If we assume that talent is evenly distributed throughout the planet, that the U.S. population is around 300 million, that the global population is 6.7 billion, and that 1/5,000 people are the top candidates to push U.S. innovation forward, that gives us a pool of 60,000 people in the U.S. and 1.28 million outside of it.
Innovation will not be spurred solely by giving those 60,000 Americans access to math or science education, but by providing the right incentives for them to enter the scientific and technical professions. More importantly, we could radically increase the number of innovation candidates through targeted immigration of the 1.28 million people that hail from elsewhere.
While this analysis isn't classically correct (America has benefited from immigration for much of the 20th century), it is compelling and provocative.
It is much too difficult for talented overseas engineers and entrepreneurs (even from Canada and the UK, let alone China or India) to come here and unleash the massive value creation they yearn to foster.