Predicting the unknown
If one takes a course in environmental economic nowadays, one is bombarded with information about environmental Kuznet curves, which has become something of an obsession in the field. Environmental Kuznet curves are simply a fancy name for the arched relationship that exists between some environmental pollutants and income: very poor and very rich countries don’t emit much sulfur dioxide, for example, while middle income countries do. Environmental economists have investigated this topic to death, inventing whole categorization schemes of pollutants, describing which obey this relationship and which do not.
Ever since I took a course in environmental economics at Duke, something’s been bothering me about this single-minded focus on one pollutant’s relationship to income levels. It struck me recently that what scares ecologists most about the future is the sheer pace at which surprising negative consequences of man’s activities appear: the rate of problem generation, defined loosely to include everything from pollutants to invasive species to land-use change, seems high. If we consider this rate, in problems per year (r ), as the product of the number of new technologies in a sector (T) and the proportion of new technologies that prove problematic (p), we can begin to grasp the conundrum. If T is increasing multiplicatively, then all else being equal r would increase at the same rate.
Here, however, my investigations as a scientist have come to a screeching halt. Scientists just don’t publish a crackpot idea like this without an example dataset, and I haven’t found a field yet with good enough data. The chemical industry comes closest. Around 800,000 new chemicals are introduced a year (excluding organic chemicals with complex sequences like DNA), and the annual rate grows by about 4% a year. Sadly, the vast majority of those chemicals are never screened for toxicity (except perhaps by computer modeling), and so the known total of dangerous chemicals (on the TCSA list) only grows by around 2000 chemicals a year. Since r is effectively unknown, p is also unknown, and it’s unclear if society’s screening capacity is really getting 4% more effective a year.
Anyway, this is all important because if, for a given sector, T grows multiplicatively in at all the same way as our general economy does, then P must fall at least that fast to maintain screening capacity. This is a very difficult thing to do: it’s far harder to halve the error rate of a screening process than it is to double the input, for example. There seems to be, barring massive technological changes in the screening processes in any given sector, a limit to how high T can safely go. In other words, there’s a limit to how fast our society can change and still filter out bad outcomes.
