This is a post that I'm virtually certain will be misinterpreted. But it's an important enough issue that I'm going to bet that my writing skills are sufficient to provide clarity to a rather muddy issue.
First off, though, a disclaimer: Science is good. Policy informed by science is good. Leadership informed by science is good. The alternative to all of the above is bad. Nothing I am about to say is to be taken as support for creationism, global warming denial, diminution of White House science advisers or the re-excommunication of Galileo.
However, there is a conflict that lies between the fuzziness that is innate to scientific inquiry and the precision that is required for policy -- and more broadly, leadership. We see this conflict whenever global warming deniers trot out scientists who disagree with mainstream theories and we are forced to explain to the deniers that while the nature of scientific inquiry invites debate, the presence of a debate per se does not imply anything about the preponderance of evidence. As Joe Romm has pointed out, Einstein's revisions to the laws of motion did not prove that Issac Newton was an insufferable quack. It just meant that science is innately fallible and subject to revision. Or as Keynes famously said, "When the facts change, I change my mind. What do you do?"
So far, I don't think I've said anything novel or controversial. But here's the catch: The same logic that compels us to acknowledge that science is fallible and evolves must also compel us to acknowledge that policy based on science might be wrong. This is not to suggest that a 1-percent doubt ought to stand in the way of policy based on 99 percent certainty, but rather to recognize that good policy must retain sufficient flexibility to "change its mind."