As I organize the history of programming, I think the OP's analogy is right. I remember that in the early days, code that just worked as long as it was implemented was called 'cowboy coding.' These days in the US, it's not considered a good coding practice, but there were many programmers who missed it.

During that time, many star programmers emerged. The legends we know today. When cowboy coding started causing problems with collaboration, rules had to be created, and people wrote those rules. Whether it was creating something from 0 to 1 or various other areas, stars seemed to emerge everywhere.

But as time passed and businesses became more sophisticated, the early implementations piled up layer after layer, and human cognitive limits started to show. It seems like it's becoming increasingly difficult for extremely talented star programmers to emerge in specific fields.

In reality, everything—networking, databases, browsers—has become so much harder than in the old days. When I first learned computers, it was opening Notepad, creating an HTML file, and loading it. These days, people start with Python.

So as an industry becomes more advanced, it seems harder for a star programmer who can control everything alone to emerge.

Star programmers are appearing in the AI field right now because it's a new, unexplored territory. They implemented a lot early on and started creating things from 0 to 1, which excites people. But no matter how much I think about it, I don't have the ability to create from 0 to 1. I can't even take something from 1 to 100. I can only combine things at the 20–30 level to make a single product. That's why I think it's hard for me to become a star programmer.

I also want to become a cowboy somewhere, so I've been wandering through an empty pasture, but I don't know where that place is