There is nothing wrong with striving for excellence and rewarding excellence. In fact I am all for it. But the problem is that we tend to reward only the highly exceptional, leaving no room for the rest that are still excellent but not "exceptional". Those exceptional few that make it to the top then justify it that they are the only ones worthy of staying in the game, and reinforce it for the next generation.
@dalias That's what more and more research grant/award panels are trialing. Studies actually show that there is no correlation between reviewer scores/ranks across reviewers (above a certain threshold), meaning that if you meet the baseline criteria reviewers don't do any better than random!