Six Key Ideas of Operations Research?

For most of us, teaching a first course in operations research involves trotting out our old friends linear programming, sensitivity analysis, integer programming, and so on. Twenty years ago, we used linear algebra; today, we concentrate more on modeling. But it is not clear that students get the Big Picture. Like “Modeling is useful” or “Decisions can (sometimes) be made on marginal information” or even “If you add constraints, the best solution gets worse”.

Steven Baker, whose Blogspotting I regularly visit and which often provides material here, has a posting entitled Economics and Braille Keyboards pointing to an interview with the economist Robert Frank. Frank has a new book out based on the idea that there are really only about six key ideas in economics, and if you can get students to think about and apply those six ideas, then you have really had a successful introductory course. The title comes from a student essay on why drive-through banks have braille on the keyboard. If blind people can’t drive, why provide braille? The answer is in cost-benefit analysis: it would cost more to have two types of keyboards, so standardization is better (assuming putting on those dots is not costly in itself). As for benefits, this holds even if the benefit is zero. If the benefit is positive (blind people in cabs, say), then it holds even stronger.

What would be the six key ideas/insights about operations research? I listed a few above. Some others might include “If there is randomness, systems without slack run out of control” and “Nonbinding constraints don’t affect the optimal solution” (these are more insights than key ideas). As a prescriptive science (rather than the descriptive science of most economists), it is not clear that this method of teaching works as well, but it is useful for us to think about what we really want to get across in our classes.

4 thoughts on “Six Key Ideas of Operations Research?”

  1. Hardness matters! Sometimes people instinctually believe that they can just throw computing power at some problem and get the answer via more-or-less brute force. I was one of those people. But then the asymmetric TSP showed me otherwise.

    This connects with Economics: the economic consequences of the inefficient use of the capital stock. There might be a free lunch to be had in applying basic LP stuff to more firm-level operational problems which are now eyeballed. (I guess it’s hard to explain why people don’t do it more, since it could be so powerful!)

  2. From conversations with Glenn Wegryn at P&G, Lee Schruben at UC Berkeley, and Vijay Mehrotra at SFSU, there is only one OR principle that stands out:

    OR is a fundamentally practical discipline — not a theoretical disicpline — and OR models must be flexible and changeable in real time, as the world changes.

    Of course, that’s more easily said than done, especially when so many models are built with technology that isn’t amenable to real-time adjustments. It may also be true that the dynamic and hugely complex problems that real companies face today are more approachable with the simulation toolset.

Leave a Reply

Your email address will not be published. Required fields are marked *