What about the following issue: I suspect that a simulation (agent-based or otherwise) of the "stock market" (e.g. the DJIA) could be developed which is statistically indistinguishable from the real thing. That is, the moments, fat tails, etc. would not serve to allow a statistician to distinguish between the simulated signal and some, as yet unobserved, actual data. Yet, such a simulation would have no predictive value except on some set of measure zero. However similar to the real world the simulation is, it won't tell you the level of the DJIA next Tuesday. How is it possible to make simulations useful for PREDICTION?
Frank --- Frank C. Wimberly 140 Calle Ojo Feliz (505) 995-8715 or (505) 670-9918 (cell) Santa Fe, NM [EMAIL PROTECTED] -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Robert Cordingley Sent: Tuesday, August 08, 2006 12:53 PM To: The Friday Morning Applied Complexity Coffee Group Subject: Re: [FRIAM] Agent-Based Modelling of a Blowback - How Terrorists are made The recent discussions on developing models for political analysis have been very interesting. I fully expect that to persuade folks (policymakers and business leaders) to part with their organization's time and money will require demonstrable results. Demonstrating one can build a model and show life-like performance is great, but proving it has value, matches reality and isn't just another SimCity seems to me to be what's missing. I know this is a bootstrap problem, if one could get the funding one would be sure it would prove itself. The question has to be answered: does the process work in this domain? Do the ethnographic studies, the incorporation of the best political advisors, etc., perhaps with the use of all the computing power you can dream of, along with the latest and sharpest computing tools produce a system that has measurable performance against the real world. What is the probability that when X is tested, Y will occur? When does chaos takeover? Is it meaningful in the time it takes to implement policy? Having performance based results are key to success and probably not readily shared. (For example, if someone has a functioning model of the stock market that works, I'd expect them to keep it a pretty closely guarded secret.) I'd recommend studies be done on a small scale perhaps to model the performance of island or tribal cultures. With solid performance data that proves the technology, one can build a case for larger implementations. Do such results exist? Robert (my 2c) Justin Lyon wrote: Jochen, 20/20 hindsight can only be turned into 20/20 foresight with simulation. Yet, for some reason, I have repeatedly failed to convince policymakers of this in numerous meetings. I did an analysis for one of my MIT classes using strategy dynamics (basically, a dumbed down version of system dynamics for non-math people) to look at the growth of islamists in Afghanistan during the 80s. I hypothesized that the use of strategy dynamics by intelligence agencies would make sense as a way of developing a framework for analyzing complex situations and providing clear insights into possible future issues, including possible blowback situations. I then worked with Dr. Warren (an LBS professor who was teaching system dynamics at MIT via distance learning) and some other colleagues to use strategy dynamics to look at the conflict in Sierra Leonne and we had the opportunity to present the findings to the director of the secretary-generals office of the UN in New York. In both cases, I tried to get more funding to explore using strategy dynamics and system dynamics to analyze terrorists issues, but failed to know the right people or how to navigate the paperwork to secure funding. Since it's easier selling work to corporations, that is where I focus. But, I still remain convinced that system dynamics, enhanced with agent based models, in a hybrid model using software like NetLogo or AnyLogic would be a powerful tool for intelligence purposes. The strategy dynamics process is well-suited to gathering data in a structured manner that can be easily fed to analysts back home. I call it developing a strategic simulation architecture (SSA). It can be taught to people in a few weeks. We even discussed training people at the UN and with the head of police in Sierra Leonne who got it but, once again, we were stymied by lack of funding. The key benefit of strategy dynamics, system dynamics and agent based models are their ability to deal with intangibles, such as the accumulation of anger in a given population and then provide insights into plausible scenarios on how that anger impacts the inflow of new recruits into terrorist organizations. See my short paper here for more: http://s158641480.onlinehome.us/public/DS-004_SSA_Terrorism_V0-5_en.doc Would love to hear your thoughts as the paper has languished in obscurity since I wrote it in 2001. :-P Best, Justin Jochen Fromm wrote: If the USA delivers weapons and military knowledge to autonomous parties in instable countries like Israel, Afghanistan and the former Iraq and even trains people there to fight, it is of course not surprising at all (perhaps even unavoidable) that eventually these weapons will be used for an unintended purpose against the will of the US, especially if all these people can do and have learned is to fight. Although it is therefore obvious that a blowback can happen in this case, it would perhaps interesting to find out the circumstances when it happens exactly, for example by simulating the phenomenon with agent-based modelling in the way Marcus mentioned http://en.wikipedia.org/wiki/Blowback_(intelligence) I guess one sequence how terrorists are made goes in a chain of events like this: 1. A superpower first delivers weapons and military knowledge to autonomous parties or groups in instable countries (according to the proverb "The enemy of my enemy is my friend") 2. The autonomous parties succeed in their conflict, fight or resistance against something, e.g. an occupier or aggressor (Bin Laden was successful against the Russian occupier) 3. The autonomous parties do something that is not intended by the superpower (for example bombing embassies in their home countries) 4. The superpower turns against the autonomous parties, threatens them or tries to eliminate them (the Clinton administration for example tried to eliminate Bin Laden with a Cruise missile attack) 5. The autonomous parties react: they are going mad (become terrorists) and plan a terrorist attack on the superpower Terms are relative: the terrorist for one is a freedom fighter for the other and vice versa. -J. -----Original Message----- From: Marcus G. Daniels Sent: Tuesday, August 08, 2006 7:32 AM To: The Friday Morning Applied Complexity Coffee Group Subject: Re: [FRIAM] Friam Digest, Vol 38, Issue 3 [...] Zbigniew Brzezinski might have pondered "if we fund the Mujahideen to fight the Soviets, what's the likelihood these people will endure and extend their narcissistic rage toward the United States [as Al-Qaeda]". Or the Mossad might have thought more carefully about how much rope they extended to the Hamas. A computer simulation that tracked these organizations as existing and intermixing with the general population (trying to spread their message) could provide some risk profile for the kind of damage they could do. It would at least remind elected officials in later years of the fact they exist at all. [...] I see such a model as sort of thermometer to answer questions like: Who is mad What are they doing now (as a group, relevant to the conflict) What could they do in the next week, month & year, if they achieve it What can't they do in the next week, month & year if they are stopped Where are they Who are they connected to as allies and as enemies What do they want What do they need What do they believe and how mutable is it ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org ============================================================ FRIAM Applied Complexity Group listserv Meets Fridays 9a-11:30 at cafe at St. John's College lectures, archives, unsubscribe, maps at http://www.friam.org
