The special issue of Interfaces journal from 2006 was entitled: Homeland Security: Operations Research Initiatives and Applications. You might find some of the papers interesting, since they touch on broad range of topics discussed on this blog. Some of the topics are bio-security/terrorism, emergency response and critical infrastructures. The article number 6 is especially interesting, since the author starts by drawing a direct link between homeland security and the genealogy of operations research expertise that we have been tracing in OEP research.
COLLABORATORY: VITAL SYSTEMS SECURITY
The Vital Systems Security collaboration examines how, today, security is being constituted as an object of knowledge, intervention, and political reflection. It proposes that the security of vital systems such as energy, transportation, communication and health is one norm in relationship to which security is being reproblematized. A central goal of the collaboration is to examine these issues through collective, conceptually driven inquiry that addresses rapidly developing contemporary problems.
Thanks for this, Onur – extremely interesting stuff. It shows that OR experts have been modeling the “vulnerability of vital systems” since at least 1961. And it indicates their expansive field of analysis: including emergency response systems, border security, critical infrastructure protection, hazardous materials management, and much more. The category of “component support profiles” seems most salient to their current homeland security efforts.
The article mentions a daunting array of OR techniques: discrete-event simulation (to determine staffing levels in a bioterrorism attack), integer programming (to predict terrorists’ use of pollutants), first and second-order catastrophic decay models (to study degradation of bioweapon agents in environment), game-theoretic models (to evaluate performance of fingerprint scanning strategies), data-mining techniques (to classify airline passenger threats), queuing theory (passenger screening), greedy algorithms (to minimize performance measures), graph and scheduling algorithms (to solve security and survivability algorithms), fault-tree and event-tree structures (to improve chemical risk assessment), data-perturbation methods (to protect information from unwanted access), linear programming (to schedule patrol cars for NYC police dept), modified MCLP algorithms (to locate emergency response capability for hazardous material spills), simulation analysis (to evaluate ambulance locations), multi-modal network flow models (to plan disaster relief) Bayesian analysis (to rank threats and prioritize safety measures), and on and on….
It would be good to have a sense of how these objects and techniques fit into a more ‘sociological’ field: what university departments, think tanks or other institutions are homes to this kind of research? Who usually sponsors it? What effect has it actually had on the development and implementation of systems-security measures? How does it relate to seemingly similar fields such as risk analysis, decision analysis, organization studies? Do these techniques mainly come out of applied mathematics and engineering?
Hi Andy,
Thanks for these extremely interesting questions.
I think the question of sociological field is extremely interesting. One way to approach the question would be to map this space, possibly using Bourdieusian field theory. I think from this perspective OR expertise seems to be an hybrid position in this universe which feeds off from more stable expertise forms, such as mathematics, physics, economics, management sciences and various types of engineering. From this perspective, then it should not be a surprise why everyone despises IE and OR experts. For instance, Bellman (the developer of dynamic programming and a colleague of Dantzig) writes how doing applied work was looked down upon from the perspective of pure mathematics. From the engineering perspective, even today OR is made fun of as not being a ‘hard’ enough science. No need to mention the economics which has an analogous relationship to the relationship between pure mathematics with applied mathematics.
Then, it must not be surprise that OR was sponsored by the military’s engineering core, RAND and national planning agency. First of all, this cannot be a coincidence that all these positions required a form of expertise that was both abstract and practical at the same time. OR’s controversiality may be due to its ability to gap this seeming opposition that is fundamental to the Western science. It is noteworthy that this is precisely how Bellman justifies his decision to leave the university for RAND. For him, his research on dynamic programming allows him to apply sophisticated mathematical techniques to real world problems. He contrasts his practice to mathematical formalization and proofs. He points out that numerical calculation is harder than abstract formal proofs, since indeed it is harder to stabilize solutions in the latter case. He points out that formalization requires assumptions (just as in the case of economic expertise) and this inevitably means simplification of the outside world. Numerical solutions of real world problems, on the other hand, requires one to accept the reality as a complex ontological existence and try to develop realistic models which will appreciate this complexity as opposed to reducing it. This appreciation of complexity and desire to simulate the world realistically is an objective that OEP/NREC models were pushing forward.
Andrew and Onur,
I can perhaps shed some light to the question, as I am an engineering student in systems sciences in the Helsinki University of Technology, which encompasses many of the methods Andrew mentions. My assessment here is that of a student, i.e. it is strictly based on taking a couple of courses (and take note, I have just restarted my engineering studies after a long pause). Out of the methods you mention, I’ve studied at least tree analysis, integer programming, game theory, modeling and linear programming.
Onur makes a point here: systems science is probable considered “softer” than the “hardest” fields of physics and mathematics. The laboratory of systems analysis is after all in the engineering physics and mathematics department at my university, and I’ve always had the feeling that it is considered one the “softest” things can study at this department.
But yet, having taken the courses, I do not think the difference to “hard” mathematics is that radical. All you basically do in these courses is exercise the methods: in other words, you calculate and apply. True enough, the outside world is present in the course exercises, but it is always simplified — most often in some problem setting from the business world, like in optimizing a factory line, modeling a nation’s consumption or deciding outsourcings in a corporation. I actually almost got a headache from one course which was constantly reducing the complex outside world to some simple market code.
Whence, as a student I do not completely agree with Onur’s distinction between abstract proofs (as simplified) and real world problems (as complex). Even if the difference is sometimes made explicit — I admit that in the introductory lessons this is done and some good teachers mention it more often –, you don’t have to think about the whole problem to pass the courses!
I think it is utterly difficult to bring sociological imagination to the level where you are actually using these methods. I know some good more pluralistic work is being done e.g. in environmental problem modelling, but I’ve always sensed it is slightly in the marginal. But on the other hand I am not competent to judge, there is probably some pragmatic reason for taking the very techno-economical stance to the methods.
As a final side note, there is also a more philosophical branch in the systems analysis at our department. They have been employing philosophers which is very interesting development, but I am not sure how it will affect the problems I have been talking about as student of the normal courses.
Antti,
I agree with most of your observations about teaching of OR techniques as I had to take some of those courses for my undergraduate degree. However, I think there are two quite different issues at stake here:
First one has to do with the distinction between training and practice. I agree in OR/systems training one often times faces hypothetical problems under clean assumptions. One might assume that this is a reasonable thing to do, since the purpose is to master the techniques themselves. Some of the ‘analytical models’ we ran into in OEP could be thought within this perspective. The developers clearly indicate that they are not suitable for real world application, since they are not realistic (e.g. they only assumed one nuclear detonation was going to take place). From this perspective, they could only serve for the training and readiness purposes under preparedness scheme.)
However, the second issue has to do with what we mean by complexity. if we mean chaos, then you are right in the sense that these models do not account for everything that is outside in the world. However, if by complexity we mean something that is in between chaos and simplicity, then I think these models are trying to unconceal something ontic (as opposed to merely ontological) that has complex properties. We might want to call that something a system. I think the idea of a network is the perfect example. Often times network theory is criticized for being reductionist (see Bourdieu), but I think this misses the point about network theory. The claim to complexity is not due to an un-reductionist model that makes no assumptions, but due to its search for complex identities and qualities that a system reveals. And it might even be the case that these qualities can be revealed only when certain assumptions are made… This would bring us to an interesting question about anti-realism: in the sense that to what extent these models come into being without systems theory and a practice around systems security? It would be interesting to think this interdependence between the expertise (epistemology) and objects of expert knowledge (ontology) in terms of the question of problematization?
Finally, I wonder what these philosophers do? What are the areas they are interested in, and what kind of a function they have? What kind of an alliance this is?
Onur,
The difference you make between training and practice is an essential one, and one that I did not address perhaps enough. It is of course different to use these methods in some pedagogical situation than to apply them to real research problems, as only in the latter case the methods have to help some problem of decision making (e.g. on how to protect critical infrastructure). I think it is an interesting problem how much the models are relied on and, on the other hand, how much there are deviations from them.
I will think about your second point. My own interest on vital systems security is on the kind of balances that exist even though the infrastructure accidents increase uncertainty — but at the same time I do not deny that there are new problematizations, whence more dynamic balances are born. I will thus keep your description of “complex identities and qualities that a system reveals” in my mind.
The philosophic research group in question is studying “intelligent behaviour in the context of complex systems involving interaction and feedback” from an acting subject’s perspective. It would be interesting to see how this applies to models of vital systems, but I do not know if it has been done.