https://third-bit.com/2018/11/03/abstraction-comprehension.html I think this is very insightful . Same code but different presentation based on user preferences

Eve like partition of codebases into multisets with each set having a projectional view of its own or the ability to write a shader to visualize the underlying computational structure depending on the user type / context would be a powerful feature to leverage in an IDE.

An interesting challenge here is that information will get lost between presentations. It would be nice to allow people to write code in their preferred presentation, but perhaps prompt them to add information that will aid automatic generation of other presentations. The

temp

variable in the example stuck out to me for this.

I don't think "high-level" and "functions of functions of functions" are synonymous as the post suggests. In an array language, element-wise subtraction of two columns as "income - expenses" doesn't add any significant cognitive load.

That post raises an interesting point, but I not a new one. The exact same problem has been around for decades concerning the use of math in science and engineering. An example from physics education is the old but ongoing debate on how to teach quantum mechanics: in terms of concrete representations (wave functions) or in terms of abstract representations in Hilbert spaces. It really depends on the prior knowledge/experience of the students, and that's something a teacher can't even know before the start of the course. So I think that programming and its teaching can learn a lot from maths in this respect.