http://en.wikipedia.org/wiki/The_Structure_of_Scientific_Rev...
His finding was that science works by accretion of facts, except that over time, a number of facts that don't fit the established theory build up. Eventually, this reaches a crisis point where you realize that the current models are unable to explain reality, and then they get thrown out (a "paradigm shift") and everything needs to be figured out again from first principles. At this point, younger scientists have the advantage, because the "knowledge frontier" contracts back to nothing (almost...the new theory still has to account for old experimental results).
The last major paradigm shift in physics was a century ago, with the twin discoveries of relativity and quantum mechanics. You could probably argue that we're due for another one soon - the weight of evidence that our understanding of the universe is flawed has been building up, with unexplained results around dark matter/dark energy/cosmology and the failure of superstring theory to make accurate testable predictions. It's pretty likely that within our lifetime physics may become a hot place for brilliant 20-somethings again.
But in the meantime, most of the action has been in computing. The field of practical applications of computer science has been undergoing a paradigm shift approximately every 10 years (wasn't yesterday's top Hacker News story about that?), hence why we see so many young tech billionaires.