I really got the feeling as being in college when I yesterday hear d talks from the likes of Dr.Raj Reddy , Prof Jhunjhunwala, Dr. Kanade etc. I really enjoyed the discussion panel as well and I am sure it would have inspired many to pursue research as it surely did inspire me. I was not in the panel and I am sure I will not be asked to be there 🙂 but I wanted to add my 2 cents into the discussion and I spell them out here.
Almost 3 years ago I took a decision that I am not going to pursue research (or let me say it that I deferred taking up research for a very very long period of time) but if I had pursued research I would have followed this approach.
– Approach the best/favourite faculty/guide that is accessible.
– Get in the field of research that this faculty is working on and make it as your favourite
(NEVER EVER do it the other way around , it is a sureshot recipe for disaster)
– Study the history of the subject ( some also call it as ‘trawling the literature’)
– Learn about the most break through ideas , concepts & solutions
– Make a list of the most challenging and the longest duration unsolved problems.
– Pick one from the list and then solve it or attempt to solve it for the rest of your life.
Essentially choose a field of science and then push the boundaries in that field. It could be either basic research or applied research. To explain it a little more I will give some examples from different fields.
Communication Theory – The most fundamental ideas in communication is that information theory laid out by Shanon. “But does communication only involve the transfer of information,we know that in general we use the tools of communication to create a sense of presence rather than for just plain information transfer” – A great advance in this space would be to thus come up with a formal theory for presence in communication like the way we have a formal definition information ( -log( posterior)/log(apriori))
Signal Processing – A transform which obsoletes & replaces the Fourier Transform,( you can name the transform to yourself:) )
Speech/Statistics – Some model which is simple but more effective Hidden Markov Model.
A better method (computationally superior) that replaces the KL Transform in the data compression community.
Work to present a unified theory of informatics( information theory), cybernetics(wiener filter), systemics( turing machine)
Propogate the work of Godel further to improve/develop our mathematics to a level where we can have true AI rather than the sipid defintion/claim of the so called strong AI proponents.
Finance – Another example is that in the area of finance drop the now obsolete & inaccurate model of Gaussian curve and redefine some of now orthodox metrics. Use some new models that better represent the real world that we live in . Possible approach could be to investigate the use of fractals. Infact this can be generalized to any area of science that uses the tractable & beautiful but inaccurate fat tail distribution .i.e Gaussian Curve.
Databases – A theory which unifies IR & relational Databases, a theoretical framework for semi structured databases like the way Codd gave theoretical foundation to relational DB
Computer Vision – A sound framework for Scale Space representation of iimage, humans use this mechanism to understand images in brain but a good scientific model is not yet available which if available would provide answers to many problems in vision.
Here are some example in the soft science
Economics. Our notion of copyright(property rights) is binary but in the digital this does not hold good , we need more thought capital spent on redefining the notions for the digital age.Geeks know it intuitively that value flows from inclusion as much as from exclusion and that fact should be captured in the new models for intellectual property rights that should replace Copyright.
Another example in the soft science is that fast eroding value of core competencies as the boundaries of firm getting redefined due to new technologies(open source) & consumer choices. What this necessitates is the creation of edge competencies.
Real knowledge value gets generated only when researchers work to push the boundaries of epistemology. Some would argue that the nature of some of these problems is that they don’t get solved for decades and centuries then why bother. My answer is that if they are simpler problems then technologists are there who will solve them ( a researcher is not needed to solve simple problems) and also when one makes an attempt to solve a difficult problem then one solves some other problem which is of great importance. The best example that I can think of is John Nash , he was obessessed with solving Reimann’s Hypothesis but could not solve it but instead gave to the world Nash Equilibirium in Game Theory.( which is one of the most profound inventions in the last 30 years)
To summarize a researcher should increase the edge of the boundary of science otherwise he should’nt be a researcher.
Update: Read Casy Zak’s alternate viewpoint some of above thoughts.