I'm sorry, but the problem is not the talented individuals or their efforts. Its that fundamentally new and different ways of computing and approaching computing problems arise as technology continues to progress.
We can sit here and grouchily state that reimplementing UNIX is the problem, but even within the ebb and flow of Linux we see substantial change and reformation over the arc of 5 years. And that's ignoring the actual GUI toolkits which have been in a constant state of flux and only partial levels of functionality.
These are challenges that the community has been happy to dismiss even as they create increasingly obvious and increasingly difficult barriers to entry in the community. They do so because they do not inherently feel the problem as acutely, they're the beneficiaries of education and opportunity (or more succinctly, privilege) that they are happy to dismiss as something that anyone could have. It is not so, but try telling them that.
I think that is a part of it, for sure. But even in first world countries like America you see lots of kids with inadequate nutrition, no access to modern education, and no cultural inculcation. And of course, that sort of pretends IQs themselves don't fall along a normal distribution and that there aren't gender and racial issues discouraging a large class of people from pursuing an education in this field.
Even if these issues were addressed socially, I think economic barriers are hard to ignore.
They will join the heritage of Donald Knuth, Ritchie, Ken Thompson, RMS, ESR etc etc. They existed in a time when there were thousands of engineers writing very complex systems for large companies. All those systems are gone, never to be seen again. Useless. gcc, emacs, vim have their source code in thousands of git repos around the world and are used daily all the time. The next step will be possible!