Yet Another End of Science Story

Friday, 1 August 2008 by bbbeard.

And maybe this time, there is a valid point to be made.

Wired Magazine’s cover story last month parleyed the “Petabyte Age” into an argument that the scientific method itself is undergoing a revolution. The thrust of the argument is that extreme computation and extreme data have changed the nature of investigation, driving scientists to abandon their paradigm. Chris Anderson writes:

Scientists are trained to recognize that correlation is not causation, that no conclusions should be drawn simply on the basis of correlation between X and Y (it could just be a coincidence). Instead, you must understand the underlying mechanisms that connect the two. Once you have a model, you can connect the data sets with confidence. Data without a model is just noise.

But faced with massive data, this approach to science — hypothesize, model, test — is becoming obsolete.

As a computational physicist, I have some sympathy with this argument. In my doctoral thesis I argued that computation is a “third discipline”. neither experimental nor theoretical, and with a quality and rigor — and limitations — all its own. Indeed, one of my proudest accomplishments from that era was a paper that reconciled moderate-temperature experimental data with the predictions of low-temperature chiral perturbation theory. The zone connecting these two disparate regimes was largely inaccessible to experimenters and theoreticians alike, but clever algorithmists like my collaborators could accomplish this reconciliation.

But Wired is talking about a different phenomenon. In essence they are lining up to escort the theoreticians and experimentalists off the stage. In my heart I cannot endorse this trend as a good thing. Personally, I would be lost without the framework of theory to explain observations of computational or experimental origin. “No theory should be believed until it has been confirmed by experiment” is the creed of the experimentalists — and conversely, “No experiment should be believed until it is confirmed by theory”. An example of the latter is Rutherford’s classic scattering experiment with alpha particles and gold foil. Without a fit to his theoretical calculation of scattering cross section, his experiment would have been a mild curiosity. But because it conclusively showed the concentration of atomic mass in the nucleus, it revolutionized our view of matter.

Conversely, I see in some of my colleagues that a sincere and debilitating ignorance of theory is costly and dangerous. A recent example is the hunting for resonant instability phenomena in fluids (e.g. Faraday waves) using computational fluid dynamics. Without theory to guide the search, extreme computation is extremely wasteful, since there are sizable regions of the spectrum (especially at low frequencies) without resonances, and a random search is unlikely to turn up these instabilities. An unguided search that is less than exhaustive is more likely than not to guide one to the wrong conclusion.

But at the same time, the time’s they are a-changin’. A whole new generation of investigators is coming of age with Google, and supercomputers on the desktop. So am I headed for the Recycle Bin of history — just as my forebears, who finally had to admit that real calculation could be accomplished without the aid of a tab sheet, and that real documents could be prepared without a pencil and yellow pad?

Posted in Science | Print | 2 Comments »