There is a popular myth that George Box's quote
All Models are wrong, some models are useful
is a license to stop thinking intelligently about models and how they are used for nearly everything we do. Models are simply aids to thought, especially thoughts about actions in systems that are complex. Models are used to understand more about the system and, in that way, can help the decision maker make decisions that will bring about better results.
I wonder if those self-proclaimed voices have actually read Box's book Robustness in the Strategy of Scientific Model Building, or had to build models of complex systems prior to constructing physical or virtual systems, or worked in a area where modeling of systems was the basis for making management decisions with other peoples money.
Box's book is about the philosophy of robust procedures. It argues that the circa 1979 emphasis by statistical researchers on ad hoc methods of robust estimation is mistaken. Classical methods of estimation should be retained using models which more appropriately represent reality. Attention should not be confined merely to discrepancies arising from outliers and heavy tailed distributions but should be extended to include serial dependence, need for transformation and other problems. Some researches of this kind using Bayes theorem are discussed.
So without actually reading the book, applying the classical estimating processes, examining the Bayesian processes, or incorporating these statistical models with the prior conditions that drive the modeling parameters, the use of Box's quote is just a glib toss off quote so common now with the 3.0 version of actually doing the hard work to solve hard problems.
Of course models are useful. The notion they are not is naive and misinformed. There is a model of the air traffic control system that is used everyday in our Enroute ATC here in Longmont, to get aircraft across the US and into the ATCT/TRACON Center at DIA. There are models of drug toxicity used for clinical trials, models of vibration damping used for stability control systems, models of cost and schedule adherence, models of environmental cleanup processes, models of climate change, models of capitalization of assets, models of nearly everything we depend on in our modern life.
What George Box was speaking about is how the underlying statistical processes are used or misused. This is the frequentist versus Bayesian conversation, that is never ending. This is also the continuous discussion in journal articles based on statistical inference. So stop reading the grocery store populist magazines and start subscribing to Science and Nature
If those Management 3.0 and Agile thought leader voices would actually read the book, go take a statistical forecasting class at the university, stop reading the populist repeating of bad quotes, read the Bayesian statistics text books used on our biopharma, probabilistic risk analysis, and non-linear dynamics programs, then go themselves work a program using their new found knowledge, maybe they'd not be so glib about All Models are Wrong (because that is a falsehood), some models are useful (a truth if you actually understand and can apply modeling).
In our daily work we depend on several modeling tools and the models they produce:
- Systems Dynamics
- Monte Carlo simulations of cost, schedule, and technical performance interactions
So time to buck up and do some homework on applying modeling to complex problems. In the end when those thought leaders complain about "populist" models like OODA they can't seem to grasp that George Box was actually talking about mathemtcial and computation models. Not their simple non-mathematical models. The difference between populist thought and mathematical and scientifica thought is in the populist world you can't calculate anything, so you have to have the dicussion using arm waving.