Search This Blog

Thursday, December 27, 2012

Should we admit students through a lottery

 Thanks to my good friend, Surya Mantha, I read this long article on the Christmas day, titled "The Myth of American Meritocracy," by Ron Unz on the American Conservative website. The URL of the fascinating article is:

I will not go into the details of discrimination charges, but the solution seemed interesting, and I was wondering if the same solution would work for IITs (not that we have a problem of discrimination).

Paraphrasing his suggestion, it would mean that in an IIT, we get some top few rankers of JEE (say, up to the rank of 1000), and then we look at the next 40-50 thousand students, and select randomly amongst them through a lottery.

The advantages of this scheme would be (again, my understanding of how this scheme would work in the IIT context):

1. The goal of the students would become to get into top 50,000, since getting into top 1000 is just too difficult a goal. This is an easier goal than the current system where every single mark can change your career completely, and hence you will try to do everything to increase the chances of getting that extra mark. And if enough good students would not go through such an extreme form of competition, coaching, etc., then more students would have a normal childhood.

2. The unfair biases that any strongly competitive admission process would introduce will not be present in a lottery based system. So, for example, the admission process would not be discriminatory to women, and we would see a more gender balanced class. (The process itself is not discriminatory, but the fact that very few women are allowed to go to Kota kind of places for coaching, makes the overall system discriminatory.)

3. It will temper the arrogance found in too many of today's IIT students. Students would behave normally with each other, and greater amount of interaction amongst them will actually improve the quality of education.

4. Not getting into IITs would not be a matter of shame or stress. There is no shame in losing a lottery. So less stress in the society for everyone.

5. Since students were simply winners of the lottery, the society will treat non-IITians with much more respect that they truly deserve. A few random marks lost in a test is no way to judge and treat human beings.

The loss in this scheme, as one will surely argue, is that we are not getting the top few students, but the next best. As the article would argue, students at the very top are truly exceptional (in academic sense), and latter students are all roughly equal. (If you think 50K is too low, make it 40K.) In a sense, trying to rank students after the few truly exceptional ones, is a lottery by itself. Luck plays too big a role today. So we are just replacing one type of lottery with another form of lottery, but this new form of lottery does not induce stress, which the current form of lottery does.

Having said all this, I am not fully convinced of the lottery, and hence seek the views of my readers.

Sunday, December 23, 2012

Changes to GATE

A strategy committee for GATE has been formed. They have a problem to worry about. The number of candidates giving GATE will be 11 lakhs in 2013. The number was less than 8 lakhs in 2012. It is obvious that the current model of conducting GATE, where faculty members of IITs and IISc are present in each and every center, cannot be scaled beyond the current numbers.

But I find the initial proposal amusing. I know that it is just an initial proposal, and it may get changed during the course of further discussions. So, in a sense, this article is really to initiate the discussion on a public forum, rather than keep that discussion only restricted to the faculty members of IITs and IISc.

The proposal is to have a two tier exam, and following the example of JEE, the two exams will be named as GATE and GATE Advanced (How I wish it were Main Gate and Advanced Gate). The first tier will be during September/October, only online, on multiple days (but each student can give it only once), and conducted with the help of a partner organization with expertise in conducting online exams. The question bank for this test will still be responsibility of the GATE committee. Each exam will be of two-hour duration, and will have only Multiple Choice Questions (MCQs).

The selected candidates in the GATE will only be able to appear for GATE Advanced. It is expected that about 15 percent students will qualify for GATE Advanced. GATE Advanced will be a 3-hour exam, and will only be in the discipline of that exam. No questions on Engineering Maths, Aptitude, etc., which would have been covered in GATE. So this will allow greater coverage of the subject, since the entire 3 hours is focused on the subject.

I have a problem with this two-tier exam. If we can have the first tier for 11 lakhs (or whatever that number becomes in 2014 and beyond), then why not just use the score of the tier 1 for admission, and any other purpose that anyone may use the score for.

The strategy committee has two reasons for not using the GATE score for admission, and insisting on GATE Advanced. (They don't give these reasons in the proposal. They don't even consider the possibility of just changing the current GATE to proposed new GATE. But the proposal has something from which one can guess these two reasons.)

The first reason is that GATE Advanced will be able to cover the syllabus better, by just focusing on the subject test.

The second reason is that it avoids the comparison of multiple tests, since some inequality of the various tests will only impact selection for the GATE Advanced, and that too of the weakest candidates, and all stronger candidates will get exactly the same paper (GATE Advanced) and thus the comparison for admission will be fair.

The first reason is going against the very reason for including aptitude and engineering maths in the GATE paper. These two items were introduced by arguing that they are a good predictor of quality of preparedness of the student for higher education. But now, we are saying that anyone ranked 1 and ranked 1.6 lakh in the GATE are equal for the purpose of admission, which will only be done on the basis of GATE Advanced. This immediately means that people who perform poorly in engineering maths and aptitude portion and do well in the technical questions may get poorer marks in the GATE, but that will not affect their admission at all. And given that selection to GATE Advanced will be at very low scores (similar to GATE qualifying marks currently), these two topics are as good as useless for most students to prepare. Then we might as well make the first paper subject only, and cover the subject well in that paper.

The second reason is an admission of incompetence. We are admitting that even after having an experience of more than 50 years of conducting competitive exams, we have no clue as to how do we scientifically compare the two tests. This admission of incompetence is not a problem by itself. After all, we are primarily experts in science or engineering and not in subjects such as testing. But what is amazing to me is that while we admit our incompetence, we do not wish to seek expert help. Surely, there must be people in this country who understand these issues. And if there are no experts in testing in India, can we not seek consultancy from people elsewhere in the world. After all, everyone faces the same problem in such testing. GRE faces the same problem, just to give an example.

The simple logic would dictate that if you are having two tests than you should test different things in those two tests. And this was the problem in the JEE debate also (and the compromise that we have all accepted for JEE for 2013). If one has to do two tier tests, the first test should be just aptitude and engineering maths. Everyone should be allowed to give the second test, with the expectation that most universities and PSUs and other users of the score will require a minimum score in the aptitude and engineering maths. This would automatically reduce the number of students for the second test drastically, since everyone will first give aptitude and if get poor score will not find any good reason to waste more good money. But even if lot of students sit for the second test, it does not matter, since that too will be online and on multiple days.

Wednesday, November 21, 2012

Quality in Indian Higher Education

A lot of us keep bemoaning that the quality in Indian Higher Education scenario is largely absent but for a few institutes in each discipline. Consider engineering, for example, with a demand of more than 10 lakh seats per year, but the most liberal interpretation of quality would still not let you classify more than 50 thousand seats as having good quality of education.

Has the higher education policy of the government failed?

Of course, not. The policy has been fantastically successful. It was designed to create a situation that we see today, and it has succeeded. A higher education policy has to balance between access/equity, cost and quality. Government has always considered access/equity and cost to be more important parameters than quality. Please note that while technology and improvement in eco-system, governance, understanding of pedagogy, etc., will hopefully keep reducing the cost, we are far from a situation where cost is irrelevant and the quality education can be provided at a low cost.

If that be the case, it follows that quality education will require greater financial inputs. (By the way, it does not follow that greater financial inputs will always result in higher quality education. So one needs to be aware of all issues, and not just blindly throw money at the problem.) Where will this money come from. It can come from the government (tax payers), or from students, or in the form of philanthropy (including CSR of companies). When we talk about students, it can be their parents or through loans that they pay back later (there are more than one model for this payback, including a slightly higher income tax, for example).

But the government policies have consistently ignored this issue of funds for higher quality. It does not want to spend a much larger part of the budget on quality. So it can support only a few of its favourites for higher quality. There is a severe fee regulation and control in most states that does not allow even a better performing college to charge a significantly higher tuition. The fee that is allowed is such that it is not theoretically possible to even pay the minimum UGC salaries to all faculty members at the student-teacher ratio required by the regulator. The students and parents keep complaining that it is not easy to get student loans in a hassle free fashion.

Fee control is justified on the basis that in the absence of easy student loans and the lack of government subsidy, the higher education will become inaccessible to poor. This is a fine argument, but then this is precisely what I have pointed out in the beginning of this article that the policy is to give higher priority to access/equity and cost, and that the policy has succeeded.

A few deemed universities started claiming that as universities their tuition cannot be controlled and started charging higher tuition and started providing higher quality of education. Of course, there were some who started charging higher tuition and still provided lower quality of education. Instead of creating a distinction between the two, the government is threatening to bring tuition control to all deemed universities. Again, cost is more important than quality in our higher education policy.

One would have thought that the policy may permit a few institutions who have got a track record of doing better than average to go up the ladder in quality, either by providing direct subsidies or allowing them to charge higher tuition. But no. Government does not have funds for quality education outside its favourite institutes, and there must not be any institute in the country which is expensive.

And it is not just the government. Even the mango men (I love this phrase instead of "Aam Admi") want the same. Low cost, low quality in all institutions is better than low cost, low quality in most, and higher cost, higher quality in some. After all, this can create a class divide. Those who can afford quality education will become superior to those who can not afford quality education. So, ban higher cost institutions. The result, unfortunately, is huge competition for the few quality seats (like IITs), and a huge exodus of students for quality institutions abroad, when we could have been the education provider to the world, if we had paid some attention to the quality in our education policy.

Please note that I am not arguing for higher tuition here. I am only pointing out that low quality of higher education reflects success of higher education policy, and not its failure. Whether policy needs to change is something that more erudite people can comment on.

Wednesday, October 10, 2012

12th class performance of incoming batch at IITK

There has been a lot of debate about eligibility of students getting admission to IITs, whether it should stay at 60 percent, may be go up slightly, say 65%, or should it become 80 percentile, or may be a slightly lower percentile. But all this has been without any data on what would be the impact of any of these eligibility conditions on the selected students.

So, I got the data of 12th class performance of all JEE admitted students this year to IIT Kanpur, and tried to analyze it. Of course, we will not know the impact of 60% eligibility, since those who get less marks than that, perhaps don't even go for counseling, and most likely would not take admission, knowing fully well that they have to submit their mark-sheet by 30th September. But JEE office informally tells me that such students are very rare.

Since COBSE has announced percentage marks equivalent to 80 percentile only for General Category students, we looked at only those students marks in the 12th class. Since 80 percentile was given only for 2012, we have assumed that the marks would have been same in 2011 as well.

Out of 398 General Category students, 14 students would not have been eligible for admission to IITs. That is 3.5 percent.

And one may note that the current COBSE list is based on 80 percentile of all students registered for the 12th class board exam in that particular board. It has already been pointed out that the IIT Council decision was to consider 80 percentile of only successful students. It means that the cut-offs of eligibility in different boards would be higher by at least one percent, and a few more students would be then ineligible, say 4 percent. This is not a small number. Assuming that a similar fraction would be seen in SC/ST/PD/OBC candidates as well, we are talking about making 400 odd candidates (out of 10,000) ineligible for admission to IITs after they have performed well in JEE Mains and JEE Advanced.

I have said in an earlier blog that in the transition year, instead of doing too many changes at once, we should bring in changes slowly, and in particularly argued that a scheme which makes just 1 percent of candidates ineligible would put strong enough pressure to take schooling seriously. I have also argued that in view of lack of data on comparison between different boards, even if IITs want to change the eligibility condition to percentile format, it should be kept low at 70 percentile in the transition year. Well, it so happens, that if we look at the percentage marks corresponding to 70 percentile, and then see how many students would not have made it to IIT system, it would have been 4 out of 398, just one percent.

Another interesting point to note is that out of 14 students identified as below 80 percentile, 12 are from CBSE and 2 from Andhra Board. Now, we all know that CBSE board is much tougher than state boards. The standard of education is much higher, and  there is more than enough evidence to show that 80 percentile of CBSE actually has a much better academic preparation than 80 percentile in many state boards. We are just waiting for these 12 students to go to court next year.

Of course, there are too many variables that we don't know. What happens to the reserved category students. Is the statistics similar at other IITs (no reason to believe that it will be different). Because of additional coaching of 12th class this year, would most students getting through JEE advanced will also get 80+ percentile scores, or will because of additional coaching of 12th class this year, the 80 percentile cutoff will increase and there will be more students who would have been selected in JEE Advanced, but would be deemed ineligible.

Only time will tell how much chaos is waiting to happen in June-July 2013.

Added on 11th October: 

More data and interesting observations:

Out of 398 students whose data I have, 296 are from CBSE (74%), 76 are state boards (19%), 25 are from ICSE (6%), and one student from another board.

The distribution of percentile is more interesting. In CBSE, out of 296, 252 have 90+ percentile, 33 have 80-90 percentile, 9 have 70-80 percentile, and 3 have less than 70 percentile. In case of state boards, pretty much all students of all states have 90+ percentile, except AP board, where 2 students are in 80-90 range, 1 student in 70-80 range, and 1 student has less than 70 percentile. In ICSE board also, 24 out of 25 have 90+ percentile, and only one student is in 80-90 percentile.

What this means is that a student who can pass JEE with a top-5000 general category rank does not have to bother about getting 80 percentile in a state board at all, in fact, not even 90 percentile. But a CBSE student who can pass JEE with a top-5000 general category rank still has to worry about clearing the 80 percentile hurdle.

This also means that if students want higher percentile, they should leave CBSE board schools and join state board schools. This will not only help them in focusing on JEE and not worry abut 80 percentile at all, but it will also help them in getting higher ranks for NITs, where the percentile score is being included in the ranking.

What an interesting idea sir jee?
To improve the quality of school education, you incentivise people to leave better schools and better boards.

Monday, October 1, 2012

JEE and admission to IIT, not branch

Many of my readers criticize me for focusing on Under-graduate programs. They get more ammunition with this post, since this is again about UG admissions :-) But seriously, the UG admission and the UG programs are totally broken in this country. PG education has a problem primarily because of poor-quality UG programs. There are lesser structural problems with PG education (and I have written about them at times).

The issue that I have been thinking about is whether it is a good idea to offer admission to a branch (like IITs and all engineering colleges do right now) or just offer admission to an IIT (I will focus on IITs), and the branch selection is done sometime later (say, after a year when they have all done some basic courses, and after taking into account their performance in these courses).

This idea is not new. I have heard of this almost every year in IITK. It was actually recommended to Senate in 2012 (where unexpectedly it was referred to another committee). Many people like the idea, many others don't. What do people feel will be the negative consequences of such a move. First, if IITK takes the leadership role and does this suo moto, without other IITs joining to do the same, then the top rankers in JEE (say, in the top 500 odd ranks) will prefer to join a specific program in another IIT then joining IIT Kanpur with uncertainty of the program that will be allotted next year. But on the other hand, students with 2000-3000 ranks may prefer to try to work hard in IITK and attempt to get a branch of their choice rather than get a guaranteed branch in which they have no interest. So, the chances are that we will have students in the narrower range of 500-3000 ranks, rather than 100-5000 ranks. (I am mentioning only ranks of unreserved seats, but similar thing should happen in reserved seats as well.) Interestingly, it means that the average "quality" (if JEE measures quality) will be no different if we do branch-less admissions. However, the problem is that the popular programs may want to have those students in the 0-500 ranks. On the other hand, the less popular programs would be afraid that while they will attract better ranked student, but they will get only those students who have switched themselves off after coming to IITK (since performance in the first year will matter). And, of course, as we have said earlier, students and parents would prefer to know which program they are getting admission in. So there is really no constituency for branch-less admissions.

The other reason that people say against branch-less admissions is that it will make the first year extremely competitive , and the stress period for the student will just get extended by one more year.

Is there any advantage for branch-less admissions. Actually, I believe that if handled properly, it can reduce stress, particularly if many institutions adopt this model of admission. As I have said before, the stress in JEE comes from the fact that a small mistake can affect your future very seriously. And branch-less admission is essentially telling the student that a small mistake will not affect the future
seriously, since you could work harder later and improve your chances of a branch closer to your interest. So instead of JEE being a pressure cooker, there will be lesser stress but spread over a longer period of time. And having lower stress for a longer duration is better for the mental health than the other way round.

But even more importantly, branch-less admissions are also more fair. During the debates on changes to JEE, those opposed to changes in JEE pointed out that there is a wide error margin in the 12th class marks, and therefore its use in ranking is unfair. Well, in JEE, the error margin may not be that wide (since the error on part of grading is not there), but still it is there. And indeed, even when two students get exactly the same total, we artificially try to rank them by saying one subject is more important than the other, etc. A single exam will always have error margin (and that is why exams like GRE give scores which are bunched in multiple of 10s). How fair is it to rank students who are within the error margin of each other, and based on that rank give them a branch for the rest of his/her life.

When we have a situation where no stake holder is interested in changing the system, and yet, it seems that the current system has problems which can be addressed by those changes, the right strategy is to change very little, show that it helps a little, then make a bit more change, and so on. One way to do that will be to say that while we admit students on the basis of JEE scores (or ranks), the final branch allocation at the end of the year will be on the basis of sum of JEE scores and some measure of the first year performance. In the beginning, the measure of the first year performance can be kept very small, say making a difference to the original JEE score by no more than a couple of percent marks. This will mean that if we rank all branches by way of popularity, then the branch that a student will get after the first year will be either same as the one s/he would have got at the time of admission, or will get a branch which is one higher or one lower in that ranking of popularity. Indeed, if one performs around the average of the batch, then s/he will get exactly the same branch after one year as s/he would have got in the beginning.

Once we do our minor tinkering with the admission process, and we do not see any major problem, we can increase the weight of the first year performance slightly.

Saturday, September 15, 2012

A 5-year plan to get into top 100 QS Ranks

I know a lot of my readers would like to argue that rankings are arbitrary, and one should ignore them. I respect their views, but still would like to suggest a plausible roadmap for the Indian universities in the top 300 or so ranks (basically the old 5 IITs) to get closer to 100th rank in about 5-years time. Be warned that the steps that I suggest may not be in the realm of improving teaching and research substantially, but none of them would hurt the cause of quality, and many will actually improve it. Also, I am focusing on IITs for two reasons. One, I am more familiar with the playground. And two, they are already in the top 300 odd positions in the QS ranking. But, of course, the kind of things I am suggesting, they are applicable more widely.

Just to put challenge in a perspective, let me point out that the total expenditure by an IIT in a year (including non-plan, plan, and R&D) is about Rs. 5 lakhs per student, while for the top ranked MIT, it is over Rs. 1 crore per student, and while there may be less costly way to reach the top, it is obvious that the society will have to make a huge investment to get even close to top.

I start by observing that if we look at the best performance amongst the five IITs on all the six parameters that QS rankings measure (IITB on Academic Reputation, Employer Reputation, and International Faculty, IITD on Faculty-Student ratio and International Students, and IITK on Citations per faculty), and combine these best performances into a composite score, such a hypothetical entity would already be at a rank of about 180. This shows that the current financial, administrative, and other constraints cannot stop an IIT from being in the top 200 (and indeed IIT Bombay was in top 200 just two years ago).

In fact, I think that there are inefficiencies of the system, which if removed, we can further improve our performance in QS Rankings (and this article suggests some steps in that direction). So I believe that a rank of 150 is quite possible within the current financial constraints of the Government of India. If we can have some improved funding both from government and non-government sources (including alumni), and have greater autonomy, etc., getting close to 100th rank should be possible. To climb up beyond 100 would require substantial financial inputs, but that call can be taken by the society after we improve to near 150 ranks.

First thing to note is that all our institutions have a very poor faculty-student ratio. All 5 IITs have a score between 28 and 35 (out of a maximum 100). This was not always the case. In fact, we used to have an excellent faculty-student ratio till 20 years ago, and we had a not so bad ratio till 2008, when the Government decided that we had to increase our student strength by 54% within 3 years. While the student strength had to go up by 54% within 3 years, it is impossible (and even undesirable, if I may add) to increase faculty by 54% within 3 years. To make matters worse for the existing universities, the Government has also set up a large number of new IITs, NITs, IISERs, Central universities, and so on. So the competition for quality faculty became intense, and we have grown at a much more modest pace. But the good news is that the student strength has mostly stabilized, and most IITs continue to grow in faculty size. We need to be far more aggressive than what I have seen so far. I am speaking mostly from IIT Kanpur experience, but what I hear from other IITs, there is a definite scope for improvement in how we keep in touch with PhD students and post-docs, how we respond to queries regarding faculty applications, how we treat them when they come for visits, formal selections, what kind of help is provided when they decide to accept our offers, and so on. If we can increase our faculty student ratio by 15 to 20 percent over the next 3-4 years, we improve our rank by about 50 places just on this count, and IITs have sufficient budget and autonomy to do this. Perhaps setting up of more chairs by well wishers of these institutes will help. A bit more money as research initiation grant will help. Some more support for Professional Development (the current number of Rs. 1 lakh per year is very low) will help. So money will help, but we don't need a coal block allocation to improve ranks by 50 places. And this is something that we should do anyway, not just for ranking.

One specific suggestion for this is to look at people working in industry, who want to move to academics for 2-5 years. What I have seen in academic world is that we have the same parameters for hiring a fresh PhD and for hiring an industry person with 25 years of experience. I do not wish to dilute the standards, and I do not wish to suggest that someone be hired just because someone is coming from industry. But I look at the two faculty candidates differently. When we are considering a fresh PhD, we are looking at a person who can potentially be with us for 35-40 years. If we made a mistake in recruitment, and this person does not perform well, it will turn out to be a huge mistake by the department and the institute. Moreover, we do not know how this fresh PhD will build his research and other academic activities when s/he starts working independently. However, when we look at a 55 year old for a fixed tenure contractual appointment, we have a fairly reasonable idea of what this person is likely to do in those years, and if the person performs poorly, it is a loss for a much smaller duration. Therefore, in case of a fresh PhD, if we are in doubt, we don't offer, but in case of a senior person, if we are in doubt, we could take the chance and make that offer.

There are also examples when a faculty member from a decent institute (but, for the sake of argument, a lower-ranked institute) wants to spend a semester at an IIT, while being on leave from his institute. Many of us would like to have the same sieve applied to such an application. For a short term faculty like this, I would only look at whether there is some benefit to us and whether there is some benefit to that faculty member and his institute, when he goes back. So if such a person is a good teacher, but not a great researcher, I would be happy to invite him to offer a course, with the hope that he will also interact with some of our faculty members, and improve his research as well. So our teaching load goes down, our QS ranking goes higher, and we have helped another institute in growing up. Again, I am not suggesting that we recruit anyone and everyone for a semester, just to show higher numbers of faculty in QS information sheet. I am only suggesting that we look at specific benefits to us instead of comparing the applicant with an applicant for a permanent faculty position.

To put in perspective the growth requirement of faculty, we need a net increase of ONE faculty in each department at these IITs for 4-5 years to improve ranking by more than 50 places. Given that we lose on an average one faculty per department to retirements and resignations, what we need to achieve is hiring (and joining) of two faculty members per department in the top few IITs. By no means, I want to portray this as a trivial task. It is a challenge, but something that can be achieved with sustained efforts, planning, strategy and leadership.

The next aspect of QS ranking where we perform miserably is the internationalization of our faculty and students. The best IIT (Bombay) in this respect has received 0.2 marks out of 10.

First the students. Let us understand how IITs admit foreign students to its under-graduate programs. Everyone has to give JEE. The foreign nationals are given admission to a program in an IIT where an Indian student in unreserved category with that rank could have got admission. So the foreign nationals have to compete with Indian students in unreserved category. (Admission to foreign students are not counted towards the number of seats available for Indian nationals. These are extra admissions.) This implies that in the entire IIT system, one would have no more than 5 foreign nationals, and these too would normally be those whose accident of birth took place in US, when their Indian parents were working there, but have since settled back in India. They have gone through the same school system, the same coaching, and gone through the same JEE. Culturally, they are as Indian as any other student, and they don't bring the advantage of exposing the classmates to a different culture, which will broaden the horizons of other students. These few students do not even bring the advantage of higher revenue. First it is just one student in a batch of 1000. And too, even he requests that he be allowed to pay "Indian" tuition, which is generally accepted.

But we could do admissions differently. What if we admit all foreign students who get marks above the cutoff marks in JEE, and for the purpose of seat allocation, we become more liberal. If we do this, we may be able to admit not 1 out of 1000, but perhaps 10 out of 1000, some of them with genuinely different cultural background, thereby bringing diversity to the classroom, all of them paying full tuition, and allowing us to climb the QS rankings by a few notches.

Other things we could do is to be liberal with our students going out for a semester to foreign universities (at least those who can afford), and pro-actively attract foreign students to spend a semester on our campuses. In particular, our MBA programs could have agreements with some good foreign MBA programs for such student exchanges. This improves education of our students, and our QS rankings go up further. The only thing we need to enable this are somewhat better living conditions than our typical hostels. There is enough interest in foreign students to experience India.

Similarly, can we not attract foreign students to our MTech and PhD programs. The impediment to attracting foreign students is really our attitude. There is a belief in the academic community that it is not our job to attract students. We must remain aloof and pure, and only consider applications from those who apply on their own. May be we need to hire a marketing guy in our institutes. (It used to be our belief that attracting faculty is not our job. But that has changed in the last decade or two. Hopefully, we will soon start believing that attracting students is also our job.)

Next, attracting international faculty. If we were to believe our bosses, the problem is that of the government. They don't allow permanent jobs to foreign nationals, only a 5-year contracts. And that is the only reason why foreign faculty is not queuing up to take positions in IITs. Frankly, that is nonsense. Not many foreign nationals are interested in life long jobs in India. In fact, most of them are interested in spending brief periods of time in India, from a semester to a couple of years. Again, most of us don't see much value accruing to us if a foreigner spends a semester in our department. We suspect that s/he is only interested in tourism and would do little research and in most cases this won't lead to long term collaborations. But why can't we invite them with just the expectation that they will perform as much as any of us do. After all, we are only paying them as much salary as we ourselves get (while he probably gets much more in his home country). So, if a foreign national faculty member with a good academic/research record wants to spend a semester, will teach a course, give a few research seminars, interact with a faculty or two, and a few graduate students, but will not have any long term collaborations, isn't it just fine.

If we do all this and our international faculty could increase from around 1% today to 3-4%, and the international students can go up from 0.2% to 2%, we would have improve the diversity, the quality of education for our students, and gone up 10-20 notches in QS ranking. And again, I think these are achievable goals within the current budgets and administrative constraints.

The major part of QS ranking is the Academic Peer Review which has 40% weight in the overall marks. This is basically the perception of academicians around the world. When these academicians fill up the survey, how much do they know about us. They would typically base their views on whether they have seen research papers from our institutes, whether they have interacted with faculty, students and alumni from our institutes, whether they have visited us, or they know someone who visited us and told them stories about us. And some of them who have heard of us through the grapevine but not really know about us, may even try to find out about us from the website.

What is the impression that they will take if they visit our website. They will see hundreds of faculty members not updating their websites for months. Even the institutional part being full of obsolete information. In today's day and age, it is absolutely essential to have a great website that clearly gives the impression of the energy present on the campus. We must be present on social media. We must inform the world through facebook and twitter every exciting thing that is happening on our campuses (and there are lots of things indeed happening out here). We must have a large mailing list of academicians around the world (at least Indians and NRIs, if not many foreigners) whom we ping once in six months and let them know the fun we are having doing our academics and research, and gently suggest to them that perhaps they want to participate in that fun for a semester or two.

I see that in most of our communication, we tend to over-emphasize that we are an Institute of Technology, even though half our PhDs are from Science departments, and even though we have other vibrant programs like Economics and MBA. By emphasizing our engineering departments, we make sure that academicians from engineering disciplines know about us (and hence we do get a good rank in engineering disciplines, mostly in top 100). If we want to get close to top 100 overall, we need to make sure that academicians in sciences and other disciplines also know about us.

It may seem like I am suggesting some marketing approaches only to get higher QS ranking, with no impact on the quality. But I believe that greater visibility will lead to easier faculty recruitment, more international students, more collaborations, student exchange programs, and several other benefits that will improve the quality of research and education at our institutes. And some of these steps will also improve perceptions of the employers (which constitute another 10% weight in the ranking). And greater visibility of our research will also improve citation of our papers (which constitute another 20% weight in the ranking).

I really believe that if we do all these things, we will get close to 100th rank in the QS ranking, and if the improvements in ranking are encouraged by the government with higher funding, we can break into top 100 in the next five years.

Perhaps, government can help in the following fashion. It can decide that universities which are more likely to break into top 100 will be given extra funds to speed up the process of doing that. It can select 20-30 universities most likely to break into top 100. These can be top 5 universities in engineering/science/liberal arts/ and so on, and perhaps top 5 large universities with all these disciplines. There can be set goals and associated increase in funding. For example, for the IITs, one could say, once you break into top 300, all new faculty members will be given an extra 50 lakhs of initiation grant. Once you break into top 250, the professional development budget of each faculty goes up from 1 lakh to 3 lakhs. Once you break into top 200, each department will get a one-time grant of Rs. 10 crores for further improvements in whatever aspect they are lacking, and so on.
 (These are just examples, and not very well thought out ones at that. I am just trying to suggest a framework of encouraging certain institutes to focus on improvement in ranking. Of course, faculty members of some institutes may reject such inducements and continue to do the good work that they have been otherwise doing, without an explicit focus on rankings.)

Thursday, September 13, 2012

No Indian University in Top 100 of QS Ranking

Make that 200. The top rank goes to IIT Delhi at 212.

Let me draw up a list of reasons that I expect to hear and some proposed solutions.

QS rankings are biased. Their methodology favors China (and Korea and Japan and Brazil and everyone else).

Of course, they are biased. Haven't they heard of coaching mandis. Haven't they heard of JEE. Am I to believe that average student getting into MIT can do better in JEE than students that get into IITs. These rankings are just a plot by ISI to show India in poor light.

I think IIT Council must call for an emergency meeting to come up with a new Indian ranking system, which will be based solely on preference by JEE selected students of various universities. Any university who wants to participate in such a ranking will have to admit students only through JEE and the percentile score of 12th class board exam. If we do this, we will become the world leaders, with 19 out of top 20 universities being from India. We will make sure that there is at least one non-Indian university in the top 20, even a Pakistani university, lest others should doubt the ranking system.

We are a poor nation, and our priority is primary and secondary education.

Yeah, sure. That focus is so clearly visible in the PISA report. But I have a suggestion, and it has been inspired from an excellent paper that Prof. Mehta of IIM Ahmedabad wrote recently on Why Harvard is Number One university in the world (needless to say that QS does not agree with the title of the paper). It includes a brief history of Harvard, how it started as a government college, but government never had sufficient funds to support it. So early in Harvard's history, when the Government couldn't provide Harvard sufficient resources, it gave Harvard the right to receive the revenue of ferry services across the river Charles. So perhaps the government of India can do something similar.

Give just one coal mine to each IIT.

The main problem is that we lose marks for internationalization of our campuses. There is five percent weight for presence of foreign students, and five percent weight for presence of foreign faculty.

There are two solutions to this. We can bar Indians from going abroad for higher education. Thus, we will save billions of dollars that these people spend on education abroad. The top universities in the world will perform poorly next year on Internationalization parameter, since Indian students form a significant part of their international students.

On the other hand, we should impress upon QS ranking folks that they should not go by the nationality mentioned in the passport. A lot of students and faculty may not know who is the Vice President of India, but they certainly know the names of each and every Senator in US Congress. We should count them as India Born Foreign Students. (Just like we have the Indian Made Foreign Liquour - IMFL.)

Only universities established in a city called Cambridge are eligible for the top ranks.

MIT, Cambridge University, Harvard.
This is a simple problem to solve. In fact, this is not even a problem. British had a very short supply of names, and everything had to be named after something in England. So they first started calling my city, "Manchester of the East." But they had the foresight to see that one day all the cotton mills will be closed, and only education business (including a world class coaching mandi) will flourish. There is sufficient historical evidence that the city was to be renamed as Cambridge by British. When the order came, the local British agent could not read the poor hand-writing. So "Cam" became "Can." And to make life easy for the locals, "bridge" was translated to its Hindi word "Pul." And over the century, "Canpul" got corrupted to "Kanpur." I think we should just respect history and rename the city as Cambridge. If not Kanpur, may be we can rename Kalyanpur as Cambridge. If even that is objected to by the locals who do not understand the value of QS ranking,we can just rename the IIT as "IIT Cambridge." If we can have IIT Bombay in Mumbai, IIT Madras in Chennai, and IIT Gandhinagar in Ahmedabad, why can't we have IIT Cambridge in Kalyanpur.

While the number of research papers from Indian universities has been going up, their citation index is still low.

Serious problem. It is in our genes, I tell you. We pull each other down. You look at top 20 universities. The researchers in those universities will liberally cite each others' papers. But what do we do. We also cite papers written by researchers in those top 20 universities. We are stabbing our professional brothers and sisters in their back.

Again, where is IIT Council, when we need them most. Why can't they make it mandatory for all papers published from all Indian universities to cite at least 10 papers published by researchers from other Indian universities. The problem of low citation index will soon be history.

Concluding remarks

We only need to think out of the box. The traditional recipe for excellence invariably means hard work, accountability, planning and long term vision backed by strategy, massive amount of resources, and so on. But our leadership has shown that with innovative thinking one can achieve excellence by doing nothing. The problem of school education in the whole country can be solved simply by tweaking admission process to colleges. I have tried to follow that leadership style, and suggest ways to get into top few ranks next year without hard work, accountability, planning, vision, strategy and money.

Sunday, August 26, 2012

Suicides at IITs

 Every year, 4-5 students are ending their lives in the IIT system. Given that the total student population in the IIT system is between 40-50 thousand, we are losing one bright young life for every 10,000 students every year. This is a serious cause for concern.

When any tragedy occurs, the first and foremost question that everyone has is: "Why did this happen." And when such tragedies happen as often as have happened in the IIT system, it is natural that people will want to know the reasons. Unfortunately, it is not easy to find the reasons. Each life is unique, and the reasons to end that life are also unique. In most cases, if not all, there are multiple reasons behind such a decision, though one of them may have acted as a trigger on that fateful day. But, if there are so many tragedies, then there must be something common between them. People want simple answers, which they can understand. And if the experts fail to give a simple answer, they will invent one. And hence the common perception that these deaths are caused by academic stress.

And this perception has ensured that the focus of the Institute authorities is on reducing academic stress, and less attention is paid to the "real" issues. When someone takes away his/her life, and a question is asked what have you done since the last such incident, you can not just say that we are trying to do things that will increase interaction amongst the students, even though that may be the most important thing to do. You have to tell how you have reduced academic stress since last such incident, because that is what most people understand as the reason.

IITs are a competitive place. The admission to IITs is the most competitive exam in the world, for which many students study for 3-4 years, and even drop one year after passing 12th class to prepare for IIT admission. The competition to perform better than average in such a group can be very intense, and someone who is used to be amongst the top few in his/her school for 12 years would feel stress on realizing that s/he is performing worse than average in this group.

If I look at the curriculum at IIT Kanpur (since I know more about it), we have less courses than whatever I know of curriculum at various NITs, we have less contact hours, we fail fewer students, we provide opportunities to recover from failure by offering summer courses, and so on. We have a lower graduating requirement (in terms of grades) than any NIT. (From 2011 onwards, one only has to pass all courses with a 'D' grade to get a degree.)

A large number of changes have happened in IIT Kanpur in the last 5 years in response to these questions about academic stress. We teach less - the working weeks have been reduced from 15 to 14 in a semester, and additional days have been given to spread the exams (to minimize the probability of two end-semester exams in a day), as well as ensuring that there is a gap of 2 days between the classes and final exam. We have reduced class timing from 55 minutes to 50 minutes, to further force a reduction in course content in every course, and enable a more leisurely movement from one class to the other, enabling the students to ask a few questions from the instructors at the end of the class without the stress of getting late for the next class. The number of fail grades is an all time low, around 2.5 percent of all grades in the Institute. I wonder if there is any university in India with a lower fail percentage. The graduation requirement has been reduced from a CPI of 5.0 to 4.0, basically allowing anyone passing all the courses to get a degree. Again, I wonder if there is any university in the world which gives degrees at "D" average, like we do. The students don't even have to bother about showing an "F" grade on their transcript. They are allowed to withdraw from a course just a week before the end-semester exam with no mention of such a withdrawal in the transcript. We have changed the rules for Academic Warning, Probation and Termination (for under-graduate students) so that only a fraction of students will get into these states. We have started giving additional chances to a student whose program has been terminated to explain his/her poor performance, and have re-admitted several such students.

Today, there is no doubt in my mind that the "real" issue causing stress to the students is competition and not the curriculum and academic rules. And hence the solution is to counsel the students to not get into a rat race. They need counseling that a five point someone can have a good life ahead. They need counseling that if they find it hard to cope up with all the courses in a semester, there is no harm in dropping one or even two courses. The stress from peer pressure in the hostels to complete the BTech degree in 4 years is intense. They need to be told that it is alright to be slow and steady and complete the program in extra time. But it is easier said than done. Remember, these are people who are intensely competitive. That is how they got through JEE. To now tell them to not worry about competition is certainly not very convincing to them.

I remember when I was the Department Under-graduate Convener several years ago, I would call each student on Academic Probation, ask them to register for courses which they have already failed once, and they think they failed narrowly (in which they have a easier chance to pass), but they will all want to do CS courses because they could do other courses in summer and still have a chance to graduate in 4 years. I would tell them that they should first focus on getting a few 'C' grades or better on their respective transcripts and get out of this cycle of Warning and Probation, and only later worry about how much time their degree will take. I will then monitor their performance in these courses, and if one is performing very poorly in some course, ask him/her to drop that course, since the termination rules at that time were based on the performance in courses that one did that semester. Again, there would be huge resistance. "I will work hard and make up and pass the course," was a common refrain. I couldn't force them, but if I was spending hours with each one of them, they reluctantly would agree to my advice. I was happy when at the end of the semester, there was not a single CSE BTech student in the termination list, but I became famous as someone whose sole aim in life was to delay everyone's graduation. Most of these students (who were on Academic Probation) felt that they could have passed more courses, that they could have passed advanced department courses, and that my advice held them back I doubt if anyone felt that because of my advice they were still students of IIT Kanpur.

So in my opinion, the real challenge is to convince someone to go slow, and ignore the competition. I recall we used to have a compulsory slow-paced program in the first year based on a diagnostic test. In this program, the student would do a particular course in a slow pace, learning the same material in two semesters, instead of one. One could do slow-paced learning in multiple courses also. The idea was that once the basics have been learnt well, it will be possible to learn other courses easily. But then there was an opinion that a forceful slow-paced program was causing stress. So we made it optional. The number of students choosing this program reduced substantially immediately, and the number of students in Warning and Probation increased, but because the slow-paced program was identified by outsiders as one of stress inducing issue, we could never make it compulsory again, and the numbers kept reducing, and finally we don't have any such program now.

The difference between stress due to competition and stress due to curriculum/academics-in-general is to be understood, and handled properly. If we do not understand this difference and keep reducing academics, we are only going to reduce the quality of education in our top institutions, without improving the experience of our students, and without making any dent towards solving the problem of excessive stress.

There are other more important reasons why we should not report every such death as linked to academic stress. First, it is simply not true. Suicide is a very complex issue, and one does not take away one's life because of one reason. Even to the extent that one reason is a trigger, academics related reasons are for very few students.

And secondly, if we simplify the reason for suicide, we are encouraging other suicides. The phenomenon is known as "Copycat Suicide." Read more about it at this Wikipedia page. In short, when someone says that a student committed suicide because he had a low CPI of 5.0, it makes the other with CPI of 4.9 think of the same step. But if it is pointed out that suicide has complex reasons, including psychiatric and medical reasons, then the student with a lower CPI does not relate that suicide to his own situation.

And this brings to the most important issue - reporting of suicides. I was browsing the net for information on how to deal with suicides, and came across this site on how to report suicides. It tells us that there has been a lot of research on effect of reporting of suicides on the next suicide, and it is agreed today that it makes a significant difference. I hope our media is aware of it, though the signs are quite to the contrary. If you look at the recent reporting of two suicides this week, the media talked about a possible problem with some relationship as the cause (which itself was not proper, if you agree with the reporting norms suggested by the site mentioned above). But soon after the second suicide, the media was talking about academic stress causing a series of suicides in IITs, completely forgetting that just the previous day, the same newspapers had mentioned a different possible reason for a suicide.

There is no doubt that a lot of people, whether in media, our alumni and other stake holders, are genuinely concerned about 4-5 suicides a year in the IIT system. I only wish that they will report, discuss, debate and talk about the issue sensitively to make a positive impact on the situation, and not a negative one. The current reporting is putting pressure on the institutes to focus on academics, while the need is to look at it from a wider perspective.


Tuesday, August 21, 2012

72 hours without a cell phone

Three days ago, on 17th evening, I lost my cell phone. A less than a year old, Samsung Galaxy S-2, the only expensive cell phone that I ever bought. And this happened when I was out of town, to Surat and Ahmedabad, which was far better than losing it in a new city without many friends and acquaintances. What I found amazing was the response of the service providers to this tragedy.

In the pre-mobile days, there were two disasters that could hit a traveler. You could lose a passport, or you could lose your credit cards. I have had the experience of getting an emergency passport in Indian Embassy at Washington in two hours flat, and Citibank sending me a new credit card within  24 hours. But today, losing a cell phone is a much bigger calamity than either of the two, since it affects you even at home, and of course, could be very serious while away. But can you recover from the loss in 2 hours or even 24 hours. Sorry, you will be in for a bigger shock than the original shock of losing your phone. Customer service is not exactly a strong point of cellular companies in India.

Director of IIT Gandhinagar, where I was visiting, assigned the task of helping me out to one of his staffers, Santosh, a young and dynamic person with amazing patience and full of ideas. The first task was to get the number blocked. The first few calls to 111 met with a standard response, file a police report, bring a copy of the report to one of the Vodafone stores or ministores, and they will block it. But that will take time (since police wanted me to show a proof of purchase and tell them the IMEI number of the phone, which will take some time to recover, sitting 1000 KM away from home). Finally, he called 9839098390, the UP(East) circle's customer service number, and they were more helpful. After several questions regarding my identity, they agreed to stop the service. I wish they had asked some question which Santosh did not know. I would have felt more secure. This way of blocking service could be used as a denial of service attack. But anyway I was thankful that the service to my phone had been blocked.

I was stupid enough not to have registered for Samsung's FindmyMobile Service, which is free. But I had enabled the 2-factor authentication for google services, and hence I could login to google to revoke access from my mobile to all google services. I was also carrying my Galaxy Tab (with an Airtel connection) and a laptop (with a Tata Photon connection). So I did not need an SMS to my phone to log in to google. (And, I was ready with one time codes from google, in case I had to login to google account from untrusted devices, which later on, I had to.) So, now I was sure that while the person finding the phone could look at my stored emails, my contacts, my calendar entries, but he would not be able to change any of them. And I had not stored any password on my mobile. There was hardly any data on the phone. And I had regularly backed up my mobile phone using Samsung Kies. So I wasn't going to lose anything important. (Though I haven't figured out yet how to restore that on a new device.)

Having taken care of the blocking of phone, I was quite at peace. I could use the SIM card of the tab in a temporary phone. The tab had all the numbers, and I was sure I could borrow a phone for a couple of days, before I bought a new one. So I thought of sending SMS to a few people telling them of my temporary number. But my tab kept telling me that SMS could not be sent. I thought it was some configuration issue, and searched google for similar problems. Couldn't find anything useful, but the SIM in the tab was on pre-paid, and all the searching, and emails, and stuff made sure that I ran out of charge on the tab. So one more device down. I switched on the laptop and thought that I could add some charge to the tab account and make that functional again. But it was midnight by then, and it kept showing me airtel site being down, perhaps for maintenance. I was paranoid by now. What if my laptop also stops functioning. I would be totally disconnected from the world. So I added 1GB charge to the Tata Photon service, when I already had 200MB left in my account, and my normal usage on a day when I am out of town is about 20-30 MB.

When I woke up in the morning, I thought I will try recharging my Airtel account for the Tab once again, and guess what. at 3:00 AM, my laptop, which I had forgotten to shutdown before sleeping, had downloaded Microsoft Windows updates, Vaiao updates, Semantec updates, and everything else that it thought it needed to become up-to-date, and had exhausted the entire 1.2 GB. Since I mostly use WiFi for connecting laptop to the net, my setting for updates has been to let it download everything, but ask me before applying those updates. I hope that in future, on laptops, they will allow me to set different update policy for WiFi and a different policy for cellular network, something common on cell phone operating systems.

Now, just imagine, how would it be, if I were in a new town without many friends. But at IIT Gandhinagar, it only meant that I had to go to office, connect to WiFi, back on Internet, pay the amount to various service providers, and you will have Internet connectivity even without WiFi from these two devices. (Of course, I wouldn't be able to use netbanking for making the payments, since that required me to type in a code which my lost phone would receive. But thankfully I remembered my credit card's second factor code.)

But now, I needed a phone. Santosh decided to work on twin strategies. Try to get Vodafone to issue me a new SIM. On the other hand, ask Airtel why my tab wasn't able to send SMS and make phone calls, and get them to fix the problem.

It appeared that Airtel problem was simpler. So he took the tab to an Airtel outlet. They checked the number in their computers, and pronounced the verdict. Since I had not used the connection for 60 days to make any phone calls or send an SMS, it had been deactivated. But why do they de-activate. How does it matter whether I have not made a call for 60 days or 600 days. Well, the all pervasive security reasons. If you have a pre-paid connection and you don't make a call/SMS for 60 days, it is assumed that you might have lost it, thrown it, or something, and it might fall into the hands of ISI, who will simply put some charge and use it for illegal communication.

But, wait a minute. You know that I have been using it daily. Every day, I have connected to the Internet using that SIM. So your assumption that I might have lost it is obviously wrong. But, Sir, this is what the computer is saying.

But, I just had a recharge 2 days ago towards phone/SMS, and I received an SMS from you that the validity of my connection has been extended to October 2014. Sir, the computer is not showing any recharge. In fact, if there is any recharge after the phone has been deactivated, that is not shown by the computer. (What he did not explicitly say, but meant was that I had lost that money, because I was stupid enough to recharge a dead phone.)

But, how come I continue to receive all commercial SMS messages if my SMS and Voice service has been blocked (and a point I did not want to argue - my number is on DND list). Sir, we reserve the right to send SMS messages in your benefit. Of course, and including the one which tells me that my service has been extended till October 2014. They will only do things for my benefit, and blocking my service was only to ensure my safety.

OK. So now, what do we do to unblock the SIM.The Airtel service agent was most polite. Sir, you only have to make an application, along with a proof of identity, a photograph, and an address proof, and it will be activated very soon. So Santosh went back with everything. No, Sir, you have to submit these things in any Airtel store in UP (East). But why can't you activate from here. Sir, each state is a different network. Of course, I know that. But it is the same company running the two networks. Sorry, Sir, we only take care of customers of Gujarat circle.

Do you think I can request someone in Kanpur to go to an Airtel store with all these documents, and they might agree to activate the SIM. No, Sir. They would like to see the SIM, to make sure that the SIM is indeed in your possession, and not with ISI, and you are not an ISI agent. Of course, national security is paramount. I felt secure for the first time since losing my phone the previous evening. I also felt that it was the right time to sell off the Bharti Airtel stock, which I have kept from their initial IPO.

So, the focus shifted to Vodafone to get a new SIM. All the Vodafone stores and mini stores that Santosh called told him that the new SIM can be issued only by a store in UP (East). Why is that. Is the SIM in Gujarat any different from a SIM in UP. Santosh being a very resourceful person, called up a senior executive in Vodafone, and explained to him the problem. Thankfully, the person agreed to help. He asked a Vodafone store to issue a SIM, and sent out a request to his counterpart in UP (East) to register that SIM. He told us that it will get activated between 4 and 24 hours.

When it did not get activated for 8 hours, Santosh called him again. He checked and told him that it has been activated and the phone should be working now. In the meanwhile in the expectation of getting a new SIM, I had bought a new phone. The phone was showing the signal strength was very good, which meant that the SIM had been installed correctly. But we did not want to disturb the executive in the night. So waited till morning. In the whole day, there were numerous calls to him and his office. And every time, we would be told to reboot the phone, do this setting or the other, check your roaming, and so on. The whole day was wasted like this.

On 3rd day morning, the SIM was still not getting registered on the network. But surprisingly, if someone were to call my number, they would get a ringing tone, and not the message that the number is blocked. When Santosh rang them up again, this time they could guess the problem. The UP East fellow who received the communication from Gujarat circle executive had activated the old SIM and not the new SIM. So the guy who has my phone could have made any number of phone calls if he wanted (though I checked on my call log on Vodafone site, there is no calls made from that phone since Friday evening). And once this problem was identified, they took only a few hours to fix it.

So I am back in action after 72 hours, only because Santosh knew a Vodafone executive, and relentlessly pursued him by making about FIFTY calls to him. Otherwise, Vodafone stores were no different in their response than the Airtel stores. Under normal circumstances, I would have had to wait till I returned to Kanpur, and made a personal trip to one of their stores, same thing that Airtel told me.

I have been a customer of Vodafone (and the earlier avatars of Hutch and Essar), right from July, 2000, when the call charges were very high and one paid for incoming calls as well, and have had earlier complaints about their service. The only reason to not change is that their competitors are no better.

But I wonder, if Government of India can issue an emergency passport in two hours in a foreign country, based on limited identity information that I have, why can't a telephone company issue a new SIM (or activate an old one) in 24 hours within the same country, when all sorts of identity proofs have been provided.

In these 72 hours, I realized that while not having a phone is certainly an inconvenience in today's world, but the inconvenience gets exaggerated because others don't expect you to not have the phone. So the maximum trouble I had was in locating the driver at the station, since they have forgotten the art of putting up a placard. They will send you an SMS with their number, expect you to call them up and discuss the place where you will meet them. If you have to meet anyone, they no longer wait for you at the pre-decided location, but will sit in a Coffee shop and expect to be called when you reach there. And, of course, the 2-factor authentication has made the mobile phone indispensable.

Wednesday, August 1, 2012

UGC Regulations on PhD and Autonomy of Universities

As I have said in my blog many times, our regulators feel comfortable in homogenizing everything. All boards must be similar. Common entrance test. State wide technical universities with 100s of affiliated institutions. No one is allowed to experiment. No one is allowed to be exceptional. Mediocrity rules.

Continuing with the process, UGC came up with UGC (Minimum Standards and Procedures for Awards of MPhil/PhD Degrees) Regulation, 2009. The goal is always noble, to set the minimum standards, and of course, anyone can try for excellence. These regulations are based on the assumptions that there can be no accreditation based system in the country, that UGC is incapable of monitoring quality and use its vast financial resource in a carrot/stick approach to encourage quality. The only way quality can be achieved is by specifying to the level of fine details on how universities should run their PhD programs.

UGC keeps coming up with its regulations. Typically, when we study the regulations, we find that many things are already being done. Amongst other things, there may be a good idea that we can adopt, and some things may not be useful in our context, which we ignore. UGC is obviously not very happy at the last part. How dare anyone has any autonomy in this country. So they have come up with an interesting idea to make us fall in line. Every university (including IITs) must issue a certificate to its PhD graduates that these regulations were followed by the university throughout the program of the student, right from admission to the graduation. If such a certificate is not issued, then the universities must not consider such PhDs as worthy of recruitment. So, now, we have the following options:

  1. Follow all regulations, and issue the certificate along with the PhD degree that we followed all regulations. This will certainly make UGC very happy.
  2. Don't follow all regulations, but issue the certificate anyway. (This will be the path taken by several universities, including some IITs. I am already aware of such certificates.) This will seemingly preserve the autonomy of the university, though by agreeing to do this, one has already surrendered one's autonomy.
  3. Don't follow all regulations, and don't issue the certificate. (Hopefully, many universities will ignore lack of this certificate, but certainly some PhD students will feel uncomfortable with this.)
Please note that the issue is not whether those regulations are reasonable as "best practices" or not. Even if they are reasonable, for a regulator to interfere so much in the running of a university, and try to micro-manage every university in the country, is totally undesirable and not a very healthy practice. Such regulations will only ensure that universities don't do any thinking on their own, and encourages a culture to do the minimum that a regulator will ask for.

Actually, as best practices, they are reasonable for a university to adopt. However, when you make them mandatory, with the requirement that a certificate of following these regulations be issued, then one has to start looking at them carefully, and there will be a few points where there may be some intended or unintended differences.

Since UGC assumes that most people on the face of this earth do an MPhil before PhD, which is not the case in engineering education, the regulations are written in a way that sometimes it makes no sense for us. For example, the way I read them, if we admit an MTech into a PhD program, we should either demand that the candidate gives GATE again (s/he already must have given GATE after BTech), or we have an IIT wide exam for PhD admission (and not a department level exam that we currently have). They require us to have a compulsory course on "research methodologies" even though most of our admissions are for those who already have some research exposure. There are several other such differences, which, I am sure, if UGC understood engineering education, particularly in IITs, they would not have written the way they have written, but now that they have written that way, it becomes difficult for us to either accept them or give a certificate that we follow them.

And such confusions have come about because the regulators want to make everything homogenous. Their pursuit of mediocrity does not make them happy in just issuing "best practices" documents but "regulations" which must be forced on everyone.

Tuesday, July 31, 2012

Fire in Tamilnadu Express

Nothing is worse for a rail fan than the news of a railway accident, particular when innocent lives are lost. And so was the news of S-11 coach of 12622 Tamilnadu Express from New Delhi to Chennai burnt by fire yesterday early morning, around 4:30 AM. Since the TN enjoys a non-stop run from Vijaywada to Chennai, there was no stop for the next three hours (till the train reached Chennai), and hence everyone must have been fast asleep, when the fire broke out.

Here are some links to the news reports:

Times of India
Indian Express
The Hindustan Times

Different accounts report the number of deaths to be between 30 and 47.

The usual things have been done. Everyone has paid condolences. Railways have announced financial compensation. (They can at least change the vocabulary. It seems that they are valuing a life at 5 lakhs. Instead, they can say something to the effect that they are providing financial support to the next of kin.) Everyone will be making an air dash to see the burnt coach, as if it has become a museum piece. An inquiry has been commissioned. And, by tomorrow, the traffic will be normal on the busy route. Media will also forget that yet another accident had happened.

Hints are being given out that it was either an electric short circuit, or a sabotage. It is being suggested that some Railway officials heard a loud bang (which apparently supports the sabotage theory), though some survivors from the coach and passengers in nearby coaches seemed to have heard nothing of the sort. Some others are talking about sound of crackers.

How does it matter whether it was a short-circuit or a sabotage. The important question is why did it kill so many. And that question, I am not seeing debated in the media, but only on the railfan website, It is not being debated, because it brings forth uncomfortable questions for the Railways to answer.

Shouldn't there be electrical safety devices which should detect a short-circuit and break the circuit before a fire gets ignited.

Shouldn't there be some emergency light system in the coach, when there is an electric failure, so that there can be at least a bit of light for people to plan escape.

It has been pointed out that the emergency window exits remained intact in the burnt coach. Why could people not open that. How do we make it more user friendly, so that they can actually be used in an emergency like this one.

Why couldn't people escape to the neighbouring coach. The train has a vestibuled rake. Was it because, it was full of trash. We notice in many trains that the pantry staff keeps trash in the vestibule area. Was the passage closed for the night. What is Railways going to do to make sure that vestibules provide unhindered passage in such emergencies.

Was the coach over-crowded, as is the norm these days. Many unauthorized passengers sleeping in the passage, again, hindering the exit path. Can something be done to make sure that the passage area in the coach is free of any encroachment, particularly in the night.

Aren't the coaches and the upholstery supposedly made of materials which are fire retardant. Then why did the fire spread so quickly, that people did not have time to escape. Have Railways done any testing of these materials.

Who and what caused fire is certainly important, but the more important issue is whether the loss of life was avoidable in such an accident.

My prayers are with the family and friends of those who lost their lives, not because of a fire, but because the fire spread quickly, and the coach design and other practices did not allow a quick exit.

Monday, July 30, 2012

MOOC: Massively Open Online Courses

Lately, a colleague of mine, Prof. T V Prabhakar, has been educating me about MOOCs. He has been forwarding various links on this, including the recent ACM article. Of course, I was aware of the AI course offered in the MOOC format by two Stanford faculty members, Sebastian Thrun and Peter Norvig, in which more than a lakh students registered.

The ACM article is by a professor of University of Massachusetts (Fred Martin) who offered the AI course last fall, using the Stanford AI course as the lecture material. The experience of the course taught by Fred is very interesting, and points to the pedagogical changes that we may be seeing in future. The lecture material will be available on the net, and the class will be essentially a face-to-face discussion session and for any incremental learning. Of course MOOCs also have their own online forums, questions and answers, links to additional information, and so on.

What will be impact of MOOCs on education in India. Very difficult to guess. But let us make some guesses.

A course by an excellent professor in a good university in some part of the world even in an online mode is likely better than the corresponding course offered in an average engineering college in the country, even though latter is a face-to-face communication. Considering this, a college in India may just ask its students to register for the MOOC, and ask students only to give the exams of the university. However, this option does not appear feasible. The language of foreign professors may not be very easily understood by our students because of differences in the accent and pronunciation. A typical student may not have sufficient writing skills to ask questions on the online forums. But, most importantly, there will be differences in a course offered in another university and the syllabus of the local college. And considering that we mostly operate in the archaic affiliation model in India, allowing any variation in syllabus is simply out of question.  Because of this, MOOCs will be used just like the courses on NPTEL. Students will just watch lectures on specific topics from the archives, and institutions will not replace their courses with MOOCs.

Of course, an institution may take care of the exam part by suggesting to the students that parts which are not covered by the MOOC will be taught separately and only that part will be taught. But the language difficulties on one hand and strict university regulations on the other will ensure that this does not happen.

I believe that language difficulties will go away soon. Many courses do have subtitles. Once you have subtitles in a course, to allow a text-to-speech software with a different accent should not be a problem.

Can a MOOC be used in the same format as by Prof. Fred Martin in the ACM article. That is, asking students to go through the lectures, but having an interactive session with them, what is known as the "flipped classroom" model. This also looks unlikely because to conduct an interactive session based on lectures by someone else is not an easy task. It is not like conducting a tutorial where the instructor has provided detailed material to be specifically discussed in that tutorial session. This interactive session invites questions and discussions on almost anything discussed by the lectures on MOOC and beyond. Again, such expertise is not available amongst the faculty of majority of the engineering colleges of the country.

I foresee that in the next 4-5 years, MOOCs will be used primarily by people outside the strict curriculum boundaries. These may be working professionals or even students who are interested in a particular subject and just want to learn it without necessarily this counting towards any degree.

Within the educational institutions, there may be some (like IIIT Delhi has done), who may allow students to go through a MOOC, and have an internal mechanism to test whether the student has indeed learned the material and give him/her credit. This way, they can expand the range of electives that their students can take.
Of course, only universities which are nimble and can approve any new course quickly, can take advantage of MOOC in this way.

My guess is that the scenario will change as and when professors in Indian institutions start offering MOOCs.

Here are some websites for looking for these courses:

Another link from CACM: The Coming Tsunami in Educational Technology by John L Hennessy.

Sunday, July 29, 2012

IIT Gandhinagar: 1st Convocation

Two students came to my office in the morning and asked me a simple question, "are all IITs similar?" And instinctively, I said, "No, IIT Gandhinagar is different." Well, may be all IITs are different, but when I visit IIT Gandhinagar, and I do that very frequently, it just seems like a very different place, a place which seems to defy Newton. There is no inertia there. There is no resistance to change, and as the cliche goes, the only constant there is change.

I visited them again last Sunday (22nd July), to attend the first convocation of the Institute, to see the pioneer batch of the Institute receive degrees, less than a month after they were notified in the Gazette of India as a new IIT.

The first thing that struck me was the dress. To quote from their convocation brochure, "the stole robe, designed specifically for the first convocation of IITGN, is a combination of the western academic attire and Indian free flowing draped garments." No one seemed to be missing the black robe and the cap.

The Chief Guest was Mr. N R Narayana Murthy, one of the most inspiring leaders of our times, and a role model for the generation that is represented by the graduating batch.

When I went through all the documents that were distributed, it was amazing to see how the young Institute and its pioneer batch has performed. Out of 86 students graduating, 8 were going for PhD programs (7 abroad, and 1 in IIT Gandhinagar itself). 10% of the batch going for PhD has not happened in older IITs for may be a couple of decades. Even Mr. Murthy mentioned that it is unheard of for Caltech to admit two students into the PhD program of their same department from the same external university. There were four more who were going for MS/MTech. There must be something right that the Institute has done over the last four years to enthuse so many of them for higher studies in engineering.

Six graduates have set up a technology innovation start up in the area of distributed computing in IITGN Incubation Center. And one graduate is spending time in improving the machine to make incense sticks (which he had designed himself earlier in the program, and which is helping the poor in Ahmedabad).

The batch had options of either studying for a vanilla BTech or embellish it with an honors or a minor. Only 37 out of 86 decided to graduate with a BTech program, which shows that students were really very keen to study engineering, very unlike the complaints that we hear from faculty of older IITs that students are only interested in a degree and they want to pursue MBAs.

The convocation brochure and the Director's speech talked about several innovations that IIT Gandhinagar has done in its march towards excellence. The focus on humanities and social sciences in the engineering curriculum has been amazing. They have the highest share of HSS courses in the engineering curriculum of all engineering institutes in the country. There is a comprehensive interview of every student in the institute every semester by a panel of faculty. This helps in detecting problem areas early and also identify strong points of the students so that they can be advised accordingly. Short courses of 8-10 lecture hours by distinguished visitors, which give the students 1 credit towards their graduation requirements is another innovation, which is actually being implemented and is not just on paper. They are building relationships with several good institutes nearby by encouraging their faculty to forge research collaborations with IITGN faculty, by encouraging their students to spend a semester or two at IITGN and earn credits which their parent institute may consider towards their graduation requirement, and so on. They realize that for having an excellent research program of their own, they will have to ensure that there is a research ecosystem around them. There were too many initiatives to mention here. Even though I have been going there so frequently, and even been part of some committees, when all the initiatives of the last 4 years were put together, it still was very surprising to me to see what IIT Gandhinagar has achieved.

(Disclosure: I am a guest professor of IIT Gandhinagar, and all my visits to IITGN are supported by them.)

Website of IIT Gandhinagar:

Sunday, July 15, 2012

Performance of Droppers in IIT Kanpur

IITs permit only two attempts in the Joint Entrance Examination. You can sit for the exam in the year you pass 12th, and the year after that. Why not be open and allow everyone. How does it matter if a student who passed 12th class 5 years ago can sit in JEE, pass it, and study in an IIT.

IITs used to allow this till 2006. So I took out data for the 2006 batch, and looked at their performance in the first year (when the courses are common, and hence no bias of some department being more liberal or tough or competitive would matter).

In 2006, if we consider the students who had also given 12th class in 2006 (that is no drop year), the number of students were 184, and the average CPI of these students was 7.9.

Those who gave 12th class in 2005 (one year drop), the number of students were 217, and the average CPI of these students was 7.2.

For 2-year droppers (12th in 2004), 81 students had an average CPI of 6.8.

For 3-year droppers (12th in 2003), 35 students had an average CPI of 6.4.

Remaining 18 students (12th in 1999 to 2002, or 4 to 7-year droppers) had an average CPI of 6.4.

If we just look at the students doing 12th in 2005 and 2006, more than 25% of the students doing 12th in 2006 have a CPI of 9.0 or higher, while less than 10% of the students doing 12th in 2005 have a CPI of 9.0 or higher.

I guess this is too small a sample and that too from many years ago to warrant any conclusions, but if similar data was available from earlier years from all IITs, and it showed a similar trend, it would justify IITs' decision in 2006 to allow only two attempts at JEE.

Tuesday, July 10, 2012

A Note on Admission Process

How should a university go about deciding its admission process? Before we can answer that question, we need to understand the goals of the admission process.

First and foremost, an admission process is a process of predicting who will perform well in the future. This is not a simple statement, of course. What does one mean by "perform well" and "future."
Performance could be getting good grades in various courses of the academic program that the student will do (and admission could be targeted for each individual program), or it could be an expectation that the student will get good grades in any program of the university that the student chooses to do. Clearly, if you are admitting a student to a specific program as opposed to admitting to the university and letting him/her follow any of the programs offered by the university, the predictor function is likely to be different.

Performance could also refer to students doing well in going to the next step. Some universities may specifically train its students for certain kinds of jobs, while others may want their students to go for higher education, and performing well may refer to getting a job or getting admission to a good university for higher education.

Performing well could also refer to what alumni achieve several years after graduating from the university. Again, it could mean success in a relatively narrow area, or success in a broader sense of the word.

As one may notice, design of a good admission process must start with deciding the goals of the university and what type of students does it want to admit. Is it interested in students who are likely to do well academically in a specific program for which it is giving admission, or is it interested in students who will be successful citizens of the world a couple of decades later.

Prediction is a very difficult process in any sphere, but when it comes to human behavior and performance, it becomes a much more difficult process. Past performance or knowledge of some topics can only predict so much. It is also important to predict whether the student will have motivation and interest to continue doing well in future. If the university is looking for performance in terms of long term success, then soft skills and life skills become important too. A uni-dimensional testing can hardly predict future performance or success. It works only when you have millions of kids interested in thousands of seats, and you never do a scientific study to figure out if you could have had “better performing alumni.”

Prediction process would normally take into account, not just the past performance, but also the circumstances under which that performance was achieved. For example, most universities, who have humans to go through admission applications, would consider 90% marks by an urban student with both parents rich and educated as being worse than 85% marks by a rural student with no role models in the family.

Is admission process all about defining the kind of students one want, and coming up with a predictor for that. Both these are very difficult problems, but unfortunately, admission process is even more complex than that.

For example, what if your predictor function results in a group which lacks diversity. The performance or success of individuals also depends on the peer group that those individuals are part of. It is well known that one learns better (in a broader sense) in a diverse group. What this means is that while the predictor function is operating on individuals, simply picking up the top N applicants may not be as good as offering admission to a few applicants lower down in that list, those who come from diverse backgrounds. Diversity here could mean students from different cultures, languages, religions, etc. It could mean that if there is a serious gender imbalance, and a more gender balanced class is expected to perform better, then introduce preference for the gender which our primary predictor function is not able to capture. If exposure to sports and cultural activities is likely to improve the performance of the class, then having a few students who are good in those activities may be considered.

Besides diversity, there may be other more controversial issues to be considered in the admission process. Consider the following. If a university has to admit 100 students, which it was going to admit based on the admission process built based on the discussion so far. Before it could offer those admissions, it receives an offer. If a particular student is given admission, then someone will do something to make sure that the quality of learning for the other 99 would be at a much better level. For the sake of argument, let us assume that that student has a decent record in whatever the university was looking for, but did not make the cut because the university could offer only a limited number of admissions. For example, if someone offers a Rs. 10 crore donation, which could revamp all the labs, stock up the library, increase the Internet bandwidth, attract additional faculty, or whatever.

This is a difficult decision. Obviously this improves the performance and success rate of the group that is being recruited by the university, which was really the goal of the admission process. We have already said that it is alright to offer admission to some students who did not make the cut based on our predictor function, because we wanted to have some diversity, give preference to sports, culture, etc., because we believed that that would improve the group performance. Now, this is yet another situation which will improve the group performance.

But, typically, no good university would like to offer admission to someone based on bargaining power of someone else. However, most universities would use admission as a carrot to get things which are likely to improve the learning and success of its students and alumni. So they may not offer admission to someone whose father is now offering to donate 10 crores, but they may offer admission to someone whose father donated 10 crores last year. (I want to re-emphasize that the student must be meeting all the expectations of the university in terms of his/her past performance.) Some universities have “political quota” whereby they may offer admission to wards of presidents and prime ministers and other top folks in the world, since it is expected that such associations will bring certain benefits to the university, which gets passed on to the students and alumni. Some universities have “alumni quota” for similar reasons. In Indian context, I have seen an example of a college offering admission to ward of an IIT faculty, if IIT faculty promises to join as a visiting faculty and offer some courses to the class. More generally, use admission to wards as an incentive to attract faculty members to the university. Again, in every such decision, the university has to think whether this is being done to improve the quality of education of the group that is being admitted.

In this context, I recall an incident at LNMIIT, Jaipur, just when I joined there in 2008. We used to admit students only through AIEEE ranks. One day, office of Mr. Lakshmi Mittal (the steel king, who was our Chairman and promoter of the Institute) sent an application of a student, and asked whether this student could be given admission. This was an excellent application. His 12th class performance was very good. He had represented a state in national games. He was very good at some performing arts, which I now forget. But his AIEEE rank was just a little below the last student whom we had admitted. He was the kind of student, we would love to have in our university. His family knew Mr. Mittal well and had approached him for admission. However, we had to inform his office that our processes did not have any scope for discretionary admission, and thus we could not admit him.

But this made us think that we should have some way of admitting such students. So we made a proposal, which essentially said that besides the 240 students that we admitted, we may admit up to 4 additional students based on criteria other than entrance exam performance. Two students were to be “management quota” but the recommendation would have to come from Academic Council, and two students were to be wards of staff of the university, and again would have to have good performance in something other than entrance exam (say, 12th class boards). We took the proposal to the board, where Mr. Mittal spoke against the idea of management quota but supported the seats for wards of staff. He said that management quota would result in pressure to admit weak students. His famous lines were, “to give admission to my relative would result in average performance of class going down, while giving admission to wards of staff would result in attracting good faculty, which will ensure that the average performance of the class goes up.” (Of course, LNMIIT later decided not to offer admission to wards of staff also, but that is another story.)

Coming back to admission process, if we look around us, we find that hardly any university has given a thought to the kind of students they want, and what would be the right method for selecting those students. Rank in a single entrance test is used for admission to a large number of universities and in very diverse set of programs. The focus of admission process is strictly on logistics and whether the process will stand a judicial scrutiny. And there is tremendous pressure from potential students, their parents, and also from alumni and faculty to keep things that way. In IITs, for example, it is not clear whether we are trying to predict good performance of the students within the academic program, or success later in life. If we are trying to predict good performance in the academic program, then how can the same prediction function be good for as diverse fields such as Computer Science, Physics, Mathematics, and Economics. If we are trying to predict success in life, then how can you do that without any aptitude test, language skills, other soft skills and life skills. I used to argue that JEE is so bad that it is difficult to think of a worse admission system (till IIT Council proved me wrong – lesson for me here – never underestimate others).

The students and parents in India think that it is their god given right to be considered for admission in every program in every university across the length and breadth of the nation. And since the process must be convenient to them, we must have only a few tests (if not a single test), common merit lists, joint counselling, and so on. Government and government appointed educationists look at vote banks. Alumni think that by changing the admission process, a message is being sent that the previous selections were not good. And faculty wants to make sure that their time is not wasted in court cases over admission process, which is a hugely exaggerated fear.

Frankly, if the nation wants excellence, it can only come about by several people trying out a lot of different things. Common processes can only lead to mediocrity and worse.