Same goes for RPG's. They are games. As much as people want to pretend that they aren't, they still are. And games have math, like it or not.
I don't see how this follows. Yes, an rpg is a game, but many games don't have math. Notably, the games than tabletop rpgs have the most in common with, such as children playing house or people of any age engaging in various forms of mock combat, don't have math. In fact, most games don't. Sports don't. Tag doesn't. Improv games don't. Make-believe is a game. Rpgs are a subtset of that.
One could certainly play an rpg that involved no math. As such, I'd argue that the math that does exist in D&D is not inherent or intrinsic to the game, and the math doesn't reflect the game, and changing the math doesn't fix the game.
If it has math, then we can apply systematic changes to the math to create better rules. And, yes, I do say better. A 1st level thief with a 15% chance to open locks is not as good of a rule as a 1st level rogue with an Open Locks of +8.
Oddly enough, this is the perfect example of why. Yes, a thief with 15% lockpick is bad, but not for any mathematical reason. Mathematically, the character works fine. 15% is a perfectly valid number. However, the character's ineptitude is likely to bog down a game session and may not represent the underlying concept well enough. What level of aptitude is appropriate to dispense is a subjective decision.
I'd say that as time has gone on, we've become much more systematic in our approach to game design. There's a lot less "throw things at the wall and see what sticks" approach to game design. So, yes, I'd say that science has certainly become a much more common approach to game design than art.
This analogy struck me because essentially the same thing happened in medicine; the pharmaceutical industry has moved away from testing folk remedies and randomly assaying exotic plants for useful compounds and towards targeted drug design, wherein computer modeling is used to shape a drug to match some molecular target and the result of the modeling is then synthesized in a lab.
Interestingly enough, even though this approach sounds really science-y, it has largely failed to produce useful drugs for humans, because we are too complicated for computer modeling to work to this extent.
But, OTOH, you can certainly make arguments for advances in materials used in fashion. It's not like they had nylon 100 years ago. There are things you can do in fashion today that you absolutely could not do before.
I do think that even where progress is not objectively measurable, it as happened. Not in a straight line necessarily, in textiles or in game design, but I do agree that there is a general impetus towards things getting better.