Search This Blog

Monday, May 30, 2011

Are IITs World Class

Everyone has heard the honorable minister, and everyone has an opinion. I have been reading a lot of those opinions, and I thought I will add my voice to the clutter as well. Of course, I only know about IITs, and hence I will not talk about IIMs.

To me the answer is, "No, IITs are not world class, but it does not matter." The whole debate is not about what is world class, and whether IITs have achieved that standard. Whether we are in this list or that list. My belief is that such statements are often a reflection of the gap between expectations and perceived reality. And, when such statements come from people who ought to know, we should not dissect their words, but look at the causes of that gap.

May be the reality is much better but people don't understand. Well, isn't that a problem then. Shouldn't IITs work on that.

May be the expectations are too high. That, in my opinion, is a good problem to have, but may be IIT administration should do some expectation management.

But, may be the expectations are reasonable, and we still don't match them. That would be very worrisome.

Let me explain with an example. I have often written about JEE on this blog and elsewhere. One can argue that in the middle of huge scandals, IITs have largely been able to offer admissions on merit. JEE has not faced any major scandal (barring some doubts in 2006) over its 50 years of existence. And that is a HUGE achievement. And whenever I criticize JEE, people love to point out all this to me. But when I criticize JEE, it is with the knowledge of this background. I criticize because my expectation from the best institutes in India is much more than conducting a scandal free UG admission. I expect the top institutes to have a transparent admission process. I expect the top institutes to do some research into their admission process and figure out whether they are getting the students of the right type. I expect the top institutes to constantly debate alternative admission strategies. So don't point to me that IITs have the best admission process. They may very well have that. But I expect them to do better than the current best.

As long as there is a gap between what is expected and what is delivered, there will be criticism. IITs can keep ignoring this criticism and stay in their comfort zone. Or IITs can dissect the exact words of the Minister and say that if we get world class budgets, then we might improve. Or they can do some introspection and say whether they can do something within the significant resources that our country spends on them and narrow that gap of expectation and performance. Unfortunately, the last part is never done.

When someone says that IITs are not working with Indian industry, IITs can counter with the statistics of how many companies have given sponsored projects, how much funds they get in these projects and consultancy. But all that means nothing to the person who complained to begin with. Shouldn't IITs be doing much more than what they are doing. I recently wrote about the Waikato Institute of Technology at Hamilton, New Zealand, who are having amazing amount of focus with industry. I am not suggesting that IITs become as focused on industry as Wintec, but certainly there is a huge scope for improving industry interaction (and without increase of budget).

When someone says that IIT graduates are not exposed to live projects and the education is very theoretical, you can counter by telling about the latest equipment in your labs, and telling about the number of lab hours that students have to spend before getting a degree, but is that what I expect from the best institutes in my country. Why can't a civil engineering student be involved with a real civil construction project on campus, and all our campuses are having a lot of construction all the time. A student who has to spend just one hour every week with the site engineer would have a great experience and understanding of how civil engineering is actually practiced. Is it too much to expect that on an average, a CSE graduate would have written 20,000 lines of code in 4 years, and I am talking about original code, not code copied from Internet (while agreeing that Computer Science is much more than coding, just like Civil Engineering is much more than construction).

I also want to comment on the supposed selectivity of IITs. We are supposed to be world class because we are the most selective. Everyone in the country has heard the numbers, one out of 50 or something to that effect, while Harvard admits one out of 10. Are we really comparing apples with apples. Harvard tells you all the parameters that they will look at for admission. They give you statistical details of some of those parameters from last year's admission (parameters which are objective like SAT score). While in case of IITs, there is only one parameter, which is unknown at the time of application, which is the JEE rank.

Let us consider an experiment for IITs. Let them conduct JEE, announce the result, and then ask students to apply for admission, knowing fully well what the last year's closing rank was in each program. I don't think you will see even a ratio of 1:2. The large number of applicants for a small number of seats only means that people believe in randomness of the admission process. And if the process is based only on objective criteria, and there is still a large ratio of applications to seats, then it is nothing to be proud of. 4.5 lakhs JEE applicants are there because there is a general feeling amongst the students that JEE is a lottery, and on a given day, anyone with coaching can score high, and also, there are now some more institutions who have started using JEE performance for admission.

(By the way, the ratio of the number of candidates in AIEEE to the number of seats in NITs is much worse. Should we say that NITs have higher selectivity, and therefore, they are also world class.)

I think the problem with IIT system is that they are satisfied with being the best in the country. And whenever anyone points out that your parameters are not what the best in the world have achieved, there is a ready made excuse - money. Double our salaries, type of thing. IITs are not doing sufficient introspection to see if they could have met the expectation of our nation to a fuller extent within the same resources.


L said...

Yes, there is deficiency of resources, but with similar salaries and resources, some departments or some individuals in the same IITs do world class work I am sure.
I believe the ISI Kolkata and some departments of the IIIT Hyderabad do world class work (hearsay from a person who hires statisticians and machine learning experts). These people are also paid similar salaries, yet do fantastic work apparently.

VJ said...

It doesn't really matter whether these institutes were started for teaching or research -- its nevertheless a good idea to assess them to see where they stand in research at the global level. I was personally curious about a quantitative indication for the same. So I compiled the number of publications at the top venues in the research area that I know a little about. My findings are at

While these numbers may not be 100 percent accurate, they are still encouraging. What I concluded from these numbers is (a) (not surprising) there are only a few individuals contributing at the international level and they are doing very good, (b) several researchers (both academia and industry) have recently come back to India so we can optimistically expect the overall CS research (at least) to improve in the coming years.

A request for you and the readers of your blog: I compiled the numbers of publications for computer vision and machine learning since I am aware of these areas. I believe similar list for other areas such as networking, OS, PL, Algo, Complexity, etc. might be beneficial have similar values. If any of you is interested in compiling a corresponding list in your area of expertise, please send me the list or a link. I believe these lists would be helpful at least: (A) to the students who might be interested in pursuing a particular area of research, and (B) for networking/collaboration among junior researchers.

Rainbow Scientist said...

I have said it before in my blog, but IITs are only undergraduate institutes focusing on technical education and for that purpose they are good enough. I will not say they are world class because as you say if IIts are becoming IAS, managers at the end of their education, then the job of making engineers failed. As for research, they are not the research institutes and should be looked as one. They do have reasonable research output, but they are not the university with all the offered subject, so they are limited in what they can offer. Also, their research is directed by what problems west is studying at the moment, not what the local needs are and this is very short sighted.

WebMiner said...

"Double our salaries?" Dheeraj, I expected to hear more from you about the abject level of non-academic support that IIT faculty get. Compare IIT profs with middle-tier profs in the USA and you won't find planetary-scale differences. Compare what a single administrative staff does per day in the USA vs. IIT and the scales will fall from your eyes.

Comparably crippling factors are that (most) undergrads have too much on their plate to invest in research, and many masters students come with an undergrad education even more mediocre or dysfunctional than what IITs can provide. Excellence cannot arise out of muck. The ministerial ramblings are great examples of what happens when people with no connection to ground reality are handed a microphone: drivel comes out.

Unknown said...

It would be useful to highlight the list of concrete accomplishments and their impact over an extended timeline. There is a tendency among the general public and even IITians themselves to elevate good stuff to extraordinary proportions. In the areas I am familiar with (Algo, TCS, OR), e.g., Narendra Karmarkar's accomplishments are more hyperbole than fact, unlike the AKS work on primality. Impact over 10-20 years is hard. Many intuitively realize that and take the easy street.