Feynmann is not too much off the mark here too

The previous post from Feynmann's book actually quite nicely summarizes the scene in India too.

Education here is much more "lexical" than "semantic." I've many times retorted that our "best students" are basically next-generation database engines -- who can retrieve paragraphs and programs given keywords.

I once remember giving the class a project problem, that involved building a model, based on observing a set of events. The main objective behind this exercise was to help the students think through this problem of model-building from first principles.

However later in the day, some students caught me in the corridor and asked me whether such problems have been addressed elsewhere, and "what can they read up." And I made the mistake of dropping a few keywords like "learning automata" and "grammatical inference."

That was it. The next thing that happens is that these keywords diffuse through the entire class. Google searches are made based on these keywords and papers retrieved. Many students also find that I myself had written a paper on learning automata some years ago.

So the final project evaluation becomes one big pain for me, giving me a splitting headache at the end of the ordeal. Over and over again, I see people having blindly implemented my paper or some other paper on model-building. They have copied formulas verbatim from the paper and unable to explain the formulas or their significance. And a large majority of the class just implements my paper. In fact, the exact problem statement that I'd specified was slightly different from what my paper was about. But this didn't seem to matter.

I even heard that some students decided to implement my paper because they felt that it would flatter me and make me more amenable to give them better marks. Our people-centric world-views at work again. Everything has to reduce to a social predicate: personality, feelings, moods, etc.

I think there was just one group (among the set of 36 or so project groups) who seemed to have actually thought through the problem and had come out with their own approach, that was not some delta-x change over some existing paper. Their approach may have its flaws, but I'd bet that these students understand the problem space much better than those who just implemented existing algorithms. These guys would be in a much better position to tackle similar problems or see the same problem semantics occurring in several other situations.

Well, I could go on and on about several other examples, but what's the use.. ;-)