There is so much talent among the current seniors who applied for the grad school - at least those who applied at Harvard. If you consider the professional growth of a physicist, the decision to accept someone to the grad school may very well be the most selective step. The members of the committees are used to see that most of the applicants are recommended as top 1%, and in many cases it is even true. And it seems that even many people described as 1% will be rejected by many schools. It's kind of crazy.
I wonder whether there exists a more efficient way to organize the brainpower and solve most of important open problems in physics within a finite time scale. I really mean to solve important problems that people want to solve, not merely to invent a meaningful problem for everyone (and certainly not just to convince other people to work on the research direction started by you in order to improve one's ego). I can imagine that sometime in the future, the scientific community will be organized in a more diverse and interactive fashion.
What do I mean? There will be people who will be specialized in checking papers. There will be people whom you will contact to solve particular problems that they may be good at solving - for example, some of them will be ready to quickly answer some of your questions that require to use computers. Some scientific jobs will become more mechanical. In the recent era, people were forced to become rather specialized because the body of knowledge and skills that are important for the current science and technology is rather large. But they have not specialized in their methods, and my guess is that it will eventually happen much like in Ford's company in the 19th century.