Dr. Norm Matloff: Foreign Programmers NOT The "Best and Brightest"—Just The Cheapest
March 21, 2009, 03:55 AM
A+
|
a-
Print Friendly and PDF
Dr. Norm Matloff writes:
One of the industry lobbyists' favorite lines is that the H-1Bs are "the best and the brightest" from around the world. Although I do support the immigration of top talents, only a small percentage of H-1Bs are in that league. I've shown this in great detail; see here and here.

But the lobbyists keep pushing it, hence my current posting. Here I'll give you a look at some other aspects, including the background of some of those who are making the "best and brightest" claims.

Let's start with Dan Siciliano, a lecturer in Stanford University's law school. He made the following comments to the Wall Street Journal, June 30, 2006:

Economists worry about another place owning the very next big thing" — the next groundbreaking technology, If the heart and mind of the next great thing emerges somewhere else because the talent is there, then we will be hurt...[an increase in the H-1B cap is needed] to avoid irreversible damage to the economy.
So, what are Siciliano's qualifications for making such strong claims about the innovative quality of the H-1Bs? Here are his "qualifications":

Siciliano was previously an immigration lawyer with Bacon and Dear, one of the most prominent immigration law firms in the nation; Sun Microsystems retained Roxanne Bacon when Sun engineer Guy Santiglia sued Sun after they laid him off while keeping H-1Bs. Siciliano also is CEO of LawLogix, a firm that develops software systems for immigration lawyers. To top it off, Siciliano is on the Board of Trustees of the American Immigration Law Foundation, which is the research arm of the American Immigration Lawyers Association. He is also a research fellow with the AILF. Note by the way that the vice chair of that board is Kirsten Schlenger, with whom I participated in a TV debate on H-1B; the video is here: [WMV]

So Siciliano has a vested interest, to put it mildly. Unfortunately, he doesn't seem to mention these things when he talks to the press. In the above WSJ article, for instance, he was described as "a Stanford economist." When he was on CNBC on March 6, he was asked his views "as an educator," and though his affiliation with the Immigration Policy Center was mentioned, there was no mention of IPC's connection to AILA/AILF.

On that CNBC show, Siciliano referred to my CIS study, the one at the www.cis.org Web link above, and grossly (though not necessarily deliberately) misinterpreted my findings. Here is what happened:

As one of several approaches I took to assessing the industry lobbyists' claim that the H-1Bs are "the best and the brightest," I compared actual salary to legal prevailing wage. If the H-1Bs were indeed outstanding talents, they would be paid well above the legal prevailing wage, so I calculated the ratio of actual wage to prevailing wage, calling it the Talent Measure (TM). Just as I suspected, the TM value was a hair above 1.0 for the industry as a whole, and for almost all prominent tech firms.

Microsoft was an exception, with TM equal to 1.19. Siciliano, who was debating CIS Fellow John Miano on the CNBC segment, tried to refute Miano's claim that H-1Bs are underpaid, saying "CIS' own study found that Microsoft pays 19% above prevailing wage." This of course ignored the fact that Microsoft's 1.19 figure was very unusual, and much more importantly, it ignored the disclaimer in my study:

Typically the employer will cite government data as the source. The legal definition of prevailing wage in both the law and regulations contains major loopholes (see my previous Backgrounder mentioned above), but the industry lobbyists insist that the foreign workers are not underpaid. Since here the focus is on another industry claim, that the foreign workers are of outstanding talent, for the purpose of the present analysis, it will be assumed that the prevailing wage is the real market wage.
In other words, to address the industry claim that the H-1Bs are "the best and the brightest," my approach was to take at face value the industry's claim that legal prevailing wage is the real market wage, and then investigate what they are paying the H-1Bs relative to prevailing wage. If the answer is that they are paying no higher than legal prevailing wage, one must conclude that one of their two claims is wrong—either they are underpaying the H-1Bs or the H-1Bs are not outstanding talents. The TM figures should be used only to assess the "best/brightest" claims, not to determine whether these firms are paying their foreign workers market wages.

Since I and others have found a 15-30% difference between the prevailing wage and real market wage, that 1.19 figure for Microsoft could in fact be as low as 0.90.

Siciliano ignored all this, or didn't read the study he was citing in the first place.

Now let's turn to the recent Washington Post op-ed by Vivek Wadhwa. [They're Taking Their Brains and Going Home, By Vivek Wadhwa Sunday, March 8, 2009]

Unlike Siciliano, Wadhwa has actually done research on H-1B and related topics, and though I've found fault with some of his work, he has done some very good studies. Yet Wadhwa has chosen some odd poster children for his own claim that we are losing "the best and the brightest," ostensibly because they weary of the long wait for green cards.

The claim is unfounded in the first place. The employer-sponsored green card categories used by the tech industry consist of three levels, EB-1, EB-2 and EB-3, in order from most to least talented. The fact is that there have been long waits in recent years only for EB-3, which is for ordinary workers of no special talents. See here.

As mentioned, Wadhwa's examples here don't support his "best and brightest" assertion. His first example, Sandeep Nijsure, attended mediocre universities in both India (per my Web search) and the U.S. (per Wadhwa's op-ed). And though I dislike judging someone by the schools they attended, there certainly is no indication that Nijsure plays in the "best and brightest" league.

Another example Wadhwa offers us is Girija Subramaniam, a test engineer at Texas Instruments. Come on, Vivek—a test engineer???? This is no job for geniuses. So this example doesn't work either.

His third example, Meijie Tang, may be different. She's mentioned a lot in an NYT article, Re-education, By Ann Hulbert,  April 1, 2007 and she seems to be an interesting person. After acing a bunch of China's standardized tests she gained some fame in that country before she came to the U.S. for study. I'm not a fan of those tests, but maybe she does qualify as "the best and the brightest" in some senses. On the other hand, it's not in the technology sense; Tang is an econ major, according to the Web. And look what our best-and-brightest economists have done to the world recently.

So, the Best and Brightest count is two no's and one maybe. Is that the best Wadhwa could come up with? Again, I strongly support facilitating the immigration of outstanding talents, but I don't regard these as the type I have in mind.

Before I go on, I'd like to point out that many U.S. native "best and brightest" are being displaced. Gene Nelson won a National Science Fair award when he was in high school, which arguably makes his case somewhat similar to Tang's. He later earned a PhD in biophysics, but once he got older (remember, even 35 is "old") the jobs started drying up for him—and taken by H-1Bs.

I've mentioned before my former student, whose innovative engineering work at a major name-brand firm got him a mention in the Wall Street Journal, arguably also a best-and-brightest quality. He was sloughed off by the industry a few years ago, around age 36, and after working only sporadically for several years, he finally bit the bullet and switched to another profession.

I've got several readers of this e-newsletter with degrees from MIT who find it hard to get tech work too. Presumably MIT only admits the best and brightest too.

A man who earned his PhD in computer security in my department about 10 years ago is out of work, has been for almost a year. He was a top student academically, had prior industry experience before coming back to grad school, and is personable and articulate. He's willing to take any tech job, and I'm sure he can do many jobs occupied by H-1Bs better than they can.

And let's not forget Douglas Prasher, the almost-Nobel laureate who's working as a van driver for a Toyota dealer. He could do a lot of those research jobs in biotech that are filled by H-1Bs.

So H-1B is crowding out many of our own best and brightest, and causing many of the best and brightest college students to avoid tech in the first place. (Some of you may have seen news articles the last couple of days showing that computer science enrollment is finally up somewhat this year. What they don't tell you is that in many cases this was accomplished by lowering the bar for admission.)

As I have shown in the studies cited at the outset of this posting, the vast majority of H-1Bs are NOT in the best-and-brightest league. On the contrary, many are rote-memory trained and quite lacking in the insight needed to develop a good product. Here is an incident reported to me recently by a reader of this e-newsletter (posted with his permission):

In 2003, I was an employee (U.S. citizen) at a major retail firm's IT shop. A team of Indians from Covansys was developing a Java-based stores POS application. The application was running painfully slow - 6 minutes to process a single transaction.

The Indians' performance recommendation was to buy a larger, faster server. Management called me in to give performance tips. I looked at the code. The Java application was creating a String object, then modifying it 100,000 times. As any Java programmer and most Java students should know, Strings are immutable. Each time a String is modified, a new object is created.

After modifying a String 100,000 times, this application had 100,001 String objects consuming memory, which naturally crippled performance. I asked the Indian guys why they didn't use StringBuffer instead of String. StringBuffers are mutable and only create one object, i.e., after modifying a StringBuffer 100,000 times, you still have only have one object in memory, not 100,001. The fewer objects clogging memory, the faster everything will run.

That one quick change improved the POS application’s performance by 60-fold, saving the cost of buying a faster, bigger server.

Once again, I do strongly support bringing in the best and the brightest, but if H-1B were limited to that, as it was (at least on paper) for the old H-1 statute, a yearly cap of 10 or 20 thousand would be plenty.

Norm