[jjoonathan]

CNBC sure had to stretch that definition pretty darn far to get to "more than half," didn't they.

In any case, immigration is the least upsetting globalization policy IMO: it is exactly what it says on the tin, almost everyone agrees that too little or too much is bad, and the amount can be tuned using quotas. H1B, however, is not that. It pretends to be about scarce skillsets (which enter this calculation one way) when it's actually usually about standard skillsets (which enter the calculation in another). Further, it suppresses the bargaining ability of the people subjected to it, depressing wages for both them those they compete with. Worse still is "regulatory arbitrage" where low tariffs encourage companies to bypass environmental, safety, and labor regulation by shipping manufacturing (or what have you) overseas. I prefer immigration to both H1Bs and regulatory arbitrage, and not by a small margin.

As a more general point, exponential growth is one helluva drug and truly does create a "floats all boats" environment, but different economics apply once an economic sector plateaus. Those economics look much closer to zero-sum. Zero-growth = zero-sum. Low growth is low sum. That's why what happened to American manufacturing over the last 20 years is a good model for what happens to American software over the next 20, if we aren't careful.

We've had it good in software, but we're no longer in the phase where we can pretend the sigmoid is a neverending exponential.