In a post filled with provocative ideas, Arunn argues for the use of GRE (or, presumably, another exam like the GRE General Test) for graduate admissions. Currently, in engineering (and to a smaller extent, in the sciences too) GATE is the exam of choice at all our top institutions, and this is really unfortunate.
GATE started its existence in the mid-eighties; I think I took the second edition of GATE in 1985. It was originally designed as an entrance exam for the masters programs in engineering. In those innocent days, the use of GATE for PhD admissions was out of the question, simply because a masters degree was a pre-requisite for getting into a PhD program.
But that era is gone; most institutions (including IISc) now have a provision for undergraduate degree holders to gain "direct admission" into the PhD program; these students have to fulfill a bunch of requirements, and a decent score in GATE is one of the first level filters (followed by, typically, an interview). As Arunn points out, GATE is not the right exam to use; a more appropriate exam -- like GRE -- would emphasize basic mathematical, statistical, analytical competence.
Arunn's arguments against GATE cover the many different ways in which it hurts our PhD programs; here's one:
... [S]ince [GATE] is a subject exam and involves concepts and number crunchers (questions) that require a thorough brushing up of the subject basics, potential students who have finished their UG degree some years back (and working somewhere) and want to take the examination are so apprehensive of what they need to study that they don’t even give it a try. This, even though they yearn for higher education.