Now who's shifting the goalposts?
Are you seriously claiming that the first time your character tries an untrained skill is the very first time that character has EVER attempted to do something like that? The very first tree you roll an Athletics check on is the very first time your character has ever tried to climb a tree? Or jump across a space? Or attempted to jimmy a lock? Or tried to remember what kills trolls? Really?
Gretzky was 17 in his rookie year. He was playing people who had ten, fifteen years more experience playing than he did and was still completely owning them. If training>natural talent, shouldn't those far more experienced players be so much better than someone with a minimum of training (ie. 1st level) and natural talent?
The overwhelming majority of skill checks are certainly not the very first time that character has used that skill. Might be the first time at the table, but, first time in that character's life? Very unlikely.
Besides, wouldn't we use the Tool proficiencies for sports players? As you said, there are a number of skills at play here, so, wouldn't it make more sense to just be proficient in Hockey? Same as I can be proficient in a musical instrument? I don't even NEED to spend character resources on that - I can buy that through time.
I have no meaningful horse in any of the multiple races currently being run in this conversation. However, comparing people who have "ten, fifteen years"
professional-level experience to people who have "zero" is something of a mistaken analysis here. Like trying to claim that George Dantzig exhibited pure "talent" in providing proofs for two major open problems in statistcs (the Neyman-Pearson lemma and the non-existence of a certain kind of test of Student's hypothesis) while still a "student." It is, of course, partially due to his natural skills that he could find a solution (or, rather, proof) for things that had not yet been solved/proved. But as a man who already possessed a Master's degree and was beginning his Ph.D. studies, it's disingenuous to try to claim that training was
obviously secondary to talent, simply because he didn't have
professional experience as a mathematician.
Is it true that Gretzky exhibited great talent. But even at 17, he had been working for years--just to be
noticed in order to be selected for professional play. According to
his Wikipedia article, he learned to skate just before he turned three years old ("aged two years, ten months") by skating in a rink his father made in their backyard, where he taught Wayne and his friends how to play hockey. According to
an ESPN retrospective on his career he was playing local hockey games by age 6, and was making headlines at age 9. Does that indicate talent? Absolutely. But, together with the other information we have? It indicates that that talent rested on a basis of focused training. Sure, he was besting people who had 10-15 years professional experience ahead of him. He'd been besting people older, stronger, and more-experienced
for eleven years already, and had been preparing for such activities for another three years before
that.
Part of the problem here is that you are focusing on the best, of the best, of the best. And in that arena--where runners and swimmers vie for tenths, sometimes
hundredths of seconds, where weightlifters push the boundaries of human musculoskeletal strength, where cyclists train in high-altitude environments because a slight difference in blood cell size
matters--then of
course you're going to be able to come up with, not merely some, but
many example people whose success is defined by raw talent, for which training is a necessary but secondary addendum. And it's not just physical pursuits; Ramanujan and Mozart provide us with meaningful mental examples as well.
But Ramanujan in particular highlights the reason why talent, no matter how great, can still only carry you so far. He was an utterly brilliant mathematician...who was almost purely self-taught, limited to something like one or two entry-level calculus texts as the sum total of his "higher-level math" education prior to being discovered by the mathematical community (partnering with G. H. Hardy). He still achieved surprising, elegant, fascinating results, but most mathematicians today agree that if he had had access to a comprehensive mathematics education he would have gone even further and done even more. Does that mean "talent" isn't a thing? Hardly. But it does mean that training definitely has a major impact on the
reach and
breadth that talent can achieve.
It's also something of an academic conversation, don't you think? I mean, the best you can get with point buy is a 17, and that only if you play a non-human and hyperfocus on your racial +2. Most PCs are only going to have a +3 in their highest modifier. Does that actually correlate with being "a Gretzky" or "a Ramanujan"? Or, hell, even "an Einstein"?