Ocam's razor is a philosopher's tool to make their lives a little easier - beyond trying to organise hypotheses into their most likely order it doesn't and can't solve a problem in science. Science needs evidence and the simplest solution is not a requirement for the correct answer.
Well, it depends. Insofar as problems can be solved in science (i.e. provisionally) sometimes Occam's razor does in fact solve them.
To take an example I often use, suppose I leave my dog alone with a slice of pizza, and when I come back the pizza is gone. I might conjecture that the pizza was stolen by pizza-stealing fairies, but this involves introducing an entity otherwise unevidenced, whereas I already possess evidence for the existence of a pizza-eating dog. There is no need to add a hypothetical entity when I have a real one to hand.
Now if I had no evidence of dogs either, then it would be impossible to say which hypothesis was simpler. But since I do, then the one without fairies is simpler in a clear and objective sense --- the universe with dogs but without fairies is a proper subset of the universe with dogs
and fairies.
What is essential here is the concept of a proper subset, since this makes it
epistemologically significant that one hypothesis is simpler than another.
Other forms of simplicity are not significant. If I hear a crash when someone throws something through my window, then a perfect sphere is simpler (in the sense: requires less information to describe) than a jagged rock, but this is not a reason to exalt the idea that the object was perfectly spherical to the status of a most likely hypothesis.
™ Version 4.2