. While not a perfect model, it's much better for predicting the performance of a thermodynamic system than most economic models are at predicting economic systems.