What happened that got economists backwards?

Time, the same thing that got physicists backwards, time and divisibility. There is no such thing as time which requires infinite divisibility. Almost everything is container ergodic, the container size in economics, the unit sphere in physics, is the stationary item.

Economists have seasonal adjustment, true, but otherwise almost the entire system, like physics is minimal redundancy networks. In the proton it is the spiral stepping  constructed from Nulls, and managed by unit spheres.  In economics is is the streets and checkout counters. In both cases it is finite bandwidth and follows the equations for finite entropy systems.

Consider this abstract from economics:

In this paper I employ Imre Lakatos's methodology of scientific research programs to scrutinize the idea that stagflation in the 1970s falsified the Keynesian research program. I point out that Keynesian models were able to account for stagflation once they included inflation expectations, so the essential tenets of the Keynesian research program are consistent with the would-be anomaly of stagflation. Furthermore, Keynesian economics exhibited both theoretical and empirical progress by evolving in a way that rendered stagflation a logical consequence of Keynesian assumptions. The transition to new classical economics did not yield such progress. Also, as Keynesian economics tends to adopt novel findings and research methods, new classical economics does not have excess theoretical or empirical content relative to the Keynesian research program. In summary, I find that the falsification of the Keynesian program is unwarranted.

Performing expectations requires near complete knowledge of the past, and that is an infinite bandwidth system. He has gotten himself into the time domain, and he does not have locality. This is literally impossible, especially in economics. So he produces a work of fiction simply to maintain some fiction from some English dandy of 90 years ago.