In lieu of an abstract, here is a brief excerpt of the content:

From Wizards to Trading Zones: Crossing the Chasm of Computers in Scientific Collaboration Jeff Shrager The Chasm Scientists are the prophets of the modern age. Whereas in the past prophets represented God to the masses, today scientists represent reality. As consumers, we are all quite used to computers as our interface to reality—your phone conversations are transmitted by computers, your bank account is virtual money, and nearly everything that you see or hear in the media is refracted through computers. The same is true, and even more so, in science: scientists don’t fly around in space sketching pictures of the earth’s weather, nor do they peer into atoms with microscopes. These are the jobs of instruments whose outputs are nearly always computer-mediated. But whereas consumers don’t worry much about being held at arm’s length from reality by computers (perhaps we should worry more), scientists worry about it quite a lot. Although we have nominated scientists as our interpreters of reality, it’s actually not scientists but engineers, and more and more often software engineers, who have direct access to reality. Software engineers stand between scientists and the instruments that really read the photons. Thus scientists are left to interpret the “massaged reality” offered up by software engineers. But software engineers are essentially carpenters with no special training in science, and so the relationship is problematic.1 No prophet would last long if, rather than claiming a direct line to God, he was just repeating what some carpenter told him! I’ve spent nearly three decades working to bridge this chasm between scientists and reality, most recently between molecular biologists and computer scientists—to arrange it so that molecular biologists don’t need to have reality mediated by software engineers, or at least so that they have a deep enough understanding of computing that they don’t worry about this arrangement. Some of my early attempts were, in retrospect, misguided. However, I believe that they are worth describing for two reasons. First, there are still many similar projects being pursued today; perhaps the 6 108 Jeff Shrager present analysis can help those researchers avoid plummeting into the chasm. More directly, the history of my (mostly misguided) ideas eventually did lead me to what I think is actually a very good solution to the problem of software engineers mediating science. Of course, I could be wrong again; this new idea could be yet another misguided one. Time will tell. Before getting into my (mostly failed) attempts at bridge building, let’s examine a few examples of the breadth and depth of the chasm that we need to cross. The history of statistics in biology is quite interesting. Biology provided many fundamental examples in the development of modern statistics. But with the advent of molecular biology, biologists’ ability to access the discrete components of life (DNA and such) has led them to feel that they could use simple tools like microscopes, gels, and imaging to access the machinery of life, making statistics irrelevant. As a result, molecular biologists over the past three decades were almost never trained in serious statistics. Unfortunately, and inevitably, as the complexity of our understanding of life mounts, sensitive instrumentation is taking over the field; one can no longer make much use of the view through the microscope lens to understand what’s going on in a cell; instead one needs laser spectrometers, microarrays, quantitative PCR, pathway analysis, multiple testing corrections, dynamical systems models, Bayes nets, and other such machinery that require quite subtle computation and statistics. Although biologists are becoming increasingly aware that they need to undertake statistical analyses, they have little idea how to do them, and so they rely upon statisticians (or, more often, students who could only play statisticians on YouTube), taking these folks’ word for the results. This gives these (sometimes pretend) statisticians enormous say and sway over what gets done in a lab, because statistics demands a certain amount of data to get significant results. Unfortunately, the lab side of biology requires enormous time and labor, and getting even small amounts of data can cost a great deal. More than once, biologists have come to me with some data that they spent (not unusually) a year and (not unusually) hundreds of thousands of dollars obtaining, only to find out, after I do just a few minutes of math, that that they would need many times that amount of data to get any useful results, because of variability...


Additional Information

Related ISBN
MARC Record
Launched on MUSE
Open Access
Back To Top

This website uses cookies to ensure you get the best experience on our website. Without cookies your experience may not be seamless.