Difference between New IIT and Old IIT

From their inception, the Indian Institutes of Technology (IITs) have been seen by state governments as prizes to wrangle from the Centre. The location of the very first IIT was to be decided by the Sarkar Committee but even before a final report was out, the then chief minister of Bengal, B.C. Roy, convinced Nehru to pick his state. So, based on a draft report and a decision to have Hijli Jail house the institution, Kharagpur became the first home of an IIT. This was 1951.

Over the next decade, IITs sprung up in Bombay, Madras, Kanpur and Delhi. It took thirty years and Rajiv Gandhi’s support for the next one to come up at Guwahati in 1994. The last of the ‘old’ IITs was in Roorkee, converted from an existing college in 2001. The next batch of IITs came up in 2008 and 2009: Bhubaneswar, Gandhinagar, Hyderabad, Jodhpur, Patna, Ropar, Indore and Mandi.

This ‘new’ group was rounded out with the conversion of the engineering department of Benaras Hindu University (BHU) into an IIT in 2012. The latest batch of IITs was announced over the last two years: Palakkad, Tirupati, Bhilai, Goa, Jammu, Dharwad and the Indian School of Mines, Dhanbad. This takes the total to 23, which covers most of the country except India’s northeast.

In the beginning, the IITs acted primarily as teaching institutions for undergraduates. But for the last few decades, they’ve been transforming into ‘world class’ research universities. A recent paper in Current Science by researchers from South Asian University (Delhi), BHU and the National Institute of Science, Technology and Development Studies (Delhi), analysed the research performances of the first and second batches of IITs over the last five years with interesting results.

Note: The data for IIT (BHU) or Varanasi is only from when it was converted till 2014 – about two years.
While the difference in scale between the output of the older and newer IITs is stark, it’s clear that for the older IITs, the last 5 years have been a hugely productive time in terms of research. Guwahati (and Roorkee to a lesser extent) has seen the majority of their papers come in the last five years despite being operational for much longer. Comparing this data from the period 2010-2014 (a five-year period) with data from 1999-2008 (a 10 year period) in an older paper, the picture becomes even more interesting. It isn’t ideal to compare periods of unequal length but the lack of data in the public domain is a major constraint.

For Kharagpur and Madras, the five year period from 2010-2014 has seen them almost equal their productivity for the ten year period from 1999-2008. Kanpur hasn’t showed much of an improvement but both Delhi and Bombay show a growth in rate of research output.

In terms of citations, each of them for which data is available (except Kanpur) show a much higher number in the recent five year period compared to the more distant ten year period. This data is hard to interpret as it could be the result of factors outside of the control of the institutions themselves such as an improvement in the indexation mechanism, improvement in access to research, global rise in citation numbers, etc. But relative observations can be safely made. Previously, Kanpur seems to have been the national leader but now it seems to have slipped to 5th place, just above Roorkee. Kharagpur (which now leads the pack), Madras and Roorkee seem to have grown the most.

It’s relevant to note that IIT (BHU) is being compared with a period before it was converted. While we don’t have the data to make any judgements, the fact that it has exceeded its output of ten years in just the two year from 2012-2014 is weak evidence that there was a ‘reputation boost’ that came with being turned into an IIT.

To clarify our analysis, this chart limits its focus to the last five years and throws citations into the mix. Roorkee’s actual age (as it was technically founded in 1847) shows through. Its performance is almost exactly the same as the older, more respected Kanpur. But a clear segregation is still visible: the older IITs to the top-right and newer ones to the bottom-left.

But faculty sizes vary greatly across these institutes so looking at papers per faculty member and citations per paper will give us a fairer basis to compare performance.

Now we see that despite having relatively high citations per paper (only Bombay has more), Guwahati has much lower number of papers per capita than the rest of the older IITs. In fact, the newer IITs at Indore and Ropar have been performing on par with the older IITs in terms of citations per paper which is very promising. While IIT (BHU) seems to be quite productive in terms of quantity of papers, it is clearly underperforming in terms of quality as measured by citation numbers, lagging behind the other members of its cohort. One study that analysed BHU’s research output for the period 2004-2013 found that 61% of paper were published in journals not included in the Science Citation Index Expanded and almost 75% of their papers were not even cited once. Another study found wildly different results which seem to be too high to be accurate.

One explanation for Bombay’s leadership in citations per paper is its large number of international collaboration papers. These papers by dint of their global nature tend to be cited more and are favoured in global university rankings.

Interactive version

The above chart shows Bombay having the most number of collaborations almost across all the countries listed. While the United States is clearly the country’s biggest research collaborator, Germany, the UK and Canada appear to be other major sources of collaboration.

The paper also provides information on the subjects in which the IITs are publishing their research. While focused on engineering education, a large number of their research papers seem to come in the areas of physics, chemistry and mathematics. But it’s relevant to note that while it looks like engineering isn’t a priority that is partially because it is broken down into sub-disciplines. If they were aggregated together, the total amount of research in engineering would look more impressive.

Professor Milind Sohoni of IIT Bombay takes a more critical stance of these priorities. He’s been a regular campaigner against the ‘internationalisation’ of research i.e. of research that takes what’s globally trending as its subject. Sohoni argues for an engineering education that goes back to its roots as a discipline that solves local problems – especially in water, sanitation, transport, etc. Since they count for a majority of the centre’s higher education spending, a renewed focus on finding engineering solutions for major infrastructural challenges would be well-received. But to avoid blinkered solutions without any connection to the ground, engineering won’t be enough. Sociology and a more interdisciplinary approach will be required.

Interactive version

As a concluding point, this analysis has very particular implications for the latest batch of IITs that were started over the last two years. It shows that it will probably take more than 10 years for these institutes to even register on the Indian research landscape. For them to become global players will require much more. Looking at the performance of IIT Roorkee and IIT (BHU), there is some evidence that the strategy of converting existing institutions works better than starting a new one from scratch. But most of the newest batch (except for the Indian School of Mines, Dhanbad) don’t have this advantage. It’s also unclear whether learnings from the other IITs are being utilised to speed up their development, other than the general mentoring process. It remains to be seen whether state governments will be happy just having IITs or whether they’ll push for them to become centres for research and change.