The elephant in the room isn't why aren't more universities accepting 'local' students. No department would choose a foreigner over a local. You wanna know why? How much the student is going to cost them. Foreign students never qualify for in-state tuition (well at a state school anyways) and right off the bat are much more expensive to their advisor's budget. Local students can also apply for NSF Graduate Fellowship, and bring in their own funding, something most foreign students can't. Both of these factors make it so that the first preference is to hire local.

That said, in a large number of departments (especially in CS, and other sciences) you see a burgeoning international student population. Why? There just weren't enough decent local applicants to go around and science needs its foot-soldiers.

That's not really true. The basic problem is that while a foreigner can take a US PhD and apply for a fairly prestigious job in their home country, an American with a US PhD gets their career stuck in a rut. Multiple post-docs, for low salaries and long hours, followed by a desperate rush to find a job in a national lab, an industrial lab, or academia. There are hundreds of applicants for each permanent job, and even if you get a permanent job, you're not really guaranteed job-security (tenure). Many "permanent" jobs are even on soft money, which means if you fail to bring in enough grants (and grant rates have fallen dreadfully low in recent years due to Congressional budget cuts), you're a "professor" who can't pay for food or housing.

American academia relies too much on having legions of cheap grad-students to do much of the footwork, and then throwing those "students" out of science entirely when they graduate. It also relies on Congress actually funding the NSF, NIH, NASA, etc. at rates proportional to how much science is being done, which of course nowadays Congress steadfastly refuses to do.

American science careers are in crisis because the career model has become based upon exploitation: getting as much cheap labor out of as many people as possible before throwing them away.

The keys are, above all, good recommendations from respectable academics and some kind of research experience followed by high undergrad grades. A high GRE score can't substitute for any of those.

Of course, the foreigners can manufacture the grades and recommendations with a little cash. Maybe even the research experience. They probably do have to study English to pass the standardized tests: The one that matters for them is the TOEFL.

Wow. Do you have any source for this? I would love to hear one. DO you have ANY idea how difficult it is to get an application accepted at a top tier grad school? Especially if you are an international student? We have to compete against millions of our own country folk at our local examinations, which are way more difficult than the american equivalents. Google IIT-JEE for a starter, and probably try to solve a few sample questions. We work hard to get to where we are and don't expect/feel entitled a job/education just because we are born in a certain country. No one is conspiring to keep local students out, it just happens to be that when graded on a uniform scale (set by the universities themselves (Not every foreign institution's scores are accepted)) the international counterparts are better. With respect to my credentials and experience, I was a grad student from India in one of the tier one schools and work as an architect in the Big 4. You see a huge international presence here in the big companies frankly because we compete at a completely different scale than what you would like to think. I have interviewed enough people to not judge them based on where they come from or how well they can talk english.

It's much harder in America because it doesn't matter how you do on the examinations. Your fate is decided by personal opinions and relationships and what your local school thinks of you. There is no simple IIT exam or A-level or Baccalaureate or Sooneung where you can just study hard, write a great exam and get admitted.

How we dreamed of the ease foreign students had with their exams! Instead it's a deep dreary slog of obsequiousness, obedience, busy work, group projects, supplication, and being seen but seldom heard. And the results are secret evaluations you're never entitled to see and which are never documented in public.

But that's all for undergrad.

The fact remains that the only exam that matters much in top tier grad admissions for US schools is the TOEFL and that's only for foreigners. Documented research work experience is much more important than any test.