Thursday, February 22, 2007

A rant on IITs, faculty and research

I usually don't court controversies but one of my previous posts has attracted the ire of a lot of people. I had questioned the need for research in IITs, and I got a scathing reply from an IIT prof saying "Balancing teaching and research is a difficult game, but asking IIT profs to be 100% (or even 60%) dedicated to teaching and letting "someone else" do the research is preposterous. I'd leave IIT sooner than I can catch a taxi if matters came to this stage." Ouch ! So, I'll now write about it, once and for all. A disclaimer: exceptions exist :)

IITs have a very strict policy in recruiting their faculty - a core faculty must have a PhD. At a time when the country is crying for teachers (esp. in mgmt education), the excuse that IITs are offering is that the number of PhDs are too low, and hence recruitment is difficult. IIT also evaluates its profs on their research (I think), and a former professor of SOM, who taught us microeconomics, used to wonder whether it would be right to decrease the amount of time he spent on preparing for the class (and he did spend a lot) and use it for research ? In other words, should a good teacher (which he was) suffer because he likes to teach and teaches well and wants to keep doing it ? What sort of a system do we have ? An intellectual (not necessarily meritorious) caste-system.

What does a PhD guarantee ? In my brushes with research, I've found something fascinating: People starting with research are advised to keep narrowing the scope of their topic, and to focus on doing it right, though in a narrow domain. So, PhD essentially guarantees that you achieve mastery over a very narrow domain. That is all. PhD students, called Research Scholars in IIT parlance, are taught that doing it in a proper way matters much more than getting the results. Means are infinitely more important than the ends ! You have to quote your sources, you have to follow a structured process, and all crap. Very high importance is given to the form. You must follow a definite system of quoting the sources (and there are rival systems), you must follow a proper line spacing and font-size and font-style. You better follow a proper bibliographic format. And don't forget to put your guide's name, for God's sake !

I argue that it (a pure PhD) makes a person narrow-minded and creates intellectual arrogance. No wonder most pure academic profs I've met don't treat students are co-learners (one of the questions in the course-end feedback form) ... And this also forces profs to offer courses about subjects that they know, instead of what we need.

This is not the disease of IITs only, but is prevalent across academia across the world. Andrew Tanenbaum, a famous computer scientist and a professor, still defends minix's micro-kernel design is better than Linux's monolithic design, because that "right way" to do it, despite being beaten pants off in performance and quality. I was taught in my college that I must not program using "goto", avoid C macros, avoid type-casts of pointers of different data types, etc. But, my work life would have been miserable without those features. Plagiarism is a crime in the academia, but its cousin called reverse-engineering is hailed in the industry as a great genius. Academics have no touch with reality - engineers and corporate employees do. Real life forces you to be tolerant of errors, and to be more pragmatic. That is why, I suppose, most of the successful people aren't from academia.

It is not difficult to comprehend this - the bad habits start early in the PhD.

We're blindly copying the American university system. Can't we innovate a bit ?

Do we want good researchers or do we want good teachers ? Since I can always learn theory from a book, I'd rather learn about writing operating systems from Linus Torvalds, rather than Andrew Tanenbaum (I would in fact, read the OS and networking text book written by Tanenbaum ... but that's it), and learn about investing from, say, a mutual fund manager than from an academic who has investigated all the models of investing. If I want to be taught, I'd rather be taught by someone with 10 years of relevant work experience and good communcation skills, rather than an academic with 100 journal publications. And please don't point to the consulting done by profs are work experience, for reasons mentioned here.

So, am I suggesting that nobody should have PhD ? Actually, no. My point is: I don't give a damn about the antecedents of a professor, but only his current ability to teach me. A good teacher, in my experience, is not necessarily someone who has mastered a subject, but instead, can inspire us to surpass him. If you want to know the attitude of a teacher, just read Richard Stevens' Network programming book - he teaches concepts as experiments that he conducts (armed with only a C compiler and a sniffer) ... and is humble to acknowledge and point out the mistakes. He takes you on an adventure and gets you curious. No wonder, I learnt more from reading that book than from any person.

The IITs already have a solution - visiting professors. But, often, such visiting professors are treated as pariahs by the system, since they don't have the "Dr." before their names. They are no better than contract employees. But it cuts both ways too. There is no proper measurement of their performance, no evaluation of their quality.... They are anomalies of the system.

So, for whatever they are worth, here are my suggestions: All IIT core faculty must have atleast 2 years relevant work experience. All PhD students must have atleast 5 years workex. btw, joining a company as a fresher is a good experience - you are not important, nobody needs you, yet you get shouted at, by your boss. It can dissolve superiority complex easily. Have a separate/parallel evaluation of research and teaching. If a prof spends 80% of the time teaching, the weightage of his teaching should be 80%. And keep the evaluation transparent to students. Allow an experienced person (10-15 yrs of relevant workex) to become IIT prof, as a sabbatical from industry. And in turn, send core faculty on sabbaticals to industry. You can knock off a lot of birds with a single stone: profs will get corporate remuneration, junior prof might get lucky ... :)

So, I'm done ranting about the IIT system. My next target - b-schools !

Labels: , , , , , , , , ,

Tuesday, February 13, 2007

Propaganda and the journalistic bias

I've mentioned before the journalistic bias, and the utility of Google News as a great leveller. This should be old-hat for Americans, what with alleged "neo-con" bias of FOX news. But the subcontinental media are proud of their independence and courageous reporting, so I had wait for a special story to highlight this in a stark way. And this week, I was presented with just such an opportunity.

Yesterday, the arbitration verdict of the Baglihar dam dispute in J&K was out. It quickly became a Google News headline, and grabbed my attention. On clicking "all X news articles" in the Google News homepage, I found some pretty juicy stuff. Here's how the newspapers covered it:

Pakistan terms Baglihar Dam ruling a victory Monsters and, UK
India has not violated water treaty with Pakistan -- World Bank expert Kuwait News Agency, Kuwait
Expert’s nod for Baglihar Hindustan Times, India
World Bank rules on Kashmir dam BBC News, UK
World Bank asks India to cut Baglihar height Pakistan Dawn, Pakistan
Baglihar Design; WB Neutral Expert gives Verdict in favour of Pakistan Pakistan Times, Pakistan
Baglihar: Pakistan wanted construction stopped Hindu, India
Neutral expert's clean chit to Baglihar dam, India
WB expert says Indian dam violating bilateral treaty People's Daily Online, China
World Bank validates Pakistan stand on Baglihar Dam, Pakistan
WB asks India to lower height of Baglihar Dam Daily Times, Pakistan
Dam stays, but not so high, rules World Bank expert Daily News & Analysis, India
'Decision to reduce dam height will not impact generation of power’ Daily News & Analysis, India
Baglihar cleared, India has its way Times of India, India
Neutral expert's clean chit to Baglihar dam

... and many more. Google News is indeed wonderful ...

I'm puzzled by the claims of "victory". If, obviously, both sides won the victory, then the solution was a "win, win" (as we MBAs are so fond of saying). In that case, why go into arbitration ?

If both claim that they have won, but the other has lost, then somebody must be lying completely or everybody must be lying partially. So what's the truth ? Isn't it the job of journalists to bring it out - report the facts and objectively analyze it ? If you want to tell the readers what they want, instead of what is, then openly say so. Or if you are too lazy to do some basic research (or Google search) and then report, then, atleast don't be partial.

QED. I rest my case.

Labels: , , , ,

Friday, February 09, 2007

Social network of the day

One of the most fascinating aspects of an internet-linked life is the buzz of the social networks. Just like some restaurant or a club or a pub (fortunately, I've not been introduced to the latter two) is considered to be the most "happening" place in a city, cyberspace has its own happening sites, where everybody who think they are somebody flock, along with the cyber version of the queen bees and queen bee wannabes.
The best and the most popular among them, is Orkut. I don't understand the psychology behind scrapping, and I can safely say that the number of scraps I've received so far (including spams) is less than the number of friends I've got. Some record that ! I don't know why people converse in scraps, when there is no context along with it. Off goes the concept of top-posting, bottom-postin, CC, BCC ...

While Orkut is great - I was amazed to find classmates I haven't met after my 7th standard ... nearly 15 years ago - the kind of rat-race that marks it is a bit nauseating. One guy, who had around 5000 scraps some 8 months ago, smugly mentioned that I don't have enough scraps. Of course not, you dolt !

You can also learn a lot about the latest Orkut clone, and it isn't hard. I keep getting invites from my well-meaning friends, who possibly think that I don't have enough things to do online and by others, who themselves have nothing to do online and/or just keep wanting to increase their friend count. Idle's mind is devil's workshop indeed, and the output is a new social network ! Anyway, coming back to the invites, once it is Hi5, another time it is Yahoo 360, then it is some (it was the rage some 3-4 months ago) and now someone has invited me for some Tagged and someone else for something else. As usual, I've marked the invites as spam in my Gmail.

A polite message to those friends: please don't bother; i'm not interested. Can't manage a single orkut account - no time to play with 100 odd sites.

Labels: , , , , ,

Sunday, February 04, 2007

Harry Potter and the Sophomore slump

There is a well-known phenomenon in software called the second-system effect where the sequel to an elegant, small and effective software is often an over-featured, under-performing bloat. I think the effect applies to literature as well.

When the release date of the last Harry Potter novel was announced, I thought "finally, it is going to end". Make no mistake, I enjoy reading Harry Potter. It is one of the few books that has managed to portray amazing consistency between two interacting alternate worlds (wizard and muggle). But somehow somewhere JK Rowling has lost her way. It isn't hard to make out - it is called the Sophomore Slump.

I loved the first HP novel - it was small, witty, had a great sense of humour, reading it was a pleasure not just because of the story, but also because of her play with the words. In the aura of the 1st, the quality of the subsequent novels came to be overlooked. But not for long. As the length of the novels started increasing, the plots started darkening, somewhere the humour was lost. The progressive decay became noticeable. The details of the plot became more important than the pleasure in reading it. Maybe the pressure on JK Rowling to make the facts of each book consistent with its predecessors is so much that she has forgotten the real need of the reader.

She wouldn't be alone.

I thought David Eddings' Belgariad was quite good. It had a crappy plot, but fantastic dry humour, which delighted the reader. But, with its sequel, the Malloreon, having a similar but even crappier plot, it was a yawn.
Terry Goodkind's Sword of Truth series started off pretty well, but then became such a bore. I stopped reading simply waiting for the story to end. The repetitive sub-plots in the book overwhelmed my desire to read it. But my hope hasn't been realized - the 10th book is out. If you can't complete a story even after writing 10 books in the series, what sort of an author are you ?
Another favourite of mine is David Weber's Honor Harrington series. The first two books are simply brilliant. But, as the series progressed, the theme subtly changed, away from a more human character to a more technological and artificial one. The 11th book in the main plot is simply too long, too mechanical. But it has set the direction where the agony will hopefully end soon. I understood how bad it has become after reading Jack Campbell's Dauntless - a fantastic novel.
The list is long - R.A. Salvatore's Forgotten Realms saga, that started with a pretty impressive Dark Elf series; the Star Wars prequels ... I could go on and on.

A few books buck the trend. Isaac Asimov's Foundation series being the prime example. Lord of the Rings is the other. But very few.

The exceptions can baffle the analysis. Just like a painting offers insight into the artist, maybe a novel could give insight into the mind of the author. If that is true, then probably the sophomore slump would show that the authors become victims of their success. They feel constrained to meet the perceived expectations of their fans.

The authors resort to the same formula but unable to replicate the success, and resorting the previously successful formula result in gradual increase of brain-damage of the plots. And, due to this, maybe they themselves change. They become too greedy and sign contracts to deliver books, and feel constrained by it. They cease to remain artists and become assembly-line workers. The flow of original thought, the sense of humour - the very things that make a book lively are lost. The reader is no longer what the author is looking to satisfy. Maybe he/she is now thinking of the history books or wikipedia entries ?
Perhaps there is a "literary entropy" that progressively increases in the novels of a series ... Food for thought.

Oh ... I'm now waiting for July 21st ...

Labels: , , , , , , , , , , ,

From Linux to OS X

I just read the announcement announcing free device driver development by kernel developers. My only surprise has been why it took so long for anybody to get tackle the root of the issue and articulate the solution so well. It also took me on a trip of nostalgia.

I've been involved with Linux and open source since 1999 (about the time I started my engg), when my friend first installed Redhat 5.2 "Apollo" on my new computer from a PCQuest CD. I then discovered that the X Windows didn't work because I had a notorious video card - SiS 6326. At that time, I found out that the college server ( domain) was administered by students, and hence I met Kalyan Varma and Sushanth and they introduced me to Linux India and BLUG. I found out over LI lists that I wasn't the only one having an SiS hangover, and there was a lot of traffic created by 2 SiS cards - 6326 and 6215c. Redhat 6.1 was released along with PCQ November 1999 issue and one of the tricks that somebody had discovered for getting X on the SiS card was to specify the card as "Diamond Speedstar A50" (a google search tells me that the advice is still online ), and it worked ! I quickly became the "SiS expert" - getting X configured on the PESIT lab comps and friends' computers. I knew the XF86Config file inside out, but all that changed when the new X11 version (version 4 ?) was introduced.

In the meantime, I wasn't sitting idle. I became the master of the shell - I read man pages with enthusiasm, discovered arcane options to arcane commands, and kept screwing up the comp time and again. Did you know that mailing your hard-disk to yourself could (atleast on RH6.1) overwrite the partition ? Or that the fastest way of (unsafely) rebooting a comp is yes > /dev/mem and the fastest way of shutting down the comp is shutdown -nh now ? Did you know that by chattr /.fastboot, fsck doesn't run on startup .... ever ? I used Debian Potato as my OS with WindowMaker as window manager for more than a year. Go figure :)

Anyway, I also found out that my sound card wasn't working. Those were the early days of ALSA and the traditional sound system was OSS. Getting ALSA going meant recompiling the kernel and all the related headaches. I learnt more about internals of an operating system by messing around with Linux than by reading any book or attending any course. Seriously. It helped me tremendously in my job at Netscaler - I was the first to use kernel modules to develop code, I was the first to use Evolution 1.4 to connect to Exchange server, I was the first to use Linux in Asia-ops for development, I was probably the 2nd to use (X)Emacs in the entire company.... I was the first to use rsync to sync source code from my development machine to the compilation boxes... I was definitely the first to use remote X instead of VNC... I was also the first developer to install and configure BIND, DJBDNS, Squid and Tux kernel accelerated HTTP server on my test-bed. My proficiency with Apache configuration file meant that porting and maintaining of an Apache 2.0 module (mod_ipmagic) was thrust upon me !

All in all, it was incredible fun ... "fun" defined in a perverse way. The biggest difference I found with Linux from Windows was that easy things were difficult, difficult things were also difficult. All you had to do was set it up once (and that took weeks of reading man pages, mailing list archives, /usr/share/doc documentation) and it ran like clockwork. In Windows, easy things were very trivial, but difficult things were impossible.

When I was about to start my MBA, I was looking for something that would just work, for simple use as well as power use. I didn't want to compile kernel modules or edit configuration files. A lot of Linux distros are almost there - Ubuntu, Suse ... but the gulf between "almost there" to actually being there is pretty wide. Not wanting to get into a mess with Windows, I looked out and found the ideal OS for me. Mac OS X. It has its limitations, but so far, its greatest feature is that it doesn't interfere with my work. Apple has been consistent in its innovations and more importantly, getting the innovations simple enough for people to use it. I bet ya the next version of OS X will overwhelm Vista - watch out !

Labels: , , , , , , , ,