Years ago when I first read Platt, I was drawn to the reference citing Karl Popper. Having read both, I concluded that Platt was really advocating Popperian hypothesis testing, not Baconian inference.

I think Beard and Kushmerick are right on target, and I'm not the first to have advanced a similar position in print, e.g. Phair RD, Misteli T. Kinetic modelling approaches to in vivo imaging. Nat Rev Mol Cell Biol. 2001 2(12):898-907.

To me, there are currently two main streams of systems biology thought: 1) the statistical inferential approach of high throughput data analysis and 2) the dynamic systems, hypothesis testing of targeted smaller-scale experiments.

One way they can be imagined working together is to treat the high-throughput statistical analyses as yielding provocative new hypotheses and the dynamical systems modelers and experimentalists as testing these hypotheses along with their own.

Platt's recommendation to test MANY hypotheses is, as Beard and Kushmerick assert, even better suited to the world of modern systems biology than it was to the world of science in 1964.

No matter how we proceed, though, we can be sure that computational models are essential to the future of biomedical research. Many of us have watched this realization slowly dawn for decades. Biologists who want to move beyond reductionism to synthesis are beginning to see their computers as more than spreadsheets and t-tests. Modeling is here to stay.

Thanks for a thoughtful and insightful paper.

Robert Phair