Sunday, July 4, 2010

The fourth paradigm of science
Tom Smith

Seems a bit overblown to me.  I'm probably missing something but I'm not sure what the points are beyond, really big studies are (or can be) better than merely big, let alone small studies; it's great to have computers to analyze data; and Serg Brin is worth $15 billion and so can afford to give $50 million for Parkinson's research, which he has done, because his mom has it and he's afraid he might get it someday.  I do certainly agree you can discover all sorts of things by looking at big data sets and computers make this possible, and the bigger and faster the computers, the bigger the datasets you can analyze.

| Permalink

TrackBack URL for this entry:

Listed below are links to weblogs that reference The fourth paradigm of science
Tom Smith


It may indeed be overblown when applied to Parkinson's in particular, but I suspect this type of large scale data mining epidemiological analysis would prove useful in advancing genome-based personalized medicine.

Posted by: Torrey Surfer | Jul 4, 2010 2:43:03 PM

Epidemiology hasn't done awfully well as a discipline. Cigarette smoking is the biggy. Occupational exposure to asbestos is another success, though a tarnished triumph given the extent to which it has been used to rip off the owners of property containing the near-harmless white asbestos. Just last week it had a small triumph with Legionnaires' Disease and screen wash water. Probably there are a couple more that have passed me by, but it's been a lot of fuss for little return.

Posted by: dearieme | Jul 4, 2010 3:14:01 PM

This remark looked odd: "the classic scientific method: hypothesis, analysis, peer review, publication". Clearly wrong: the the classic scientific method is hypothesis, controlled experiment, analysis, publication. Data pummelling is just a way of generating hypotheses I presume. Except in "Climate Science", of course, which is no science at all.

Posted by: dearieme | Jul 4, 2010 3:20:10 PM

Who knows, maybe he'll find something useful. Isn't that the only thing that matters?

Posted by: Jonathan | Jul 5, 2010 12:38:00 AM