Well first you have to come up with at-least two theories that _both_ explain the phenomenon. Then you apply the principle to choose one that's "simpler".
You cant do a 'premature' optimization and not even attempt to piece together a theory that sounds complex, when maybe that is currently the only theory that explains the phenomenon.
You cant do a 'premature' optimization and not even attempt to piece together a theory that sounds complex, when maybe that is currently the only theory that explains the phenomenon.