Saturday, December 30, 2006

Game Theory

There is an ongoing investigation into the murder of the wife of a University of Pennsylvania game theorist. The game theorist is currently considered a suspect. The damning piece of evidence in the case seems to be his profession. Not to worry -- the district attorney, Bruce Castor, is considering all of the evidence:

According to what Castor has been able to glean, Game Theory is an economic philosophy wherein a person can apply factual scenarios for a desired outcome."In a criminal context, somebody applying it would calculate all the angles and then go ahead and commit the crime," said Castor. "It could also be a coincidence, too (that Robb taught classes on Game Theory)."

So while I suppose it could be a coincidence, its awfully suspicious. I wonder how Castor was able to glean this... Very impressive.

Full Story: http://www.timesherald.com/site/news.cfm?newsid=17646875&BRD=1672&PAG=461&dept_id=33380&rfi=6

Thursday, December 28, 2006

Awards and Honors

Browsing Barnes and Noble today, I was reminded of an important award that I forgot to list on my NDSEG application! I was the Time Magazine Person of the Year (shared) for 2006!

Drat.

Tuesday, December 26, 2006

Fellowship Applications Submitted

As of yesterday, I am done with applying for things (hopefully for a little while). I sent off my application for the NDSEG fellowship, which was a request that the defense department give me about a hundred thousand dollars.

Applying for these fellowships is a funny thing, because at least in the sciences (and particularly at the CS department at Carnegie Mellon), no one needs them. We are already well paid. Receiving this fellowship will not allow me to pursue groundbreaking research for the benefit of our nation that I would not be able to otherwise. The benefit would mostly go to Carnegie Mellon and to my advisers -- they would no longer have to pay me, and as a result, could more easily support other students. Because of this, its interesting that the government has chosen to distribute money through graduate fellowships like this, rather than just giving it directly to academic departments of their choice.

The whole thing is a study in incentives. The main benefit of the fellowship goes to the school, but I apply because it will provide me with a line on my resume and a few hundred dollars extra a month. To apply, students need recommendations, and who better to write them than their advisers? But of course, advisers have a vested interest in having their students receive outside support.

We find out in April.

Saturday, November 25, 2006

Abra-Kadabra, nonlinear alakazam!

An article today in the New York Times about algorithmic trading gives a very very silly description of machine learning, and uses the word "nonlinear" as if it means magic. "Higher-level computer science" indeed.

The result was nonlinear decision making processes more akin to how a brain operates. So-called “neural networks” and “genetic algorithms” have become common in higher-level computer science. Neural networks permit computers to create new rules and automatically change underlying assumptions by experimenting with thousands of random sequences and processes. Genetic algorithms encourage software to “evolve” by letting different rules compete, and combining the most successful outcomes.
Wall Street has rushed to mimic the techniques. Because arbitrage opportunities disappear so quickly now, neural networks have emerged that can consider thousands of scenarios at once. It is unlikely, for instance, that Microsoft will begin selling ice-cream or I.B.M. will declare bankruptcy, but a nonlinear system can consider such possibilities, and thousands of others, without overtaxing computers that must be ready to react in milliseconds.



Sunday, October 29, 2006

Trader Joe's

Trader Joe's just opened in Pittsburgh! The mayor was there for the ribbon cutting ceremony on friday.

This is fabulous; Its lack of Trader Joe's was one of Pittsburgh's major failings. Now it is alleviated!

I would write a longer post, but I have portobello and ricotta ravioli to eat.

Wednesday, June 28, 2006

High School Math and Computer Science Curriculum

I went to Brookline High School, which had a very good math and science program. Still, though, (as I expect it is almost everywhere) the focus of the math was on calculations. We saw proofs in geometry, and calculus, but these are often technical, and only steps along the way to deriving some theorem that can help solve some poorly motivated practical problem (Why do I want to calculate the rate that water rises while filling a solid created by a swept polygon??). Also, ideas from computer science were completely missing.

Its not that the more beautiful ideas in mathematics are out of reach of high schoolers -- indeed, my favorite proofs are short, simple, and profound.

If I were designing a highschool senior level introduction to mathematics, I would include:
-- The infinitude of primes
-- The uncountability of reals
-- The notion of polynomial time computation, NP completeness, and the power of randomness
-- The idea of undecidability, through the halting problem

Each of these topics can be presented with elementary mathematics, and each seems to say something profound about the universe. I think that with a course like this, rather than another course focussing on calculation, mathematics and computer science would seem like sexier topics.

Maybe the course could be called "The Nature of Infinity and the Possible"

Thursday, June 15, 2006

The government is creepy

New Jersey has filed subpoenas to telephone service providers to try and discover if they violated New Jersey consumer protection acts when they disclosed thousands of phone records to the NSA. The federal government, however, has filed a lawsuit to block these subpoenas.
Check out this quote from the Times:
"Compliance with the subpoenas issued by those officers would first place the carriers in a position of having to confirm or deny the existence of information that cannot be confirmed or denied without causing exceptionally grave harm to national security," wrote Irene Dowdy, an assistant United States Attorney, in the 14-page lawsuit. "And if particular carriers are indeed supplying foreign intelligence information to the Federal Government, compliance with the subpoenas would require disclosure of the details of that activity."

I cannot imagine what sort of "grave harm" it would do to national security for the public to know exactly what data was given to the government, and under what circumstances. Given all of the publicity that has surrounded this, if I were a terrorist, I would have to assume at this point that the NSA has all of the data that is available about all of my electronic communications. What seems to be at issue now is whether the NSA did this legally. Perhaps the grave harm that will be done if the truth is revealed is that the government will be forced to abide by its own laws? If this is the case, the correct course of action should be to change the laws, not prevent state investigations into violations of those laws...

Tuesday, April 25, 2006

New CS Building

Carnegie Mellon will be joining the long list of schools to have a CS building named after Bill Gates. (I'm counting Harvard's Maxwell-Dworkin, which is named after his mom...) See the new building here. The thing won't be complete for at least another two years, so I will still have to do my time in a windowless basement office, but it sounds nice -- collaborative areas, lots of natural light, etc. And BEST of all, "The main north-south axis, however, will make it so that the building can still be easily navigated" -- this is great! It will be possible to navigate the building!

I can't help but think that this is a not-so-subtle jab at MIT's Stata center, which is -not- easy to navigate. Otherwise, this seems like a building feature that would go without saying...

Theory and Systems

A very nice post about the need to study systems, even for theorists. This is good to hear; Some of the people I know are spending all summer studying theory, beginning their research. This gives me hope that learning about engineering challenges at Google will be a fruitful experience as well.

Wednesday, April 19, 2006

This is getting serious

We graduate in less than a month. On May 16th, John McCain (hopefully) will come speak to us on class day, and I will have a diploma in hand. Or rather, the opportunity to wait in a long line to get a diploma, which will otherwise be mailed to me (Why do I need a diploma anyhow?) All well and good.

On May 18th, Columbia throws me out on the street. So I've got about a month to find an apartment in the city.

Wednesday, April 12, 2006

Its official.

Without too much extra thought, I just emailed Carnegie Mellon and let them know that I will be enrolling in the fall. I guess this is it then -- I'm going to Carnegie Mellon. This will be exciting!

But first... My summer internship at Google.

Carnegie Mellon

I'm almost there. I sent out emails today turning down Penn and Princeton. In the emails I said I was choosing Carnegie Mellon. I've filled out the acceptance form for CMU. I just have to mail it, and then turn down Stanford and Berkeley (Ouch)

Monday, April 10, 2006

Fiber and making life changing decisions with very little information.

What is the daily recommended fiber intake? Before my graph theory class, I ate a bag of potato chips and a bag of popcorn. The bag of potato chips contained 1 g of fiber, which translated into 4% of my daily value. The bag of popcorn contained 1 g of fiber, which translated into 5% of my daily value. Is this roundoff error, or does the popcorn company play fast and loose with the fiber rules?

I'm leaning very heavily towards Carnegie Mellon now. I'll almost certainly go there. I notified some of the schools yesterday that I would not be attending them (Harvard, Michigan, Washington, Brown). I still have a bit of nagging doubt though... There are still so many areas of computer science that I haven't been exposed to, and maybe I will love. CMU is the best place to do learning/mechanism design. Where is the best place to discover new interests? Today in information theory we covered network coding. Never seen it before. Looks cool. CMU has you take your courses after you pick your advisor, Berkeley does it the other way around. Why can't the faculty I'm interested in work at Berkeley? Then the choice would be clear.

In other news, there's a new young faculty member who might end up at CMU next year, who seems very interesting. He's not going there for sure, but he's definitely not going to Berkeley, so I can add 50% of his value, in expectation, to CMU's column...

Decision in 5 days.

Friday, April 07, 2006

Theory Girl

I just found the University of Washington's rendition of theory girl today (set to the tune of uptown girl). As computer science music goes, its pretty good. Of course the mathematicians may still have us beat with the love ballad "Finite Simple Group (Of Order Two)" , by the Klein Four Group (But aren't there 5 of them?) I like "theory girl" much better than the work of MC++ though -- he's got higher production values, but borders more closely on bad rap than funny CS rap.

Also good: "The Longest Path"