The former includes the type of abstract simplifications that people laugh at physicists for ("consider a spherical cow in a vacuum..."); the latter refers to the naive concept of the scientific method as an iterative "hypothesize, experiment, repeat" approach.

Physics sits far in the "easy" corner, which is why it's the shining city-on-a-hill ideal of scientific success. Cosmology and the like are way down the "non-intervention" axis, where testing is awkward. Down the context-heavy axis are things like biology, studying systems that can't be neatly decomposed.

In the far "difficult" corner, both analysis and testing are sufficiently difficult that constant vigilance is necessary to prevent the entire field from being overrun by convincing-sounding bullshit. Evolutionary biology is the only obvious success story I can think of in this quadrant, and even that has been intermittently plagued by dubious ideas (e.g., the "group selection" fallacy).

To take something like "management science", and do actual science is not just hard, it's fantastically difficult and requires a level of clear-thinking rational discipline far beyond the demands of most fields.