One More Negative Data Point on the Quality of the H-1Bs?

Seldom does a serious, number-filled blog cause me to laugh out loud, but this occurred today when I read today’s “gotcha” post in RealityCheck, Alan Tonelson’s outstanding blog site on economics and foreign policy.

Alan deftly juxtaposes two diametrically opposing views — by the same person:

“Trump’s crackdown on H-1B visas could prevent the next US unicorn born of Indian immigrants”

–Ananya Bhattacharya, Quartz India, February 5, 2017

Share of Indian engineering graduates capable of writing “the correct logic for a program, a minimum requirement for any programming job”: 4.77%

–Ananya Bhattacharya, Quartz India, April 20, 2017

Outrageous and funny, yes, and yet another reason to stay informed via RealityCheck. By the way, some years ago, McKinsey released a report finding weakness among the Chinese graduates as well.

But let’s take a closer look at these issues. Much of what I will say here will pertain to the Indians and Chinese, as they form the two largest H-1B groups, but I will first speak more generally:

As some of you will recall, my EPI study showed that the quality of foreign students (of all nationalities) in CS at U.S. schools is somewhat below that of their American peers. Measures included patenting, work in R&D, selectivity of U.S. institution attended and ACM dissertation awards. I also cited similar research by others.

Now looking at the Indian case,  first note that technically Bhattacharya could be right on both counts. Presumably the weaker Indian graduates don’t ever come to the U.S. or even work for the major Indian firms, and even if the overall quality were weak, there still could be some unicorn formers among them.

On the other hand, a little-known fact is that the Indian IT firms actually take mediocrity as their business model. That minimizes labor costs, and their view is that by partitioning a software project into small, well-defined components, the level of complexity of the work has been reduced to the point at which one does not need programming geniuses to get the job done. (See my article in IEEE IT Pro for references and counterarguments.)

Even at the Indian Institutes of Technologies, India’s “MITs,” with their extremely high admissions standards, the curricula are pedestrian and many professors are not strong (the pay is way too low to attract high-quality people). In addition, things appear not to be nearly as intense at the Indian schools. A survey done a few years ago by a researcher at Stanford asked CS students at Stanford, Cambridge and IIT how they spent their time. The IIT students had far more leisure time than did their California and UK counterparts, who were going through extremely demanding workloads. Mind you, I do think that there is far too much intensity at the U.S. and UK schools in CS, so I am not faulting IIT at all. But it does affect the quality of the new graduates, even though the IIT grads, all top intellects, do eventually catch up.

I would also be quick to point out, as I have often done over the years, that the average level of programming skill among (U.S.) graduates of U.S. schools is not so strong either, even with that intensity. The ones who really love CS are fantastically good, but they are probably only about 15% of the U.S. grads, depending on the school.

Now, what would India and China say to the above? India might point to the success of Indian immigrants in the U.S. at the executive level. The CEOs of both Google and Microsoft are from India, they would point out. Unfortunately, business acumen is not my field, so I really have no comment there, except to say good for Mr. Pichai and Mr. Nadella, whom I presume are indeed tops in business management.

What about the Chinese? They might point to the success of Chinese universities in the ACM Programming Contest and the Top 500 list of most powerful supercomputers. Sorry, but I don’t consider either of these to be of any relevance to the issues here, or for that matter, of any practical significance.

The programming contest is exactly the kind of thing a developing nation like China needs to gain exposure on the national stage — attention-getting, and an acheivable goal if money and resources are devoted to it — but basically meaningless. As I wrote a few years ago for Bloomberg,

The Jiaoda [China’s Shanghai Jiaotong University] contestants are essentially student-athletes, spending all their time training for the event, according to a Jiaoda public information officer, Xu Jun. And the skills needed for the competition are indeed trainable. Although the problems posed each year are unique, their solutions usually fall into a handful of mathematical patterns.

This gives a huge benefit to those who can devote themselves to full-time, year-round practice. By contrast, most top U.S. computer-science students have better things to do with their time, including founding startups that might become billion-dollar companies.

China’s performance in that contest has no bearing on its ability to come up with the next killer app.

For the supercomputer list, my analogy has been to building the world’s tallest skyscraper. Given enough money and will power, one can always put up a new one taller than the last one. Again, for countries that feel the need to prove themselves on the world stage, such as China, supercomputers might be good investments, but otherwise they are simply the result of throwing a lot of money at some very narrow applications.

And finally, what about those killer apps? Skeptics argue that India and China have yet to produce any, and frankly I agree. China might point, for instance, to WeChat, which does messaging, point-of-sale payment and the like. The Economist ran a long article praising the remarkably widespread usage the app enjoys in China, and I myself am a WeChat user. But there is nothing innovative about it.

So who was right, the Ananya Bhattacharya of February 5 or the Ananya Bhattacharya of April 20? Neither Pichai nor Nadella is an entrepreneur, much less a unicorn former, so they don’t fall into the February 5 category. Perhaps readers can supply some better examples. On the other hand, I assure you that the H-1Bs, though generally not “the best and the brightest,” deserve more than what is implied in the April 20 remark.