The quality of the IITs - some (anecdotal) supporting evidence

Via Naveen Koorakula, who also happens to be an IITian, I came across a link to this somewhat peculiar story on a conversation between an IIT graduate and a rickshaw driver. I don't have the time to translate it (the conversation is in Hindi) but it is really quite the interesting chat.

I would like to present this as supporting evidence of my previous claim regarding the public reputation of the IITs. I am, of course, using public information (the reputation thing) such as this, and so it is theoretically possible that there is some privately held information which would show that this public information is, in fact, false. But given the transparency of the admission process, this seems highly unlikely. It is possible that someone is somehow tampering with the admission process while being careful enough to leave no traces of the act. But evidence for this would be very easy to find - from time to time, you would get an utter idiot who somehow manages to sneak in - but evidence such as this has, thus far, not been forthcoming.

It is very hard for me, therefore, to reject the hypothesis that the public reputation of the IITs is due to its high quality.

=+=

There were two rickshaw-walas vying for our business when we wanted to go to Sankat-Mochan temple in Benaras. I agreed to go with the one who was about 20, seemed like a regular young rickshaw-wala, but I found something interesting about this fellow in his eyes. I was not proved wrong.
He wanted Rs 50, we said Rs 30. We settled for 40. Here are the highlights of the conversation that ensued while he rode the rickshaw:

Have the subsidies given to IITs been a good investment for the Indian government? - Further comments

Five further comments on the IIT post (I am moving my blog updates from Thursday into a post of its own and have added a fifth comment):

Comment 1: A very short analysis of what is wrong with Atanu Dey's argument : He assumes that education in the sciences does not have any value over and beyond getting into a university, institute or college, and he also assumes that education does not have sufficiently large positive externalities. That $1.3 billion estimate by Atanu Dey is a measure of value added to the Indian economy, not value destroyed. If the IITs were not there, this sector of the economy would not exist and so that is a net gain to the economy as a whole. The value of the education that students in these coaching classes get should be measured not in terms of whether they actually get into IIT but in terms of what value they get from it over their lifetimes.

Comment 2: Fixed up some typos. Note that $1.3 billion is a measure of the value added to the Indian economy, not an estimate. A measure simply gives you some idea of the variable being measured whereas an estimate must be at least somewhere in the ballpark of the actual value of the variable being measured.

Comment 3: It is not really very time optimal for me right now - given that I am on a long-awaited vacation and given my various other commitments - to spend any more time on this, and so I am going to be brief. Here are some very brief overall remarks on the matter of the IITs.

1. A more rigorous analysis is called for to make the sort of bold claims that Atanu is making: It would be useful to take a look at the balamce sheets of the some of the companies I mentioned - Infosys - et cetera to get some idea of the size of the numbers in question when comparing costs (the cost per year of umdergraduate education vi the IITs) and benefits (the net taxes being paid by the IT companies in India). At any rate, to do any convincing on this matter, a more rigorous analysis is called for which makes clear what, if any, were the assumptions made in the analysis.

I am prepared to agree with almost anyone but I don't think I have enough reason to do so in this case. If you read any economics paper in any of the management journals, you at least have some idea of where the disagreements may lie. These posts are surely not well argued and seem to be full of holes. Now, whether these holes cold be later papered over is not the question. The executive summary, the thesis and the supporting arguments have to be front and center in the posts. Furthermore, there has to be some technical paper of some sort somewhere that tells me what the model being used is. So far, I have found very little.

2. Quick-and-dirty back-of-the-envelope calculations seem to indicate the IITs to be a good investment: Basically, here is how I did my back-of-the-envelope calculation. I just took a look at the balance sheets of Infosys and some of the other software companies, and used that to get some ballpark estimates of the benefits of the IITs. I wish I had more time to write up an actual post, but I leave this as an exercise for the reader. Just draw up an Excel spreadsheet listing out the cost factor, apply some time discounting of costs in the past and add some reasonable premium to just be conservative. The aim here is to get a very rough ballpark estimate of the net cost of the investment. Assume that a portion of the taxes paid by (I sometimes hate to use the word 'rent' because it is so non-specific in terms of actual numbers) some of the software behemoths is attributable to the 'founder effect'. (Nobody wanted to invest in Narayana Murthy's company when it was small. Trust me on this one. He came to a number of professors' faculty houses at IIMA to get some initial investment for the firm now known as Infosys. Very few wanted to invest.) This is because the only reason he even got his foot in the door was because he had an academic background from the IITs. You now have some idea of the benefits accrued. By way of comparison, Pakistan and Bangladesh don't have the IITs and don't have an IT industry to speak of either.

3. Small investment value: We are not talking about massive dollar investments here. There actually isn't a whole lot of corruption in the IITs and so the money actually gets spent on what it is set aside for. One must really believe that the Indian gvoernment is going to be able to better utilize the 30 million dollars or so it spends on each IIT since that is after all the opportunity cost of the investment. And to agree with the opportunity cost argument, one must believe - and I am not prepared to believe it - that there will be little or no corruption involved. 30 million dollars after the pigs have fed at the trough would be something like 5 million dollars - if you are lucky. (That is about 1/3th of the Las Vegas buffet's cost. Just sayin') I fail to see how this could bring about anything like the sort of results one can reasonably attribute to the IITs.

Comment 4: What are some empirically sound things that one could say about the IITs. a) They are probably never going to feature very high in World Rankings of Universities; b) The very high selectivity of the undergraduate program (at least until the mid-2000's) creates an undergraduate student population that is high in terms of motivation, scientific and/or technological talent and application, and given the very high level of selectivity, the undergraduate student population is - on academic terms- comparable to the student population in other institutes of technology such as MIT and Caltech ; c) The students in the Masters programs are harder to empirically analyze since less is known about them other than the fact that their entrance exam scores are high (the GATE entrance exams used to be multiple-choice and choosing people based on multiple-choice tests usually leads to skewed populations). This does not mean that they are less smart. It just means that we have a population that is harder to make strong statements regarding.

Comment 5: A couple of things that have not been sufficiently addressed in the arguments on IITs : the matter of the PhDs and this other business of public reputation. First, the PhDs. It is odd to completely ignore the PhDs in any analysis of an academic department or institution. What one would expect here is that (a) low investment research areas (projects in computer science can be very low investment) would be of excellent quality and (b) the quality of PhD theses in these areas would also be of very high quality. This is, in fact, what one sees. The Database Group at IIT Mumbai is very strong and is arguably one of the world's best. The PhD theses in theoretical computer science are often excellent as well. And second, the matter of public reputation. It would be very odd if an academic institution could sustain a very high public reputation over decades and yet have serious organizational failures. If anything, the IITs are a good example of how governments can run high performance organizations. If it is indeed the case that the IITs suffer from major organizational failures (just keep it real, folks!), then it is, by all means, okay to post about these. But nothing I have read so far in the couple of posts that I have gone through seems to suggest that major organizational failures actually exist. And finally, an acknowledgement : the phrasing "a good investment for the Indian government" is due to Prof. Ananth Raman at HBS.

Have the subsidies given to IITs been a good investment for the Indian government?

So I emailed Prof. Ananth Raman, Professor at Harvard Business School and alumnus of the Indian Institute of Technology Madras, regarding the IITs. The matter in question is whether the IITs have been a good investment for the Indian government. The short answer to the question is "Yes". Prof. Ananth Raman concurs with me on this matter.

My email to him is pasted below. Please note that the EdNotes (in italics) were added by me later in preparation for this blog post. Please note the typo in the usage of the word 'alumnae'. The correct term is 'alumni'. One of the great things about having been to Stanford and Harvard is that when you have typos in your emails, people do the auto-correction themselves. Stanford + Harvard joint reference FTW! 

I should add that I was at the Bacchanal buffet yesterday in Las Vegas. How much did the buffet cost to build? 17 MILLION DOLLARS. Two buffets such as this one and we are already there. In this age of easy, I wonder how anyone can believe that an investment in STEM education that amounts to less than 40 million dollars such as this one could be considered a bad one. Remember that we are talking about the Indian economy which is a >$1.8 trillion dollar economy in nominal terms and a >$4.5 trillion dollar economy in PPP terms. 

I fail to see Atanu's point regarding welfare maximization. How could the cost of the coaching classes be considered welfare decreasing even from a purely neo-classical point of view? I mean - let's forget about institutions. Let us do a simple purely self-interest based analysis. Here is a large number of people who independently, freely choose to spend money on a service. Would they do it if the option was not increasing their utility? And if it is increasing the utility of so many people, then how could the activity be welfare decreasing? I don't want to spend any more time reading through Atanu Dey's numerous posts on this topic. There are times when one must decide that there are other things that are a better use of one's time and this, frankly, is one of them.

P.S. This may well be the first ever post on the Indian economy which uses the sentence "Stanford + Harvard joint reference FTW!".

-+-

Dear Prof. Ananth Raman,

I trust this finds you well.

I recently came across a blog post by Berkeley Ph.D. Atanu Dey arguing that the IITs are : (a) "not all that they are cracked up to be" (which is a safe thing for anyone to say since it could mean anything); (b) a net welfare loss for the economy as a whole (wow! I mean, wow!) (EdNote: Exact quote from Atanu Dey : "One of my major points is that the subsidies given to IITs (and other tertiary education institutes) is regressive and welfare decreasing.") and (c) a drain on the economy since they represent a 'subsidy' and so the subsidy should, of course, be stopped (Link : http://www.deeshaa.org/2013/03/29/the-iits-are-not-really-what-they-are-cracked-up-to-be/).

This sounds like one of those loony rightwing talking points memos which propose tax cuts as a solution for everything, and I say this even though my political beliefs are somewhat to the right of center even in America. To put it really bluntly, I find the post to be almost entirely nonsensical. IITs have turned up in a couple (if not more) of our discussions in the past (and speaking of our conversations, I must confess that I did not quite pick up on your cues on IIT during my conversations with you - there are several reasons for that which I shall not get into). Anyway, as alumnae (EdNote: sic), I hope I may broach the matter of whether the IITs should be paid for the Indian government (EdNote: <stuff deleted>).

Hindu Studies Post 4 - Hello, I am George Ringo and I am your new Pope

No, seriously. How come Hinduism doesn't have anyone with the 'Authority' that somebody at the head of an organization such as the Catholic Church has? One can blame colonialism, racism, and a bunch of other -isms, but these explanations seem not to address the fact that colonialism is an -ism of the past, the fact that racism doesn't seem to have blunted the aspirations of the non-white contenders for Popedom, et cetera, et cetera.

As part of the main thesis of this series of posts, I would like to argue that there are actually three main reasons that Hinduism has until now lacked an 'Authority':

1. the constitution of India : India is, of course, the birthplace of Hinduism. The constitution of India has significant lacunae in its treatment of free speech and this makes it impossible for people to freely speak their minds;
2. the network structure of Hinduism makes it impossible for a large 'Population' of Hindus to agree on anything.  
3. A third reason is that, and this is important too, edcuated Indians are generally willing to give anyone who has an opinion a chance to speak, and are often more concerned about fairness than about correctness.

Now, correctness in academic inquiry comes from both knowledge and methodology and the problem is that very few of the intellectuals have good methodology. As recently as 2004, we have had someone wih the reputation of Pankaj Mishra strenuously arguing that 'Hinduism is largely a fiction'. Even a first pass on his article shows that the argument lacks any methodology whatsoever. Here is Pankaj Mishra in Axess magazine, his article now having been picked up by somebody at Columbia. 

Hinduism is largely a fiction, formulated in the 18th and 19th centuries out of a multiplicity of sub-continental religions, and enthusiastically endorsed by Indian modernisers. Unlike Muslims, Hindus have tended to borrow more than reject, and it has now been reconfigured as a global rival to the big three monotheisms.

The Scrooge of Internet debates

'Anand Manikutty Smackdown Watch' alert. As part of this blog's mission to provide only the best quality opinions, we have now decided to include a Blogger Smackdown Watch. The particular thread on the mailing list silklist linked to in my previous post on the Singularity on silklist turned out to be one where a bunch of people tried to do a smackdown on me. Thankfully, I was prepared. I will describe how I did it below - it is part of my short online lesson on Internet debating.

But first, as part of this blog's 'Anand Manikutty smackdown watch', I would like to alert you to a few facts : (1) the Singularitarian(s) on the list did not provide a set of hypotheses that could be taken as the basis of this theory. This was the case even after they were pressed to do so; (2) the 'innovation is not programmable'/'innovation requires humans' argument was never refuted; (3) each and every single counter-argument (*including* the procedural issue on not linking to the Wikipedia page) was demolished by me on the List that I used to maintain (linked here); (4) I am stingy (in fact, extremely stingy) in terms of the amount of intellectual capital I am willing to spend (that is, I won't spend any intellectual capital by agreeing that I am wrong in Internet debates unless I really am). As it happened, I refused to spend any intellectual capital in this particular debate, but still ended up coming out ahead.

In fact, thus far, although I have debated with a lot of people, even with Nobel laureates, not on a single occasion have they ever been able to prove me wrong. I almost always win debates. Except with the Nobel laureates and such where the debates are just a tie. So, I am just stingy, stingy, stingy all round. Also, note that despite what you might gather from just reading the thread on silklist, I was actually extremely nice to  all the people on silklist, even those who were not nice to me. The silklist is one of the few places left on the Indian blogosphere/forums where civil conversations on topics related to India can be had. You don't see any angry outbursts from me on this silklist thread and if you had read my List, you would have seen a lot of well researched and well thought out comments and, equally importantly, comments that were considerate and thoughtful.  Anyway, given that it is Christmas, I feel like calling myself the Ebenezer Scrooge of Internet debates would be appropriate. So in the spirit of Christmas, I will ask that you take everything I say in good spirit and that my comments should be taken in the spirit of 'giving' and 'shedding light'. I do intend to continue to remain the Ebenezer Scrooge of Internet debates since I believe I shed light on issues on many an occasion. Of course, I have been parsimonious in acknowledging defeat in Internet debates, and I believe that that is an excellent strategy. You should never have to apologize for any Internet comment as long as it was intended in the right spirit and as long as you are right.

The first rule of Internet debates is that there are no rules. The main problem is that in Internet debates, you just don't have that sense of organizational process that makes this sort of thing impossible in business settings. There is no phone number to call and no management hierarchy to resolve things via.   The other thing about Internet debates is that in game theoretical terms - and this is the most useful mathematical model here - they can be zero sum games. Once the debate has gone past a certain stage, one party must win and the other must lose. When a debate looks like a zero sum game, things dramatically change. You have to try really hard to either win or get your point of view across. It is almost like you have to be a sort of One Man Army. You can't rely on anyone else, you have to try hard to block everything and you have to improvise constantly based on what you subjectively see. Now, the big problem is this : if the debate is in a prominent source (also known as an Authority in technical circles) or is in a place that later becomes a prominent source, that particular debate from four years ago might start ranking very high on Internet searches. That may not seem fair, but then life isn't fair. (Just Google for my name and you will find that this particular thread that was practically inconsequential in terms of what I got out of it has only caused me a lot of pain. The benefit was zero.). Also, debaters are often not fair. Not only aren't debaters fair, time is also of essence. You can easily lose an Internet debate in a matter of minutes. Lose you might not only that debate but also your online reputation.

In fact, it is just like war in another way. Different strategies are called for in different situations. In a not-quite-fully-moderated mailing list-type forum like silklist (where there is at least some trolling and many, many of the discussions on matters of national policy go way off track notwithstanding the presence of at least one professor on the list), you are better off linking to your opinion from your own list because you can never control what other people say about you. All you can do is control your own reaction. Note that in this silklist thread, you need to read my comments in conjunction with the silklist thread. Otherwise, you are likely to make the same mistake as at least one person on the discussion did. He assumed that what I said on silklist was all there was to what I had said (but, of course, that was not true). The strategy I used was simple : realizing that there was trolling going on on silk list, I simply started linking to my own List from the first post on silk list (and it continued with every post from there onwards). This way, I prevented what could have been an utter disaster for me because moderators will sometimes moderate out your replies to a forum but not that of others. By maintaining control of what gets said about me, I got to control the outcome of the debate. And finally, one caveat : do not try this at home or at work. You might make people extremely unhappy. End of Internet debating lesson.

So anyway, Merry Christmas, people. Have a happy holiday season!

Update (Jan 3): Updated post a little to fix typos, etc.

A slight miscalculation

Time to look back at all the people who were busily preparing for the Mayan Apocalypse -- and laugh. The prediction that the world would end on Friday, December 21st, 2012 appears to have been a case of a slight miscalculation.

In Moscow 1,000 people who had packed into Josef Stalin's bunker were able to go back home after Armageddon was averted. 

Chinese authorities dismissed outright rumours that Jesus had reappeared as a woman somewhere in the middle of the country, and also denied that they had built an "ark" as a contingency plan. 

At Pic de Bugarach, the French mountain some had believed to be a place of salvation, the sun came out from behind the clouds and a flock of birds flew past as the official end of the world struck after 11am GMT. 

The mountain had been identified as an "alien garage" from where a vast intergalactic flying saucer would emerge to rescue nearby humans.

Comment - Email to William Nordhaus : the Singularity is not near

From the "End is Near" to the "Singularity Is Near". Below is an excerpt from my email to William Nordhaus wherein I expressed my skepticism towards the entire idea of the Technological Singularity itself.

-+-

Dear Prof. Nordhaus:

I have been thinking about the concept of a Technological Singularity proposed by some (Kurzweil, Vinge, Yudkowsky, et al.). I concur with Prof. Hofstadter in that a singularity does not seem likely in the near future. It is quite unclear to me that possibly the Singularitarians could mean by a process of 'continuous self-improvement' in artificial intelligence insofar as what has already been discovered.

<... stuff deleted ... >

Anand

-+-

I also emailed Tom Davenport on the same issue and told him about my reservations about this idea of a singularity. This was in the context of a post in the Harvard Business Review on IBM's Watson computer program. In the context of the Singularity, it is pretty clear what  Davenport's post is implying : innovation is not programmable. This is more or less the same point that I made over on the silk list (see this thread) as well.

When I emailed Tom Davenport about his opinion on the Singularity, his answer was what I expected it to be - that he was skeptical as well. He also sent me one of his articles which gets into more detail on what happens when machines are assumed to be substitutable for humans. That all of us (Davenport, me, Nordhaus) are arguing along very similar lines has convinced me further that the intellectual position against the Singularity is quite strong.

Mayan apocalypse: panic spreads as December 21 nears

From the Telegraph :

In the French Pyrenees the mayor of Bugarach, population 179, has attempted to prevent pandemonium by banning UFO watchers and light aircraft from the flat topped mount Pic de Bugarach. 

According to New Age lore it as an "alien garage" where extraterrestrials are waiting to abandon Earth, taking a lucky few humans with them. 

Russia saw people in Omutninsk, in Kirov region, rushing to buy kerosene and supplies after a newspaper article, supposedly written by a Tibetan monk, confirmed the end of the world. 

The city of Novokuznetsk faced a run on salt. In Barnaul, close to the Altai Mountains, panic-buyers snapped up all the torches and Thermos flasks.

Comment : the marginal utility of 2000 years of history versus 50 years versus zero years

The marginal utility of 2000 years of history versus 50 years versus zero years. This quantity is hard to quantify. May be it is zero. Maybe it is even negative. Maybe we cannot know what this value is for sure. But it is hard to see why it is necessarily positive notwithstanding Eric Hobsbawn comment that "Nations without a history  are contradictions in terms.'

Historians studying India often make a big deal of the fact that India has 2000 years and more of history. But what is the marginal utility for a country to have a long history? The marginal utility seems small. Below is a comment on an article by Namit Arora on "Three Quarks Daily" :

-+-

Nicely done Namit. Some of these Marxist historians are pretty, pretty good. The only trouble is that they cannot always be trusted. Every once in a while, they will say something supremely ridiculous.

Three comments :

1. Regarding the point on historical awareness in premodern India: this is already known. Nothing new is being said here.

2. Regarding the points on Gandhi : this is also already known. Nothing new is being said here. It is useful to bear in mind that we are talking about a fellow form the mid-19th century. Some of those guys were very smart and intelligent, but hardly 'enlightened with a global level of awareness'. Like, say, you and I. :)

3. ‘Nations without a past are contradictions in terms' : It would be useful to think of this in terms of the marginal utility for a country to have 2000 years of history versus 50 years of history. The marginal utility of a very long history seems to be negligible. If, say, there was massive global warming (and this is purely hypothetical) and oil were discovered in Antarctica and they decided to form 10 new countries there which decided to live peacefully and responsibly, they would surely be no worse having zero years of history. In fact, they would be probably be better off.  

-+-