With magic, you can turn a frog into a prince. With science, you can turn a frog into a Ph.D and you still have the frog you started with.
That's the conventional view of Roundworld science. It misses a lot of what actually makes science tick. Science doesn't just exist in the abstract. You could grind the universe into its component particles without finding a single trace of Science. Science is a structure created and maintained by people. And people choose what interests them, and what they consider to be significant and, quite often, they have thought narratively.
Narrativium is powerful stuff. We have always had a drive to paint stories on to the Universe. When humans first looked at the stars, which are great flaming suns an unimaginable distance away, they saw in amongst them giant bulls, dragons, and local heroes.
This human trait doesn't affect what the rules say, not much, anyway, but it does determine which rules we are willing to contemplate in the first place. Moreover, the rules of the universe have to be able to produce everything that we humans observe, which introduces a kind of narrative imperative into science too. Humans think in stories. Classically, at least, science itself has been the discovery of 'stories', think of all those books that had titles like The Story of Mankind, The Descent ofMan, and, if it comes to that, A Brief History of Time.
Over and above the stories of science, though, Discworld can play an even more important role: What if? We can use Discworld for thought experiments about what science might have looked like if the universe had been different, or if the history of science had followed a different route. We can look at science from the outside.
To a scientist, a thought experiment is an argument that you can run through in your head, after which you understand what's going on so well that there's no need to do a real experiment, which is of course a great saving in time and money and prevents you from getting embarrassingly inconvenient results.