Right, I was just trying to bring up the point that however the code is written, what really matters is, how easily humans can understand it.
Now maybe there is, or can be an algorithm that takes a piece of code and spits out a number saying how easy it is for a typical programmer to understand it correctly. That would be the measure of "cognitive load".
If we just speak of cognitive load without really specifying how to measure it, we are not where we would like to be.
How do we define, and measure "cognitive load"? It is an easy word to use, but how to measure it?
And how overcomplicated the code is has a huge (overwhelming everything else) influence on how hard it is to communicate what it does.
So in practice it's fair to call it the property of the code, even if bad documentation or mentoring can make things unnecessarily worse.
Solar system can be modeled as geocentric with enough epicycles, or as heliocentric system with eliptical orbits.
One of these is inherently easier to communicate.