For 2000 years or so, Euclid's system of geometry was geometry. There was no other system around, and no good reason to doubt its veracity. It had the infallibility of mathematical proofs – as final as "1+1=2" – with the added bonus that it wasn't only an idea on paper, but also referred to direct physical phenomena in (and for) our world. It consisted of logically impeccable proofs, leading any right-minded individual to conclude that, quite simply, there is nothing more to be said on the matter. But with the advancement of various non-Euclidean forms of geometry in the last 200 years or so, while Euclidean geometry hasn't exactly been disproved or dismissed, it certainly has been shown to be inaccurate or fundamentally limited when describing the real world.
I think so-called "neoclassical" economics is the econ equivalent of Euclidean geometry. The traditional base for economics makes some basic, fairly sensible assumptions about matter, production, and human nature, and proceeds to develop a complete science from it. For a certain kind of person, it's extremely satisfying in its careful, methodical advance from a simple, almost one-celled economic model to an entire system that, on the face of it, seems to be very close to ours. Moreover, in its ostensible ability to neatly and concisely explain certain phenomena that may have otherwise seemed impossibly complex to untangle, it has created its fair share of True Believers; people whose eyes have been opened, and who now see everything through the particular prism of this model. As an Econ major who studied with some of these people for 3 years, I can attest that that wonderful phrase "when all you have is a hammer, everything looks like a nail" fits them perfectly.
The neoclassical economics offers much for us to learn, but when it comes to its interaction with the real world, there are warps, often significant warps, both in its descriptive ability – to accurately explain what is happening – and in its normative capacity – to convey to us what ought to be done. In many respects, it amounts to armchair philosophy. Merely making a succession of seemingly solid, common-sense assumptions and deductions, and then proceeding to assert that this is how the world works, doesn't fly. The complexity of the system as a whole, as well as that of the individual human animal (as well as firms, governments, etc.), means that results will sometimes fly contrary to what an intellectual act of solid deduction might tell us.
So what should be done? Well, Karl Popper's quote in the last post below put it about as well as I could hope: Be humble about this. Continue revising (and even discarding, if need be) our conjectures, and test, test, test, using the enormous – and growing – amount of data and computing power at our disposal. Most importantly, we need to retain what Popper referred to as the "critical attitude" (also described in the same quote below).
The significant advances, especially over the last 40 years, in statistical and data analysis techniques, in research on human psychology and choice theory, game theory, information asymmetries, irrationality, etc. have greatly expanded our analytical tool box. If one chooses to ignore these facets – as many continue to do across the economic spectrum, and in popular discussion – one will invoke a Lego-type world, not the real one.