Gould's most valuable conclusion is that the racist rubbish he deconstructs in his book was produced not by crackpots working in the basements of policy institutes of dubious reputation -- but by the centers of analytics excellence like Stanford or Princeton, and complete with all standards of scientific rigor -- Brad de Long notwithstanding.
I think the main problem lies not in what John Kenneth Galbraith once said of the economic profession -- the business of providing needed conclusions to those in the position to pay for for them -- but in self-policing. Corrupt scientists manufacturing needed conclusions can be detected sooner rather than later, because such practice is unacceptable in scientifi community. The problem begins when instead of needed conclusions, scientists start manufacturing standards by which their theories are to be measured.
There is an old joke "operation was a success, but the patient died" mocking the tendency of acadmicians to bend the standards of proof to fit the theory rather than the other way around. That tendency might be apparent in medical sciences because the standards of proof are measured independently of the theoretic constructs. The same might hold for marketing, as someone aptly observed in this discussion. If the product flops, scientific rigor will do little to save the marketer from being fired.
Elasticity of definitions, mentioned in Max's posting in this thread, certainly contributes to that, but the intellectual flaw is much deeper than that. As Gould convincingly shows, the racist theories became immune from empirical proof when they invented the standard of proof of their own - the IQ test whose construction was based on the same theoretical constructs that were sought to be validated by the test. To my recollection, the French philosopher of science Henri Poincare, voiced similar criticism of proof in the natural sciences.
As I understand it - correct me if I am wrong -- very few, if any, standards of proof in economics (especially of the neoclassical variety) are measured independently of the theoretical constructs they are supposed to test. To illustrate, take the public good theory that explains market failures by the fact that for certain goods it is very difficult, or impossible, to exclude non-payers form benefiting from that good. Hence, the theory goes, markets cannot efficiently produce that good (producers would not get paid due to consumer free riding), so the public sector assumes the production and collect fees in the form of compulsory taxes. But the key theoretical construct -- public good -- cannot be measured independently of the instiutional venue of its distribution. Thus, we know which good is 'public' only by knowing which institutional sector produced it. Knowing that, the public good theory goes in circles using that construct to 'prove' why a particular institutional sector produces ceratin goods but not other.
BTW, the structurally flawed public good theory is the key 'problem shift' - or conceptual device that saves the core of economic theorizing from empirical refutation. Without that problem shift the theory is unable to explain why certain good are produced/distributed by different institutional sectors, and without being able to do so, its spiel 'Market, Market uber alles' becomes a mere value judgment.
I think the propensity toward manufacturing its own standard of proof grows with the level of abstraction -- the less odirectly observable the phenomenon to be explained, the greater the likelihood that some form of circular reasoning device will be used to 'prove' the theory that 'explains' it.
So one of the benefits of poor thought policing practices, or lax scientific rigor if you will (or is it intellectual rigor mortis?), is that such circular reasoning devices catch the attention of fellow professionals who mercilessly expose them to public view and trash them. The fate of functionalist theorizing in sociology is a case in point.
Regards,
Wojtek Sokolowski