The general idea here is easy enough to get behind!
I think the whole “simple as possible but not moreso” rule will have varying mileage for different kinds of work.
It’s often a very good idea to isolate different terms of your calculations into discrete values: Having a temporary variable for some part of a transformation on one line, and another variable for a different stage of transformation on another line. And yes, this is an artifact of limited human working memory, BUT it also makes debugging so much easier. I welcome a compiler/transpiler that does the algebraic substitution/reduction on the production code; but I don’t apologize for keeping the source code in a long form.
Our job as engineers has been made more complex by transpilation. We now must think about how to write the most readable, easiest-to-maintain source code, and now we MUST understand what the transpiler is doing to our code.