Thursday, October 07, 2010

A Terence Tao Talk -- check

Last night I got a chance to listen to Terence Tao speak. And thankfully it was non-technical so the mathematically-challenged like me could understand. It was on the "Cosmic Distance Ladder" which at first glance seems like a highly technical term/object in physics/cosmology. But it turns out to be a fascinating story of how human beings over the last 3000 years or so have measured astronomical distances using clever indirect measures where direct measurement was technologically impossible. Here's a thought experiment: Lets say you knew that a lunar eclipse spanned 3 hours. And you knew that the Moon takes a month to go around the earth. Can you approximately compute the distance from the Earth to the Moon in terms of Earth radii using no other piece of information?

The most clever trick he talked about was how Kepler estimated the orbits of the Earth and Mars (assuming them to be non-circular) using the multi-year data on Mars' declination collected by Tycho Brahe. This led Kepler to formulate his laws of planetary motion which led Newton to formulate his theory of gravitation. A great story of giants standing on the shoulders of giants...

Here's the slides in case someone's interested.

It was surprising to see how comfortable Tao was discussing a subject thats not his main area of expertise. He even fielded a few questions from the audience in the end.

All in all, a great treat!

Sunday, June 07, 2009

Rajeev Motwani passed away

Just saw this shocking piece of news:

I took a class on Randomized Algorithms with him last fall, my first quarter at Stanford. I also used to see him almost weekly at RAIN meetings.

The shock of this tragic news is compounded by my proximity to him. This is a huge loss to the Stanford community and larger Silicon Valley community and most certainly to the Computer Science community...

Saturday, April 25, 2009

The timeless magic of Pink Floyd

Listened to The Wall after a very long time, and with the whole shebang-- Floyd's lyrics on the laptop, a dark room, a dark wine to nurse, and of course, the solitude.

It dialed back the years to a time when I used to listen to it all the time. Reminded me why NYC 2003 is one of my most cherished trips ever (thanks Ashish -- aka srrdn, I love you!). Reminded me why Floyd is magical, entrancing. Reminded me why 'Comfortably Numb' is a masterpiece. Also reminded me that I have gotten old...

PS: To the uninitiated who need help understanding why Floyd means so much to so many, here are a few reviews of Floyd albums from Amazon:
Dark Side of the Moon:
The Wall:

Thursday, February 05, 2009

Life at Stanford

A lot of people in and outside of Stanford have asked me how I feel about being here. The question carries a lot more nuance and meaning when asked by people who know that I left a fat paycheck at Amazon to come here.

So here's my one-line answer to the question: Stanford is my Disneyland. It is everything I thought a grad school experience should be like. A pseudo-math way of thinking about it is: Stanford = Amazon + research - money. It has some very smart, high intensity people working on interesting, high impact problems. You don't get paid nearly as much, but that is a very small price to pay for the experience.

Monday, December 15, 2008

Redemption at Chepauk

Two close matches against two different teams at the same venue almost 10 years apart. Then he imperiously ruled the cricketing world, today he is an old lion who has to show the hyenas that he can still hunt. Then the god had failed his followers, today he gave them another occasion to believe, another sweet memory to cherish. Today he redeemed himself.

SRT wins close matches for India. Enough said.

Wednesday, December 10, 2008

Response to Terrorism: India v/s the US

Much has been said about the recent terrorist attack in Mumbai in the mainstream media, in the blogosphere and at water-cooler conversations everywhere.

The one thing that I haven't seen much of is to compare and contrast the US/Israeli approach to the war on terror against the Indian approach. I don't mean the logistics, the intelligence gathering, the geopolitical (though all of those are areas where Indian govt can learn a thing or two from the US); I am talking about the mindset of the people and of the government. And to me, that mindset is captured in the simple statement that most major US politicians have made in speeches, interviews, etc. since 9/11, which in essence says: "We will capture or kill Osama Bin Laden." Contrast that will Indian politicians who only say something to the effect: "We will bring the terrorists to justice". Note the omission of the word "kill" or its synonyms.

Why is that?

My claim: Indians are a meek, non-confrontational, war-averse people. It is a very deep characteristic of our individual and collective psyche; something taught to us by parents and held up as a virtue in society. I don't recall a single person (family, friends, teachers, etc), course or teaching (modulo isolated stories of Shivaji, Rana Pratap, etc.) that forcefully and unequivocally said: Fight for whats right. Don't just fight and die for it, but fight and kill for it.

And the most amazing, the most baffling, the most tragic thing about this thought is that fighting for whats right should come to us naturally, it should be second nature to us; afterall it is in the spiritual guidebook of the land. Bhagvadgita is full of exhortations from Srikrishna to Arjun (e.g. Chapter 2, verse 31, Chapter 3, verse 16, etc.) on why fighting and *killing* for whats right is the only honorable thing to do.

What a tragedy, what a shame...

Tuesday, December 02, 2008

Theory v/s Applications, Platform v/s Apps

The topic of this post can be summarized in one line as: "One man's theory is another man's application", or alternatively, "One man's platform is another man's app".

Let me elaborate, first using CS research and then using software as examples.

In CS research (and may be in other fields too; I just happen to know a bit about CS), there's a spectrum of research from the purely theoretical (that almost belongs in a Math dept) to the ridiculously applied (which basically belongs in a software company). Whats funny though is how folks working at any point in that spectrum view the fields "upstream" and those "downstream". I'll illustrate with a section of the spectrum that I am most familiar with (without saying anything about the extremities of the spectrum). Mathematicians who do pure Math take great pride in it and look down upon applied research as something best left for lesser mortals. However, logicians (e.g. Godel, Boole, etc.) would argue that each field of Math is nothing but the application of logic to a set of axioms (e.g. the fundamental axioms of algebra, or of Euclidean geometry, etc.). Similarly, CS theory folks take great pride in working on very abstract models (e.g. modelling a communication network as a graph, modeling text documents as points in n-dimensional space, etc.) and not worrying about "implementation details" that they leave for systems folks to deal with. However, Mathematicians think of CS theory as being utterly applied stuff since they are only concerned with existence problems (e.g. "is there is prime number between any n and 2n?") whereas theorists ask "implementation" questions like "can we find a prime between n and 2n efficiently?". Similarly, database researchers work on "database theory" problems like indexing, privacy, etc. but don't worry about implementing any of their results in a real database systems (they leave those "implementation details" to software engineers aka code-monkeys). And on it goes...

A very analogous situation exists in the software world where one man's platform is another man's app. For example, at Amazon there is a Website Platform group that is tasked with running the fleet that hosts the retail website. All the website features work as apps on top of the platform which provides basic services (monitoring tools, deployment systems, etc.) to all app owners. However, the website platform (called Gurupa) is itself an app running on the Linux box. And the deployment tools etc. that they build are built on top of compilers, linkers, and other linux apps. And those apps are themselves running on top of the linux OS. And on it goes...

In each case, there is a healthy tension between folks working on systems upstream v/s downstream. The platform folks think that engineers who write apps don't care as much about performance as they should, their badly-engineered systems break the platform, etc., whereas the app-writers think that they're working on the coolest problems and running the platform is just leg-work, neither realizing that the platform folks are really the app-folks for someone further upstream and the app-folks are writing a platform for someone downstream.

It is fun to conjecture what lies at the extremities of this spectrum. May be it is turtles all the way down...