Kermit Weis, at the 2016 Innovations in Travel Modeling Conference, gave a plenary speech that was honest, soul-bearing, and depressing. He was lamenting about the inaccuracies of travel models and his inability to defend them. Kermit is not alone. Many travel modelers have experienced this sinking feeling that their model might be spewing out seriously wrong information. Thus, we can explain the overwhelming disinterest in back-casting as a validation method. We know the back-cast will look bad, so why expose old wounds? Maybe a dose of planning theory might just make us improve our self-images.
When in the UCLA planning program, I was required to take a course in planning theory from John Friedman. Friedman has been hailed as one of the foremost planning thinkers of the last 4 decades. At the time of my course, Friedman had just written Retracking America: A Theory of Transactive Planning. The key to getting an “A” in Friedman’s class, I was told by a previous-year student, was for me to regurgitate his concepts in my own words. Unfortunately, Friedman wrote so abstractly that I had almost no idea of what he was trying to say. In spite of the fact that the Planning program at UCLA (it was not yet a department) shared a physical and administrative home with the Architecture program, Freidman was reacting against a philosophy of planning as espoused by Daniel Burnham and others with architectural backgrounds who defined urban planning in the early part of the 20th century. My only hope was to go back to Friedman’s sources in the sociological and political science literature and pray they explained things more clearly.
I was particularly drawn to the writings of Charles Lindblom and his disciples. Lindblom was perhaps best known for his 1959 article with the amusing title, “The Science of ‘Muddling Through’”. There was something about Lindblom’s incrementalism that resonated with me. Incrementalism is now weakly embedded into today’s LRTP process, but it is poorly understood by planners and citizens alike. An incrementalist just assumes that long range plans will fail to materialize as intended. This is something us more experienced transportation planners have witnessed often in our careers. There are too many unanticipated events. It is nonsensical to think that we can convincingly model the impacts of transportation improvements over a 30- or 40-year time horizon.
I am not trying to imply that modeling future transportation systems is doomed to failure. I am instead saying that it matters little whether we can correctly make that 40 year forecast with any degree of accuracy. An incrementalist will also tell us that 30- or 40-year forecasts are preposterous, prima facie.
From time to time I have tried to write out the case for “incremental forecasting”, but each time I met with resistance from editors or co-authors. My last attempt was excised from an early draft of NCHRP Report 765. I may never be able to publish a full exposition on the subject because my writing doesn’t seem to be any less muddy than Friedman’s.
When I took systems analysis as an undergraduate I learned the first day that systems were composed of components. A component in the parlance of transportation planning is a “project”. And I can tell you for a fact that project-level forecasts can often be quite satisfying. Creating a system-level forecast from a sequence of project-level forecasts would not be simple; nor would it get us much closer to a solution for our quandary without some fundamental changes in our thought processes and our conversations.
We need to make the incremental nature of planning far more integral to all our procedures, including forecasting. There are different ways this can happen. Most importantly, we need measures and vocabulary to show how mistakes accumulate over time through weak assumptions about scenarios, misspecifications of our alternatives, and inaccuracies in our theories of travel behavior.
I am not suggesting a lowering of expectations. I am saying that we need a style of forecasting that does not unreasonably and irresponsibly raise expectations in the first place, but still fulfills the needs of long range planning.
If you believe I am on the right track or can otherwise contribute to this discussion, please speak up.
Alan Horowitz, Chapel Hill, June 7, 2017