The Perils of Forecasting
This week, we've invited Strong Towns members to respond to a series of questions on Nassim Taleb's book, Antifragile. You should really read the book (it's a big inspiration for Strong Towns thinking), but if you haven't, you'll still find it easy to jump in on these topics and conversations, based on the first four chapters of Antifragile.
The following response is Strong Towns member, Zvi Leve's, answer to this question:
Taleb contends that fragile systems require predictions -- to ensure calamity doesn't occur -- and that these predictions necessarily make the system more fragile. What are implications for traffic modeling, zoning, sizing utility systems, etc...?
An ill-fated megaproject. Read more about the Memphis pyramid here.
In her fascinating series of Massy Lectures, ‘The Real World of Technology’, Ursula Franklin discusses how “technology ... has changed our realities of time and space”. Her emphasis on “technology as a practice ... as a system which is far more than its individual material components” provides a useful framework for understanding our current planning and forecasting practices.
Given the speed and scale of the ‘practice’ of modern city and town development, planning and forecasting are essential, yet the fragility of this practice is rarely appreciated. We glibly apply complex forecasting techniques to the evaluation of mega-projects and assume that the results will be accurate enough to create equally complex financing vehicles. It is well known that we plan for what we measure (and quantify) and this has significant implications for how we approach problems. Taleb notes that under the ‘fragalista’ approach “We avoid interference with things we don’t understand.... What we do not see is not there; what we do not understand does not exist.... We mistake the unknown for the nonexistent.” We feel housing costs and congestion, yet we do not recognize that the root cause is ‘growth’ itself.
Land use development patterns and transportation are intimately related, but our efforts to understand and forecast the linkages between them have had only limited success. We still tend to evaluate large-scale projects based on a mountain of assumptions about associated impacts. Key assumptions about seemingly simple variables such as the “value of time” or the “discount rate” over the long term can have significant impact on project evaluations. Change the assumptions and the ‘value’ of a project can change dramatically!
Our ‘modern’ development paradigm
In our ‘modern’ development paradigm, infrastructure investments have traditionally been funded from tax revenues whereas land-use developments have tended to be financed privately. Public-private partnerships and the wide use of tax incentives to attract private ‘investments’ have somewhat blurred these distinctions and created co-dependencies between the different actors involved in the development process. But how do we decide if something is worth investing in? According to Taleb, “Moderns today try to create inventions from situations of comfort, safety and predictability... [yet] the richer we become, the harder it gets to live within our means.” At Strong Towns we are all familiar with the Growth Ponzi Scheme, but it is less clear how we can extricate ourselves from the ever expanding cycle of debt-financed ‘growth’.
“Perhaps if we recognized the utility of less precise measures of success, we might also recognize the inherent fragility of our ‘leveraged’ forecasting and planning models. ”
So what makes our forecasting techniques so fragile? The value of “real estate” lends itself to measurement in monetary terms, as do certain characteristics of the transportation system (number of vehicles, travel time, congestion levels, tolls and fares, etc.), but how do we measure the ‘fragility’ of such evaluations? Should we be looking at monetary measures, or are these simply paper promises based on debt which will eventually come due? How can we evaluate resilience to shocks, both real and financial? Disaster-planning has become a ‘high-growth’ area of transportation planning, but Taleb emphasizes that ‘post-event adaptation' is too late: “Nature does not need to predict future events precisely.”
Economists recognize that there is a degree of arbitrariness in all “performance metrics.” Attributes associated with ‘quality of life’ are not easily quantified – how does one measure the ‘happy city’ or the ‘smart city’? Can an ‘economics of well-being’ help us plan for the things which are truly important? Perhaps if we recognized the utility of less precise measures of success, we might also recognize the inherent fragility of our ‘leveraged’ forecasting and planning models.
Related stories
About the author
Zvi Leve is a transportation modelling expert whose interests focus on the linkages between transportation, land use, and sustainable development. With an educational background in economics, statistics, operations research and urban planning and many years of international experience he brings an ideal blend of technical and creative skills to projects. His PhD research on “Using parking pricing policy as a travel demand management tool” highlighted the importance of price signals in system performance. When he is not spending his time evaluating the implications of large-scale infrastructure projects, he can be found promoting human-scaled cities and active transportation. He is active in numerous community groups dedicated to rethinking the role of transportation and the built environment in our lives.