Ullman and Patterson Write on Research, Computer Science and How to Advise Students

29 Jun

When I was finishing my Ph.D., I read a book based on interviews of people talking about their jobs to help decide what I would do next. What I learned from the book was that people were happy with their careers if they designed or built objects that lasted, such as the Empire State Building or the Golden Gate Bridge, or if they shaped people’s lives, such as patients or parishioners. Thus, I went into the job of assistant professor with the hypothesis that my long-lasting impact was not the papers but the people.

I was a student, and later faculty member, in an electrical engineering department, where the widely held opinion was that the way you wrote a thesis was to read many papers. Look at the last section, where there were always some “open problems.” Pick one, and work on it, until you are able to make a little progress. Then write a paper of your own about your progress, and don’t forget to include an “open problems” section, where you put in everything you were unable to do.

Unfortunately this approach, still widely practiced today, encourages mediocrity. It gives the illusion that research is about making small increments to someone else’s work. But worse, it almost guarantees that after a while, the work is driven by what can be solved, rather than what needs to be solved. People write papers, and the papers get accepted because they are reviewed by the people who wrote the papers being improved incrementally, but the influence beyond the world of paper-writing is minimal.

Jeffrey D. Ullman (who taught us compilers) and David A. Patterson (who taught us about computer architecture) have recently written two viewpoints which were published in Communications of the ACM. The title of Ullman’s paper is ‘Advising students for success‘ and Patterson’s is ‘Your students are your legacy

In these short articles those valuable computer scientists provide some very useful insights about what it means to do research, what can go wrong and the measures that should be taken in order to nurture a productive environment to boost new ideas and implementations, such as:

“* Show initiative, for fortune favors the bold. Don’t wait for professors to tell you what to do; if we were good managers, we probably wouldn’t be faculty. Explore, challenge assumptions, and don’t let lots of prior art discourage you.
* Sink or swim. We’ll offer you what we think are great projects with plenty of potential, and we’ll support you the best we can, but it’s what you do with the opportunity that makes or breaks your graduate student career.
* Educate your professor. We’re in a fast-moving field, so for us to give you good advice we need to know what you’re working on. Teach us!”

“Open collaborative laboratory. We were increasingly seeing people optimize their schedules to avoid disruptions by working from home when they didn’t have classes or meetings, since computers and networks were just as fast at home as in the office. The negative global impact of such a local optimization can be thought of as corollary of Metcalf’s Law: if the value of a network is proportional to the square of the number of connected users, even a small group leaving a network can significantly decrease its value. This drop in value can in turn cause others to leave, with the negative feedback loop continuing until the network nearly collapses.

In 2006, we experimented by creating a physical office area with contiguous open space for everyone in the project, including the faculty. We hoped that easy access to faculty would draw students to campus and that the open space would inspire innovation by increasing the chances of spontaneous discussions.

The open space makes it very convenient to quickly grab a group of interested people on a moment’s notice for a discussion rather than trying to wander around the building or exchange a volley of email messages to schedule a meeting. We have also been surprised to see new students in this space quickly act like senior graduate students. Apparently, easy access to faculty plus watching how senior graduate students operate helps new students move up the learning curve quickly.

The research retreats and open space also build esprit de corps, as we play together one afternoon at retreats?for example, skiing, paint ball, and river rafting?and in the lab we collectively watch presidential debates, movies, and big sports events.

The challenge of our open space is then to preserve concentration while enhancing communication, for otherwise people will still stay home. Distractions are reduced with large displays, headphones, and relying on cellphones instead of landline phones; the custom is to make and take calls outside the open space. We also included many small meeting rooms in which to hold vigorous conversations. The result is an open space about as quiet as a library or coffee shop, which is good enough for most to concentrate while encouraging spontaneous communication.”

“When I was finishing my Ph.D., I read a book based on interviews of people talking about their jobs to help decide what I would do next. What I learned from the book was that people were happy with their careers if they designed or built objects that lasted, such as the Empire State Building or the Golden Gate Bridge, or if they shaped people’s lives, such as patients or parishioners. Thus, I went into the job of assistant professor with the hypothesis that my long-lasting impact was not the papers but the people.

Thirty-two years later, I can confirm that hypothesis: your main academic legacy is the dozens of students you mentor, not the hundreds of papers you publish. My advice to advisors is to get your students off to a good start, create stimulating research environments, help them acquire research taste, be a good role model, bolster student confidence, teach them to speak well publicly, and help them up if they stumble, for students are the real coins of the academic realm.”

A Newer Model: Project-Oriented Theses

It took many years to reach this point, but it is now fairly routine to have substantial software projects carried out in an academic setting. While there will always be the occasional thesis that is purely “pencil-and-paper,” a much more productive approach is to introduce beginning Ph.D. students to a project. Often they enjoy “learning by doing,” contributing to the software development, while learning the new notions that are being investigated by the project. Senior students often get the opportunity to help, and even to supervise, junior students.

The best example I’ve seen of how to use this mode effectively comes from my colleague Jennifer Widom. In a series of innovative projects (semistructured data, stream databases, and now uncertain databases), she has perfected a routine, consisting of:

1. Define a general goal for the research, and get a team of doctoral students working together.
2. Spend a substantial period of time, perhaps 6?12 months, in which the theory and models underlying the problem area are developed. (Jennifer says that this step?making the students part of the planning and modeling?is what distinguishes her approach.)
3. Then, start an implementation project. Get the students working on pieces. The goal of each project is a robust, distributable prototype, not something that can be carried intact to commercialization.
4. Allow students to identify their own aspect of the broader problem area on whose difficulties they will focus. Students develop their own ideas, which form the core of their thesis, and are able to validate the ideas by installing them in the larger system.

It is sad that many research-funding agencies, such as DARPA, have become so “mission-oriented” recently. While it may be possible to support a Ph.D. student doing part of a project implementation, Step 4 is left out; there is no room on the project for a student to explore original work outside the boundaries of the project. For example, I have heard from several independent sources that while the European Union has been supporting “research” generously, the support is sufficiently constrained by concrete deliverables that there is no way to support Step 4 on the projects. In countries where Ph.D. support comes from a state source, this arrangement presents no serious impediment. However, in countries where Ph.D. students are dependent on project support, it becomes hard to train first-rate researchers.”

Students and Startups

One of the trickiest decisions an advisor has to make is how to deal with the student who wants to found a startup while they are working on their doctorate. Few people agree with me on this point, but I believe that, unless the startup idea is insane, they should go out and do the startup. My theory is that, while getting a doctorate and entering the research arena is a high calling, it is not the highest possible calling. A startup can have more impact on our lives than a thesis. Moreover, if they miss the opportunity to do a successful startup, then they have lost a great deal. If the startup flops, as many do, they have lost only a few years, and can resume work on a doctorate if they wish.

Sergey Brin never asked me whether or not he should quit the Ph.D. program and found Google, but I would have told him to do so had he asked. Another student, Anand Rajaraman, did ask my advice on this matter when he was about half a year from finishing. I told him to leave and be a founder of Junglee. The venture was quite successful. A few years later he returned to Stanford, started an entirely new thesis topic that abstracted some of what he had learned at Junglee, and is now Dr. Rajaraman.

You don’t have to be in Silicon Valley to think about startups. Great ideas can develop anywhere, and a responsible advisor will, when appropriate, present to their students the option that their work might form the basis of a commercial venture. I recall an email message from a student at another school asking the question: “can a piece of work be both a thesis and useful?” When I replied in the affirmative, I was then asked to explain this point to their advisor. That advisor was serving the student poorly, although their attitude seems fairly common. Even in the course of reviewing this Viewpoint, I encountered the view that a piece of technical work is more to be admired if it cannot be commercialized.”

Leave a comment

Posted by on June 29, 2009 in General, Programlama, Science


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: