I would like to formulate “lex parsimoniae” as “less the assumptions less the liability”. Its not so much about simplicity or complexity but how simplicity or complexity are entertained by assumptions. Do we want more liability or less? Obviously less, and assumptions are a liability. They are only to be taken to make a matter more accessible and not less.
So assumption is inversely proportional to accessibility and directly proportional to the invalidity of a theory. We do not assume there is one sun, we know there is one sun, there is no liability here. But “for any theory” if we assume “two suns” thats a liability because thats an assumption. So add another assumption, that, the one sun is small in size and another bigger, than say our planet. Then thats another liability on our theory vs how valid it is to reality.
We are not proving a theory to be more real by adding more complexity in this sense. It has ultimately to be subjected to tests of reality and thats where lesser assumptions have therefore a bigger chance of success as a good theory. A bad student does not become better at passing his tests because he has more and more of the tests. Its the opposite.
A good theory will pass the tests even if there are more tests, but its a subjective manipulation to let theories be squandered in abstract delusional tinkering. If a good theory is not good, it will fail sooner than not. So why give it more and more test. A bad theory without “subjective manipulations” anyway is clear enough to be bad. So why, more tests. This is the zest of “lex parsimoniae” or as I would call “cut it out”. Its also called as Okham’s Razor. “Among the hypotheses one that carries the least number of assumptions is the one to be chosen.”