This debate comes up again and again, and any time one view is purported strongly over the other we just get education that is half-assed. This includes people

This debate comes up again and again, and any time one view is purported strongly over the other we just get education that is half-assed. This includes people with your point of view that deliberate attempts at committing propositions and procedures to memory should be avoided. The issue is one of cognitive load. You can't get to mathematical thinking and appeal to both a higher-level "intuition" or "key idea" if you aren't fluent in the details. Else it becomes hand-waving that doesn't cash out to proof-ready reasoning. Likewise, you can't internalize the details if you don't have a principle which compresses them and lets you unpack them appropriately for the circumstance of the problem. This is fair, but first there has to be something to compress. The question is, "How much detail can be advanced at this point in the student's growth?" If the key issue is to dispel the notion that mathematics is arcane, you should narrow down the scale of the problem so that you can quickly see how the details contribute to generalizations, and how a procedure can be taken from it. This achieves what you are looking for with less effort and less risk. With that point of view illustrated with a simpler case, it can be developed or built upon with bigger cases. Bigger cases inevitably means more details to internalize. This protracts the phase where memorization would be appropriate. But the student, having seen how mathematics is developed in a simpler case, would understand that these details have a context that will be more adequately realized at a second or third pass of the material. Eschewing the internalization of detail as you are purporting would encourage a sense that math is subjective, because the high-level impressions of mathematics would not have the memories of detail to back them up, allowing for more to be said, with less control.