Mihai Pătraşcu on teaching

[gusl]

Mihai Pătraşcu, like his teacher, Erik Demaine, is a precocious theoretical Computer Scientist. I identify with his personal story, but I'm attaching this excerpt of his CV because of his philosophy on teaching.

Like many enthusiastic computer scientists, I started in AI. Much of 5th grade was spent
implementing a chess program, from minimax exploration of the game tree, to a user interface on
my 320x200-pixels CGA screen. Eventually, my program managed to beat my young sister. But
that rewarding moment came too late for my future as an AI researcher: computers had already
become a purpose in themselves, so I switched to systems. I spent the next 2 years implementing
a compiler for FORTRAN IV, and a Common Lisp system (arguably, the antiquated collection
of books that I could borrow made for an eccentric choice). By 8th grade, I was finally ready to
embark on my search for the philosopher’s stone: I switched to theory.
My experience has profoundly shaped my personal goal in teaching: I have no intention to ever
teach computer science. I want to teach the love for computer science, and let the learning happen.
Fancy statements aside, I do not believe I would have been able to appreciate lectures on,
say, alpha-beta exploration or context-free grammars, before seeing motivating applications and
developing my own ad hoc solutions for them. But once I had done some independent thinking
about the problem, it was easy and rewarding to understand the “official” solution based on general
principles. Thus, I find the most challenging aspect of teaching is to motivate the material, and
highlight the wonderful connections across the breadth of computer science. Teaching the actual
material to students who want to learn it is easy.
As an illustration of the importance I attach to motivation, I would sacrifice problem sets
to this altar. Rather than assigning problem sets on a recent lecture, I would assign problems
sets on a motivating application of a future lecture. Once they develop the courage to think
creatively, students will produce some solution, and they will be ready to understand and appreciate
a principled solution in a general framework. By contrast, problem sets consisting of applications
to recent material tend to generate a kind of “parrot understanding” in students, who know they
will forget everything soon after then course is over.

I hope he returns to academia, and gets a shot at making some changes.

Comments

[infryq]

This sounds an awful lot like the "Do-Learn" strategy we used at Olin. Theory courses were paired with project courses, with project content leading the lectures. We built a kinetic sculpture designed to juggle ping-pong balls, and did the math modeling of flight to figure out what dimensions things should be -- by the time drag lectures and dimensional analysis came around, we were definitely desperate for some actual facts. Repeat for the tesla coils we built second semester, and the vox board we built third semester, and the failure analysis of a fencing foil that snapped off when I stabbed someone, and all sorts of circuits -- it's a lot of work to teach, and sometimes stressful to learn in such a constant state of oh-god-this-is-never-gonna-work, but, damn, you remember the stuff.

So, never fear -- it's being done. I do hope it spreads, though.

[the-locster.livejournal.com]

I totally agree. I always had a project on the go throughout university and my later school years. This basically meant attaining high test scores on some specific tests related to areas I had worked on and being bored senseless (and scoring poorly) by the bulk of the work. Only in hindsight has it become clear that the problem was really with the quality of teaching - the herd them into a lecture theatre philosophy. Present topics one after the other with little or no explanation of how they relate, what problems they solve and why that is important, etc.

These last few years I feel like I'm fascinated by everything and don't have enough [spare] hours in the day to learn everything I want to and to spend time developing my ideas. With funding I'd be more than happy to quit the day job and just allow my interests and ideas to progress naturally - they do anyway it's just they get squeezed into evenings and weekends.