Wednesday, April 23, 2008

What Happened to the Indians?

I lamented to some Indian colleagues that we had no IIT applicants to the CS theory graduate program at Northwestern. Chicago and many other US institutions drew many of their best students from the Indian Institute of Technology campuses over the years.

I suggested that Northwestern was not yet on the theory map, at least in India. Likely true, but in addition the number of IIT CS majors going to the US for Ph.D.s has dropped by about two-thirds over the last couple of years. The culprit: Large, mostly US, banks are hiring the top graduates at salaries extremely high by Indian standards to work in their India offices. We had seen a smaller drop earlier with the software industry hiring but the software doesn't pay nearly as well as the banking industry. The Hindu writes about this trend.

I'm happy for India's success but worry about the impact on US science and CS theory in particular. You don't have to look far at the best theorists to see a large number of Indians, mostly IIT alumni. Imagine if most of them ended up as bankers in Mumbai. What a loss!

Back in the US, where do we get our graduate students from now? The Israeli's have long since stopped coming here, now that they can get quality Ph.D.s in Israel. Most Europeans also stay in Europe. We've also seen a drop in Chinese applicants. The US needs to start developing new sources for foreign students, or maybe, just maybe, find a way to attract more Americans.


  1. The US needs to start developing new sources for foreign students, or maybe, just maybe, find a way to attract more Americans.

    Or maybe, given the current job market trends, it is good that there are fewer students. Just a thought.

  2. Imagine if most of them ended up as bankers in Mumbai.

    Working for banks does not make one a banker. I wikipedia'ed the banking firm that first came to my mind (and which I knew was hiring the IITians) and found this: . Look under the employment section. Not quite "bankers", are they? They seem to be doing some research, maybe even math research.

  3. Try Iranian students from Sharif University of Tech. They are pretty good

    here is why

  4. IITians majoring in CS (or for that matter other majors as well) are getting huge salaries and in India. Hence, the incentive is quite high to stay back.

    US universities should also look to attract top talent from NITs which are considered one rung below IITs. The top students in NITs are comparable to the top students in IITs.

    One more thing. Reduce the application fees. Fees of 60-80 dollars limits the number of schools that a person applies. More often than not, it will be universities like Northwestern that will be in higher competition of being ruled out of the short list.

  5. No educational/hiring system that solely based on getting people from other, less rich, countries can sustain itself for a long term because other countries tend to get richer too.

    The largest failure for the US system is the inability of the state to give a good basic education to their own student, and a good motivation to continue in research.

  6. Israelis generally no longer require a CS Ph.D. from the US, true, but many of them after getting a Ph.D. are looking hard for temporary Post-Doc positions in the US, of which there are not nearly enough. Maybe the solution is (partly) in cultivating those?

  7. The US needs to start developing new sources for foreign students, ...

    An increase in students from mainland China is the next logical source, I imagine. All those trade-imbalance dollars need to find some way back to the U.S. from China.

    ... or maybe, just maybe, find a way to attract more Americans.

    Speaking as a potential American student, theory graduate programs could try making themselves more accessible to part-time students and students returning to school after being in the workplace. There are plenty of programs on the applied side that do this, but apparently not so much in TCS or pure math.

  8. In general, media has gone a long way to remove higher education as a value in the US. The colleges in the US are trying to get water from a most dry well. Until higher education is restored to the status of "essential" it will never get the talent.

    Talent goes where its trained to go. In the US we train talent to go where the money is, and then are upset when it goes to work for Banks.

  9. Lance, I think the US immigration process has done its fair share to dissuade people from India to come here. I gave an invited talk at IIT Bombay as an alumni and told them great things about Berkeley. To my surprise, no one wanted to learn anything about research at Berkley and almost everyone wanted to know if the US immigration process still treats Indians as criminals and "strip searches" people on airports?

    Plus the word out there's that Americans don't value education. Note that most comedy shows in US are available in India, and they often make fun of mathematicians as geeks, nerds, and Unabomber.

    If I had not come to the US before 2000, probably I wouldn't have come here either. Just makes no sense.

  10. Well if I'd known that the job situation would be so dire, I might have not gone to do TCS in grad school. And unless something changes soon I'll probably be telling people to do something else. Like work for a bank. And as many people pointed out we should not be relying on foreign students to be the major force in graduate study!

  11. Why would you want students whose primary goal is to get a PhD so they can work for the financial industry later? (Or even worse, people who come to the PhD program so they can immigrate more easily?)

    I'd say things have sorted themselves out quite nicely. Now that they have the opportunity, these people from IIT go straight into banking (what they want) rather than waste both their time and your time. And I can tell you that there's nothing that can make students who are really serious about TCS go into banking.

  12. Indian still coming
    Top 10 countries of Origin for H-1B year 2005

    * 1. India 102,382
    * 2. UK 30,755
    * 3. Mexico 17,063
    * 4. Canada 24,086
    * 5. France 15,403
    * 6. Germany 14,022
    * 7. Japan 14,858
    * 8. China 11,801
    * 9. Venezuela 11,703
    * 10. S. Korea 10,041

    from wikipedia

  13. Since there aren't any TCS jobs, graduates will all be going into banking eventually anyway. I can't really feel sorry for you. On the other hand, with less competition from a country with 1/10 the per capita GDP as the US, maybe the job situation will slowly turn around and you'll be able to get more Americans.

  14. Lance says: The US needs to start developing new sources for foreign students, or maybe, just maybe, find a way to attract more Americans.

    More than finding ways to attract more Americans, I think it is important to prepare more Americans for a career in advanced computer science, esp. TCS. If they are better prepared, they will get better jobs, and new students can be attracted much more easily by those prospects.

    Theoretical CS research in the US is shooting itself in the foot by not preparing their grads to be eligible for a variety of career possibilities (academia, industry research labs, engineering jobs, start-up companies, etc). Our brethren in pseudo-theoretical fields like machine learning have done a much better job of this, getting their students more hands-on experience with "real-world" problems. As a result, a mediocre PhD in machine learning (and most of them are mediocre) opens more doors than a strong PhD in discrete algorithms.

    I would encourage all theoretical CS students (except perhaps the top 5% or so) to gain a broader perspective on CS by working with other researchers in your department, to try and write a paper or two in bogus conferences like KDD, CIKM, WWW, and "less bogus but still nowhere near the quality of FOCS/STOC" conferences like VLDB, SIGMOD, etc. You will be much more marketable.

  15. I agree with some anonymous' opinions. The TCS job market doesn't justify a lot of graduate students in this field at all. Of course, professors always want to see more students so that their career can prosper.

  16. Or maybe, just maybe, CS theory is not as important as we all think. At least the market doesn't seem to think it is.

    Seriously, I think we get too many grad students in CS theory, more than there is demand in the job market. Compare this with econ departments or business schools, they have much fewer grad students, and on average they manage to get better jobs after they graduate. They know the law of demand and supply, and have limited supply to keep their field more desirable. We haven't, and as a result market has responded by giving us fewer applicants of lower average (and top) quality.

  17. This comment has been removed by the author.

  18. Hey, (Professor) Lance; I am really sorry, but since you have mentioned Indians and US universities, I might as well ask you this question:

    I am an Indian high school student with an interest in theoretical CS and mathematics - I have been trying to plough through Knuth for the past three years. Do you think I can benefit from applying to Northwestern as an undergraduate? Thanks a lot! Do any commenters recommend anything?

  19. The US needs to start developing new sources for foreign students

    Try indian students outside IITs. The non-IIT theory people (or other researchers) have done well despite starting out in non-top ranked schools and not having a tag with a fancy brand name attached.

    Except for a handful of academics the rest of IITians write code for microsoft or financial companies anyway. There is nothing to rue if a few more go to industry.

  20. Re: the previous comment .. I agree with its sentiment, and as an IITian, am embarrassed by the extra attention given to the IITs. However, I must disagree with the "handful of academics" stuff: just in theory, I can think (off the top of my head) of Sudan, Kannan, Mulmuley, Guruswami, Khot, Srinivasan, Ravi, Chekuri, Varadarajan, Charikar, Guha, Munagala, Goel, Kumar, Garg, Agrawal, Khuller, Radhakrishnan, Muthukrishnan, Vempala, Marathe, Mahajan, Motwani, Gupta, Pandurangan, Hariharan, Telikepalli, Venkatasubramanian, Rajaraman, Sundaram ... (as well as Raghavan, Kannan, Bansal, Jayram, ... in research labs) :)

  21. The Indians made my Ph.D experience a joy. I still remember on the first day of orientation meeting some young cocky Indian guy telling me "I was the topper in my class at IIT Bombay." And I said, "Oh really, I thought I met the topper 2 minutes ago." (Ok, calling it Bombay dates me.) Are they really gone? That's such a shame. It is just another reason to hate what happened to America in the past 8 years

  22. Oh yeah! I remember Pandurangan. He's my favorite theoretician.

    Honestly, dude. I've never heard many of those names you mention. Maybe they have more local than international significance.

  23. commenter 20:
    by handful I mean in comparison to the 150+ CS undergrads per year from the IITs less than a handful become academics (theory or otherwise).

  24. I agree with several commenters - I wish I would have gone straight to banking instead of being abused by professors for 5 years as a grad student/post doc, hoping that if I will lick their arses hard enough it will lead to something better.

    Best step I ever took was to cut my losses and join a bank.

  25. hey , i really agree with some of the people here . IITs , no doubt, are considered the best institutes for applied science in India, but there are others too. the toppers from most of the indian universities compare with the toppers of IITs, even in terms of motivation.

    moreover , it is very true that the tedious american laws are a big hindrance for the people who wanna immigrate.

  26. so many anonymous comments!!!

    Well @comment 24:
    Well I think that you have gone out of your senses. Though you were good but you followed the trend of your seniors to go and get a PhD from a US university. You should have thought about it before applying for a PhD and you should have the guts to leave the PhD if you do not like your advisors.

    Ya some professors do abuse their students. The best way to counter this is to stick to your own moral values and research ethics. Some day or the other he will feel ashamed. In the worse case you should leave him.

    @comment 25
    If you have an admit to any US university for a PhD program, it is rarely the case that your visa is rejected.

  27. Or even worse, people who come to the PhD program so they can immigrate more easily?

    As an American, I want these people immigrating. They're smart, they work hard. And they make more jobs than they take. Maybe PhD programs aren't they ideal way to bring them in, but I'll take an imperfect system to none at all.

  28. @26, anon25 is talking about getting a permanent resident or something, not a F-1 visa. Also, just search "H1b quota", you will understand what he/she meant.

    @27, apparently, your fellow American citizens disagree. Very interestingly, I found that Republicans are probably much more friendly to legal immigrants than Democrats.

  29. "@27, apparently, your fellow American citizens disagree. Very interestingly, I found that Republicans are probably much more friendly to legal immigrants than Democrats."

    Well, probably less Republicans are scientists, and therefore they are happy that someone else is doing the science. Meanwhile, maybe American scientists, despite all the benefits they get from foreign scientists, think that it would be slightly easier for them to find jobs if there were less foreign scientists. This is just a possibility.

  30. I don't intend to trivialize the discussion, since recruiting good students to US universities is important. On the other hand, did anyone notice the ads at the bottom of this posting? See this image. I'm not sure what message we should read from this, but maybe you should recruit more females in theory.

    I'm going to avoid dwelling on the obvious question of how to formalize the question of ads placement on web's more algorithms than complexity.

  31. I think it's good to look beyond the IITs for potential students -- it's the same reason why IITians look at other places than Northwestern. Simply because a school gets more attention, doesn't mean that other schools don't do quality work as well. Just a thought. I have many Indian colleagues, from IITs and also from other schools in India -- they're still good!

  32. I am from the CS depts at one of the IITs.

    Reducing Application costs, and increasing aids will go a huge way into getting people to come.

    People would now prefer to app to the ivy leagues, or to join the VERY high paying banks.

    It would make even more sense in something like TCS, where the number of opportunities are small.

  33. Some of the attitudes shown in these comments show why theory is dying. The list of theory researchers of IIT origin was not exactly star studded, nor does it show many names who actually made any difference. Doing theory for the sake of theory and publishing in conferences is not exactly a contribution to science in the real sense.

  34. ...not exactly star studded?Hmm?

    Have you heard of Karmarkar's algorithm? Or looked at faculty lists at top US universities? Chicago has two IIT graduates on the tenured CS faculty (Niyogi and Mulmuley). The Vazirani brothers are not exactly unknown people who had no impact in CS, nor is being the CTO of Yahoo an uninfluential position, and this is just off the top of my head late at night, firing off an instant reply.....

  35. yeah IITs are good and even best in India but that doesn't mean the students from other colleges in India are not as good as IITians.

    And the fact that the number has reduced would imply that people who do apply and enrol for PhD are the ones who really want to do it, especially theory.

    As far as theory is concerned it is sad that it is not as lucrative as Computer Networks or Systems area, even to do a Phd

  36. I wonder what makes IITians so much wanted for theory..

    Is that they are more hardworking and grasp easily or have done prior research at undergraduate level ?

    I'm not sure whether they have more courses in theory as compared to other schools (correct me if I'm wrong) even in India or across world and yes, I would prefer a math major more than CS applying for theory Phd.