Posted on June 11, 2009 by

Let me tell you how I came to be curious about the issue of model simulations and predictive analysis. First, I am not a statistician. By day, I am an internet media lawyer in a mid-sized law firm in the heart of Boston . Up until now, all the statistics I know came from one stats course in my recently completed MBA at Suffolk University in and some SPSS work and stats discussions in a Marketing Research course in the same program. As a child, I was math-limited and learning this takes a lot of effort for me. However, studying and learning statistics and regression analysis has now become an absolute maniacal obsession for me and in every non-work time waking moment, I am reading every worthwhile book and article I can get my hands on regarding statistical modeling; predictive analysis, algorithm creation; machine learning and computer modeling. I am even re-entering a math and algebra class to better understand this arena. In particular, I want to learn how and whether these tools can help to reliably predict world affairs. I choose those last words carefully. There are some pretty controversial claims being made out there about prognostications, predictive analysis and the future of the world. The key words are reliably predict.
For those of you who are up on this topic, you may find my book list humorous as it shows how much of a rookie I am. Here are a few titles; 1) Super Crunchers by Ian Ayers; 2) The Black Swan by Nassim Talen; 3) Global Catastrophes and Trends in The Next Fifty Years by Vacliv Smil and 4) Global Catastrophic Risks by Milan Cirkovic. These are just a few. I’ve read a number of others about Futurism and Predictions and was unhappy with those as they were unscientific and in some cases religiously based.
Recently, I came upon the work of New York University Professor Bruce Bueno de Mesquita, who says he has a computer model that can reliably (90% of the time) predict the outcome of almost any international conflict, provided the basic input is accurate. He has done work for the Pentagon and a number of Fortune 500 companies and has some ideas about how to resolve the Israeli-Palestinian conflict. Basically, he suggests enhancing the tourist industry in the country and dividing the revenue between Israel and Palestine based on a population based formula. If there is violence, the tourists won’t come and both sides lose. Bueno de Mesquita ascribes to “game theory” or “rational choice” as the core of his model.
I digress again. I won’t delve into the details of Bueno de Mesquita’s model yet, as I plan to study it and what he is saying about it first. I have an open mind on his work, given the inherent limitations of such models outlined in The Black Swan. I have seen Bueno de Mesquita speak on You Tube, however, and he makes some pretty large claims. In my future blogs, I will discuss his work in detail and what others seem to say about it.
Back in my MBA classes I took a course called Strategic Management where we operated a simulation model game called “CapSim” in which we operated companies on the computer model for several weeks, analyzing data and advancing our companies in teams. It was this experience that caused me to wonder whether computer simulations could help us in the stewardship of world events in the near and not so near future.
With all the problems we have to face—global warming; limited natural resources and coming conflicts over water and food; population declines and all this means—we can use all the tools at our disposal to help us out. I am worried about the future and what it will mean for my son and daughter. So my question is whether machine learning and predictive analysis will help us see new ways to guide us through difficult times. Please let me know your thoughts. I am particularly interested in recommendations for my studies in this area. Thanks! Jeff

Comments (3)


  1. Robert says:

    All science is, in effect, predictive modeling, and so it represents some of the best in what humanity can produce. However, I think it will always be impossible to predict the next disruptive element, and such elements end up having a disproportionally important effect on history. Just 12 years ago, there was no Google. It might be argued that once the computer had been invented, Google was inevitable. I'm prepared to accept this. However, what about the works of JS Bach, Mozart, Pablo Picasso, or Frank Gerry? I read once that the most widely recognized piece of music is the opening to Beethoven’s Fifth Symphony. If Beethoven hadn’t conceived of it, was it inevitable that some else would? I doubt it. Much public policy is geared toward the heart of the bell curve, but I suspect history is often shaped by the outliers of bell curves and it is this group that is so inaccessible to predictive models.
    — Bob McNulty

  2. smarterfleet says:

    Correlation is not causality;
    This is where the all problem is with predictive statictics-regression;
    To identify the cause of a living system is quite a challange, this is way the black swan still tends to appear;

  3. Jay Gary PhD says:

    Jeffrey, I teach courses on ‘global futures and system dynamics’ and I have M.A. and Doctoral students cut their teeth on a 5th generation integrated world model, Barry Hughes IF model, The textbook is worth a read, Hughes, B., & Hillebrand, E. E. (2006). Exploring and shaping international futures. Boulder, CO: Paradigm. This as much public policy analysis/learning at the macro level as it is forecasting. There is always a tension between the computer model and our own mental models– who is traning who!

Leave a Reply