I just wanted to add a few thoughts to my post yesterday. Readers may be wondering what is wrong with using science to identify the best educational practices and then implementing those best practices. If they are best, why wouldn’t we want to do them?
Let me answer by analogy. We could use science to identify where we could get the highest return on capital. If science can tell us where the highest returns can be found, why would we want to let markets allocate capital and potentially make a lot of mistakes? Government could just use science and avoid all of those errors by making sure capital went to where it could best be used.
Of course, we tried this approach in the Soviet Union and it failed miserably. The primary problem is that science is always uncertain and susceptible to corruption. We can run models to measure returns on capital, but we have uncertainty about the models and we have uncertainty about the future. Markets provide a reality test to scientific models by allowing us to choose among competing models and experience the consequences of choosing wisely or not. Science can advise us, but only choice, freedom, and experience permit us to benefit from what science has to offer.
And even more dangerous is that in the absence of choice and competition among scientific models, authorities will allow their own interests or preferences to distort what they claim science has to say. For an excellent example of this, check out the story of Lysenko and Soviet research on genetics. For decades Soviet science was compelled to believe that environmental influences could be inherited.
Science facilitates progress through the crucible of market tests. Science without markets facilitates stronger authoritarianism.
Posted by Jay P. Greene 
Jay P. Greene's Blog 