Holiday Party 2006

A couple weeks ago, the computer science department at Cornell held its annual holiday party. I've since managed to collect some pictures, and figured I'd share a few memories of that splendid event. I apologize for the lack of color consistency between pictures. They were snagged from a variety of cameras (and photographers), and my Photoshop skills are not advanced enough to homogenize the color balance.


We begin with a shot of Professor Graeme Bailey showcasing is musical talents during the reception which preceded dinner.


Ymir decided to join in at the piano. Dan looks very impressed.


The reception quickly becomes crowded as more faculty, students and staff arrive to join the festivities.


A pair of Turing Award winners.


Pondering my own infinitesimal chances of winning a Turing Award. My housemate Yunpeng (in background) could not care less.


The migration from the lounge to the ballroom commences as dinner-time nears.


What's the best kind of grad student? Why a hungry one, of course!


Alex and I waiting in line for food. It seems the grad students were particularly ravished this year. As a result, the line was very long.


The submissions to the gingerbread house contest.


Department Chair Eva Tardos presenting the awards to the gingerbread house contest.


Every year, the 2nd year PhD students performs a skit at the holiday party. This year, we decided to do a parody of the orientation presentation given to the first year PhD students when they first arrive. As the acting Director of Graduate Studies, I was the only one who "performed" anything.


Luckily, my hours of rehearsing paid off and I delivered a well-received performance. You can view the slides we made for the skit here.


The actual Director of Graduate Studies, Dexter Kozen, splits his time between being a professor and a rock star.


He really got into it.


Relaxing with my other housemate Yunsong.

This year's holiday party was a lot more stressful than last year's, since I spent many hours prior to the event rehearsing the skit. To give you an idea of my state of being, I barely indulged in the free wine, taking a few sips at most. Anyways, next year's should be much more relaxing.

The Art of Academia

In my recent conversations with friends fortunate enough to have evaded the long journey otherwise known as grad school, I noticed a few recurring misconceptions about my chosen profession. Being the kind soul that I am, I've decided to take some time out of my extremely busy day to clear up some misunderstandings regarding academia and discuss my reasons for choosing it. Bear in mind that this is my own personal (and probably heavily biased) view on the trade-offs between different types of jobs in computer science.

While most university positions include some type of teaching responsibility, there are many who don't. In fact, a large percentage of professors actually dislike teaching. So why, you ask, would these people become professors in the first place? The truth is that at research institutions, teaching usually comprises only a small portion of the expected workload.

So what kind of activities take up a professor's time? For the kind of job I'm interested in, it's a plethora of things, which is part of why I find it so appealing. But a large part of it is research. From a high level, one can think of a professor as managing a small business. The employees are students (mostly grad students, but some undergrad as well), and the products are research publications.

How is this different from research jobs in other settings? In the field of computer science, industrial research can range from having a strong applications oriented focus to more theoretical work. By comparison, research at universities typically fall more towards the theoretical end and is usually at least one level removed from mainstream applicability.

So why is such research appealing? Well, one of the biggest things which turned me away from development jobs at places like Microsoft is the fact that any complex software intended for mass distribution needs a great deal of attention to detail. Given that this is not something I particularly enjoy, it would follow that I'd probably give a subpar effort in this regard. As such, I enjoy working on projects whose end goals do not include a finished product for widespread public use.

Another annoying aspect of standard development jobs (as well as application focused research) is this tendency to hack at things until they work. This can come in many flavors and guises, but it all boils down to the fact that deadlines must be met, and the elegance of a solution will typically only gain you a little self-satisfaction.

I enjoy taking a more principled approach in tackling these problems. Rather than trying to work within the current known limitations, I prefer to find a reformulation of the problem which yields a simpler solution. This point is perhaps best illustrated with a short example. Suppose you're given a bunch of data points and you're trying to spot a trend. You've tried least-squares line-fitting as well as a few other methods. Nothing seems to work great, but certain fitting methods let you tune parameters until the line seems to fit OK. By doing this, you are implicitly making an assumption that the data you're trying to fit was sampled from a distribution which is geometrically similar to one of the models you're using to generate the fit. Since nothing seems to work really well, you end up tuning a bunch of parameters until something looks decent. By contrast, if you tried a broader range of models, you might find that a log-linear rescaling of the data yields a near-perfect straight line, which is a much simpler solution than the previous ones you hacked up. The trick then is to find the correct mathematical tool in order to abstract the problem into something that is simpler and therefore more elegant. As you might expect, this is usually pretty difficult for real-world problems.

But progress has been made in the past, and progress will be made in the future. I find this type of research very exciting because the results can open up entire new ways to view a certain problem, which might suddenly turn what used to be very difficult into something much more tractable. The elegance of these results holds a certain aesthetic appeal to me, which I find very compelling and inspiring.

So what's my dream job? In its current incarnation, I would ideally, at some point in the foreseeable future, be locked in as a tenured professor at a major research university. If I'm really lucky, I might even get invited on a few federal committees, though that typically doesn't happen until the latter stages of one's career. I'd have my share of teaching duties (which I enjoy immensely), as well as a horde of grad students at my beck and call. My daily routine will be interjected by periodic travels to various research conferences. And every few years I'll be able to take an extended leave (known as a sabbatical) to focus purely on research.

While the picture I've painted thus far might seem laden with roses, there are plenty of thorns in sight as well. First off, the pay, though solid, will never be spectacular. Furthermore, the process to even matriculate from grad school requires many long, lonely nights at the office. After receiving a PhD, it's a dogfight to gain a professorship at a good school. Should you survive the interview gauntlet and be awarded a position, you'll spend the next six years or so working your ass off to gain tenure. Then, and only then, can you breathe a small sigh of relief and consider the storm mostly weathered.

So is worth it? The profession is obviously not for everyone. It may be that I'll stray from this path ere I reach the end. But in terms of finding a job which affords incredible intellectual freedom, can offer up potentially groundbreaking advances in science, as well as steers relatively clear from frustrating corporate mandates, a professorship is, in my opinion, without peer. Given that such positions are extremely competitive despite all the negatives I've mentioned, it seems there are many others who would agree with me.

The Al Gore Review

It seems like everyone has an opinion on global warming these days. Of course, we have An Inconvenient Truth to thank for that. I think Al Gore did a commendable job publicizing global warming. However, I feel that the documentary raises more serious and complex questions than the simple ones it answered, and I believe this is ultimately lost on the general public.

Gore is right about one important thing. Unchecked global warming will have unforeseen consequences, though we are slowly discovering some of them.

I also believe Gore is right about the lack of political will in the United States to be more actively conscious of the environment. After all, market potential dictates any business decision, and it's up to the government to add additional penalties or rewards to prevent against unchecked exploitation. In some cases (such as this one), when public interest is lacking, politicians also have little motivation to care.

But it seems to me that more questions are opened up by Gore's documentary than answered. After all, while it's great for the public to be more aware of the problem, are we sure they understand what the problems are? I'm certainly not sure where I stand on the issue, and I've spent quite a while reading up on the topic from a number of sources.

Gore's primary goal was to convince the general public to be more concerned with the reduction of greenhouse gas emissions. I still don't buy the evidence he presented. But even if I did, there's no reason to assume that global warming would stop once we cut down our own greenhouse gas emissions. And thus, the consequences we risk facing might still come to pass. For example, Gore states that the melting icebergs will cause sea levels to rise dramatically, thus destroying many populated coastal regions. But we know that sea levels have, in the past, been at levels even higher than it is now.

So this begs the question of whether or not Gore missed the main points of this issue completely. The answer to that question will probably not be known without further work by scientists. Maybe five years down the road, Gore will make another movie themed, "The world is warming up whether we like or not. Let's find efficient means to deal with it."

My last criticism deals with the mini-biography spliced into the documentary. It's possible that a straight-up documentary would be too boring for the general public, but I feel that it disrupts the purity of his work. And if he does end up going down in history as a pivotal figure, it would be a shame that his work was scarred by such petty trivialities. But such is the price for progress.

Who knows, maybe Bill Gates is the answer.