Specialization in science, part 1: Specialization of the scientist

This post originally appeared on my old blog

Tyler Cowen of Marginal Revolution requested writings on whether science has become overly specialized and what the ramifications of this might be. I didn’t have any of the ‘leads’ he asked for so I decided to write one. My take is largely anecdotal, and especially based on my experience. I am not familiar with any literature on the subject, although I am sure some exists. However, despite all the whining about this, I am not sure that science is getting more specialized. It is just getting more specific.

First off, there is an important distinction to make: the specialization of the scientist or the specialization of a unit of scientific product. ’Science’ itself is more of a continuum, like music or any other cultural phenomenon. There are very few real boundaries, although some parts may not mix as readily as others. Is music too specialized?

Specialization of the scientist?

While all experts have specializations, there are generally two categories of this, methodological specializations and subject specializations. The expertise for either has to be based on a very large and diverse knowledge base.

Most of the senior scientists I collaborate with or have worked for are ‘methods’ people: they run labs that produce highly specialized measurements, but these measurements may be applied to quite a variety of things]uatrl. For example, I worked for a [lab that measured the widths of tree rings. My co-workers and supervisors in that lab devote their careers to this, because trees are remarkable recorders of their environment, and analysis of their growth patterns can give great insight into past precipitation and temperature, as well as occurrences of discrete events such as hurricanes, fires and landslides. Largely due to the vision of the lab director, Dave Stahle, many of these phenomena were studied. Annual precipitation chronologies were used to make streamflow reconstructions, to reconstruct oceanic circulation variability (e.g., El Niño), and that kind of stuff. And of course they map old-growth forest and do other forestry/conservation type work. But the lab also produces some amazing interdisciplinary science: Showing how extreme drought followed by rather wet years contributed to the spread of disease that decimated 16th century Mexico. Correlating the complete failure of the Roanoke colony and near-failure of Jamestown to extreme drought. Earthquake recurrence on the New Madrid fault in the Mississippi Valley. Provenance and authenticity of the Messiah (Stradivarius) violin. Etc. It is worth mentioning that analysis of tree rings was developed by A. E. Douglass to create a chronology of sunspot activity.

My undergrad advisor, Walt Manger, studied how sea level change interacts with changes in sediment depositional systems (e.g., sequence stratigraphy), as well as Paleozoic invertebrate evolution. My Master’s advisor, Glen Mattioli, began his research career studying thermodynamics and the role of oxygen in the mantle, and now uses GPS to study tectonics of plate margins as well as volcanoes’ plumbing systems both during and in between eruptions. The geochronology labs I work in now all have very specialized and sophisticated instruments for measuring the content of trace elements in certain minerals, and are generally set up for understanding when magma crystallized, metamorphism occurred, or when rocks cooled below certain temperatures due to exhumation. However, in the past year or so, this equipment has also been used to track bison migration patterns by looking at the strontium content of the bison teeth (I think) and some project involving measuring trace element concentrations of a piece of human skin, for reasons that I have forgotten (medical or forensic).

My Ph.D. advisor, Mike Taylor, and I are more specialists in subject matter: in general, we study faults and the deformation of continents. We are interested in a variety of timescales: the seconds timescale of individual earthquake ruptures to fault motion over tens of millions of years, so 1o-7 to 1o7 years, which is a range of what, 14 orders of magnitude? There is quite a depth there, and the behavior of faults is pretty different at opposite ends of that spectrum. The methods we employ are suitably diverse, and call for a wide range of knowledge to understand what we’re doing. I want to know how faults work, so I have to know how mountains erode, how rivers modify their channels, how cosmic radiation is modified in the atmosphere before breaking apart atoms in sand grains (reaction types that nuclear physics textbooks don’t mention), the non-Euclidian geometries of plate motions on the earth, the influence of temperature on the diffusion of everything from helium to lead in different types of crystals. I have to know how to process and interpret satellite radar data, how to efficiently dig several meter deep holes in rocky soil by hand, how to write computer programs to perform Monte Carlo simulations and to solve partial differential equations, how to explain to elderly and probably illiterate Nicaraguans why I want to install monitoring equipment on their land (in Spanish), how to non-verbally direct my Tibetan drivers how to get to a valley with no name and maybe no roads that I have never been to before.

I’m not an expert in all of these things, or even most, but I know enough to get by, and I am an expert in a bit. I also know a whole lot more about all kinds of geology than researchers did a century, or even several decades ago, because the hard work of those researchers, with their myriad techniques and backgrounds, allowed for that knowledge to be incorporated into the general education of later students. While the breadth of my knowledge may not be greater than theirs, it is more advanced. And I certainly don’t feel more specialized.

And I would guess that if you look at the career of a prominent researcher now versus 50 years ago, the amount of ground covered is likely very similar, and there is likely no meaningful change in specialization (although I have no real data on this). Other than a handful of true geniuses involved in scientific revolutions (when new light is shed on a huge range of subjects), few people manage to work widely, and the frequency of them may not have decreased in the past 150 years. Whereas some previous phase of science had Newton, Euler, Kelvin, we’ve had Einstein, Erdos, Von Neumann. And we also have a thousand-fold additional smart and talented, instead of blindingly brilliant, people doing more applied science on a day-to-day basis. These people, myself included, can’t be held to the standards of a Pascal.

However, that breadth of a career’s contribution is quantized very differently, because it’s quite rare to publish the long and exhaustive monographs that blanket a subject. Instead, there are scores of smaller contributions that are more specific, or more specialized, in scope. I will discuss this later, tomorrow if I have time.