[jfengel]

That's true of the hard sciences as well, not just the soft ones. The types of observations you think to make, the types of experiments you can perform, are "laden" with your theory of what there is in the world:

https://en.wikipedia.org/wiki/Theory-ladenness

Scientific revolutions often start with puzzling "errors" in the measurements, and end up re-defining terms so that the new "correct" theory makes the error go away. But you're still theory-laden: the new theory is itself subject to eventual revolution.

Even really fundamental concepts like mass, length, entropy, and temperature undergo redefinition. When they do, the way we measure them changes. The old way, which was considered scientifically solid and sound, seems quaint and backwards in retrospect.

That's even more obvious in soft sciences, where the terms are even less rigorously defined than notions like "mass". I think it's important to recognize it in the hard sciences as well, because it's easy to get complacent thinking that we have all the terms perfectly defined. Scientists are human regardless of their field and are always burdened with assumptions that they don't realize.