Friday, April 25, 2008

the philosophy of transport modeling (draft)

a new paper by timms from the most recent issue of transportation caught my eye and is the subject of this post. entitled "transport models, philosophy and language" the author attempts to ask a fundamental question about transportation modeling in general: "is it reasonable to expect any transport model to produce accurate results over a medium/long term time horizon, even if the effects of inaccurate exogenous inputs are 'screened out'?" wow. a serious question to ask. to answer it, timms turns to the philosophy of science for direction.

i was going to comment on the fact that accuracy is not the most important quality of a transport model. what we want is for it to be useful for us and our planning purposes. with respect to climate change, we want a travel demand model that can approximate traveler responses to policies aimed at climate change mitigation. timms headed me off during his discussion of the positivist influence on transport modeling (i.e. that we should let the data speak for itself—this seems to me to be a very econometrics (as opposed to economics) way of thinking): "…pure positivism was in fact widely abandoned in the 1970s,to be replaced by … pragmatic positivism..." just how does he define pragmatic positivism? "pragmatic positivism downplays…the importance of model accuracy … emphasising (sic) more a loosely conceived idea of model usefulness." hm. sounds like i'm influenced by the literature. i guess that's not the worst, and it's very nice to have the author investigating the assumptions of our modeling methods. however, we then learn that this approach is not really practical for long-term modeling, since the real-world data which can be used to calibrate and validate model results is displaced in time vis-à-vis model use (i.e. decisions are being made today based on a model that predicts travel in 2030--by the time 2030 rolls around, it's too late for us to do anything about today's decision).

the nugget is that we should begin to think about models in terms of how they're actually used which is as linguistic tools in a communicative planning process. this just means that models should give us insights into system operation and help us communicate those insights to stakeholders.

the main problem i had with the paper was that its conclusion that "…there is a need for a widescale (paradigmatic) change in the field of transport modeling, in terms of both its underlying philosophy and practice." while i agree with this statement, it was simply not supported by the type of analysis provided. additionally, the notion that "[s]uch changes…would be greatly aided if transport modelers were to become more aware of formal philosophical concepts and arguments, in particular from Continental philosophy, and were to incorporate the insights from such philosophy in their writing about models," is hilarious. i'd be the first to encourage more teaching of philosophy to undergraduate engineering students, the future transportation modelers. but, srsly? there would be an uproar. expecting current practitioners to pick up habermas is also not bloody likely. so i felt there was also a disconnect between what was realistically possible in terms of baby steps and what the author had suggested. i am not discouraging this type of inquiry. frankly, i think it's excellent scholarship and i'm really pleased that transportation chose to publish it. it's precisely the type of multidisciplinary work i'd like to pursue.

1 comment:

Anonymous said...