I'm not reading a DFW quote because fuck that guy but the serious answer to your question:

Or are the inputs and outputs so multitudinously abundant that it's impossible to measure?

This gets to the problem of modeling. I've always been a cynic of statistics because a successful statistical model is predicated on accounting for all the relevant inputs and outputs in order to get a meaningful answer. In engineering, this is easy when you have a broad basis of data from which to model, and tricky when you don't. Take airplanes: we know the stress characteristics of steel in a boundless bounty of conditions because we've been using it and testing it for 150 years. Aluminum is a little trickier because it's a more reactive metal and has no yield strength (you can't bend it back and forth without fucking it up) so initially, aluminum components failed more unexpectedly than steel components. Things got really bad when everybody shifted to carbon fiber and other composites because the makeup of composites is subject to much greater variation and can be tested in far fewer ways. So your model of steel reveals few surprises, your model of aluminum reveals a few more, and your model of carbon fiber reveals more. But then, the more of these systems you have in the world, the fewer uncertainties you have in your system and the more likely it is that your model accurately reflects the problem.

The issue with economics is that you can introduce things like mark to market accounting and suddenly every assumption is challenged. An investment bank can come up with the idea of collateralized debt obligations and have them on the fucking market within a few months. We're here in the land of Zero Interest Rate Policy which is, in economic terms, the equivalent of saying "what if gravity were positive rather than negative?"

So in part, they're blameless because modeling a plane (or a mouse, or a supernova) has pretty good boundaries where you need to model and a good sense of where you don't while modeling an economy is a giant loosey-goosey unhappy no-fun zone. But in part, they're fully to blame because economic "science" is founded on the notion that you can model some aspect of that giant loosey-goosey unhappy no-fun zone in order to learn something valuable. And they take that assumption and extrapolate it out to the entire goddamn human race.

An aircraft manufacturer will get an idea that maybe jet canopies could be made of a stronger material and, along the way, invent crazy glue. An economist will fail to notice that they're making aircraft out of crazy glue.