It would be really odd to assume that rpg designers have made no progress over the decades, other than creating new games. There have been refinements and experiments, some of which have been tossed, some kept.
The original D&D rules are rough, and even AD&D has rules mostly ignored, not because of preference, but because they are cludgy or unworkable.
Game designers have a much fuller tool box, and generally a better feel for what will work smoothly. Game design requires a knowledge of math and statistics. Grasping the best formulas for that tricky balance between fun and a workable ruleset requires practice and knowledge of what came before.
Theses ideas can shift around depending on what's in vogue, and of course it's not a smooth straight line upwards with every newly published game being better than earlier ones. Generally, however, games have improved. Designers are learning and building from previous knowledge, widening the field, and polishing techniques. This is what we do, whether it's art, technology, or RPGs.
For what I'm talking about, whether art has advanced or not is not material. I can stick to discussing system-as-technology, and I'm fine, thanks.
I'm rather sympathetic to
@GreyLord's comparison of RPG design to art rather than technology.
There's no doubt that contemporary artists have a "much fuller tool box" (to use Arilyn's phrase). But that doesn't mean that it is beyond argument that art is getting better. Probably my single favourite piece of visual art is Michelangelo's Pieta. I'm not sort of critic, and so don't have a very rich or sophisticated vocabulary for trying to explain why. But one thing that is striking about it, for me - especially in comparison to many of the other statues in St Peter's - is how "modern" it looks. Eg for me it felt much closer to Rodin in the way it conveys emotion, than to many of the other early modern pieces it is surrounded by.
Perhaps my single favourite piece of music is Wagner's Parsifal. Wagner had to invent new technology (instruments,methods of orchestral orginsation, etc) to make his music possible, but it's not the technology on its own that explains the quality of his work.
In RPG design there's no doubt that knowledge of a wider range of possible techniques - which are developed over time - can help. But it's not enough, and not all those techniques ultimately turn out to be satisfying. Just to give on example which obviously will be controversial - nearly all aesthetic judgement is - but is true for me: the RPG technique of
GM authored plotline combined with
downplaying the role of resolution mechanics to produce an experience of
the players working their way through the GM's fiction began to be popularised in the 80s and perhaps reached its peak in the 90s. But for me, the development and refinement of that particular technique did not improve RPGing.
The impact of this technique on individual games did not improve them either - eg in the first (1977) version of Classic Traveller the description of the Streetwise skill seems to me to clearly contemplate the fiction being authored in response to player action declarations for their PCs (ie a successful Streetwise check can result in the GM narrating the existence of such-and-such a useful contact); whereas in the revised (1981) version the skill description has been altered to reflect the emerging "consensus" that the GM authors all the fiction and that the best a skill check can do is give the player's access to fiction that exists - which means a Streetwise check will always fail (even if numerically successful) if the GM has already decided that no contact of the relevant sort exists.
I don't think that first version of Traveller is perfect, and in our campaign we have supplemented it in various ways (new skills from some of the later books, and correspondingly revised and slightly more generous PC generation tables), but I think it is better in some ways than the 1981 version and far superior to MegaTraveller in nearly every respect. (I'm not across the contemporary Mongoose versions but to the extent that they adopt uniform target numbers for skill checks I feel they may have lost some of the strengths of the original.) Whether through luck or genius, it design fastened on many aspects of satisfying RPG play - good spreads of probability, structuring task resolution so as (in most cases) to achieve conflict resolution, recognising the interplay of participant input and mechanics in establishing the fiction - that many later games do not have. (Eg I would say Classic Traveller c 1977 is much better on out-of-combat conflict resolution than D&D 5e c 2020).
Of course it goes almost without saying that the editing and presentation of Classic Traveller could be improved - but even in that sense the has not been a steady upward trajectory for RPGs. While Gygax's AD&D books, especially his DMG, are a bit wonky in their editing/compilation, Gygax's section in his PHB on Successful Adventures makes for much clearer "how to play" text than many other later RPGs (certainly clearer than the corresponding text for 4e D&D despite the latter's 30 year advantage).
Another great RPG, which bucked the 80s/90s trend and is incredibly light yet powerful both in PC build and action resolution is Greg Stafford's Prince Valiant (1989, I think). I'm not sure what more contemporary system compares to it for a combination of simplicity, clarity and power.
None of this is to deny that some new techniques are good ones, nor that improvements in understanding make it easier to get good design through deliberation rather than luck. But I agree with GreyLord that a comparison to purely utilitarian technology like the automotive functioning of a car in not really apt. Modern sculptors have many more tools available to them than Michelangelo did, but can they create something like the Pieta?
