Memorable Defences

So it turns out I survived my PhD defense. It was held on Match 23, a month and a half ago roughly. I must say it went very smoothly, almost too smoothly perhaps. The opponent was Jakob Rehof.

I should perhaps say something about they way defenses work here in Sweden. They're always public, anyone can come and listen. There is an opponent who is an expert in the field and who asks most of the questions. Then there is a grading committee, usually consisting of three people, with varying degree of expertise in the field. My committee consisted of Fritz Henglein, Kostis Sagonas and Sibylle Schupp. They also get to ask questions, after the opponent is done. Lastly, anyone in the audience can ask questions to the defendant. After the defense the grading committee will meet and decide whether to pass or fail the candidate. They always pass. Some of you might think that I'm joking here but I'm not, in Sweden they always pass. The reason is that if the PhD candidate might not pass then the opponent should say so before the defense, not after. So once the opponent has agreed to having the defense you know you're going to pass.

Having had my own defense got me thinking about all the defenses I've attended through the years and I thought I'd give a little tour of the ones that I particularly remember.

The First Defense I attended I think was for my cousin Åke Wallin who did his PhD in the area of neuro-psychology. He studied the episodic memory of very old people, in their 80's or 90's. Studying such old people is rather unique, there are very few places in the world were there are large enough groups of old people to make experiments statistically valid. Stockholm is one of those places and is where he conducted his research.

The First Computer Science Defense I attended was that of Urban Boquist. He had written a whole program optimizing compiler for Haskell. I found his thesis a very exciting read, it's really well written and has lots of cool optimization described in a very clear and concrete way. Very appealing. A guy call John Meacham also found this appealing and has taken the ideas from Urban's thesis and put them into a publicly available Haskell compiler called JHC. John has also made sure that you can order Urban's thesis online.

The Best Defense is without a doubt that of Johan Agat. Johan worked in the area of computer security and studies timing leaks. He pioneered the technique of cross-copying to pad programs to eliminate timing leaks. The reason I found is defense so good is two-fold. First, the opponent, Jens Palsberg, gave an exemplary introductory explanation to Johan's work. Everyone understood it. It was as clear an explanation I have ever seen. Normally on these occasions it can be hard, even for other computer scientists who are not experts to follow these explanations. Second, Johan did an amazing job at answering all the questions he had. Even hard, deep question he had very good and well thought through answer to. Very impressive. For these two reasons I find this defense the best I've ever attended.

The Most Pathetic Defense was that of Koen Claessen. If you know about Koen you know that he's an extremely good researcher and his work during his PhD was no different. In fact, it was exemplary. Which is why the opponent and the grading committee really didn't have anything to ask about. It was all just praises from their side, no criticism or hard question. So it wasn't much of a defense, more like an appraisal. Hmmmm.

The Longest Defense goes to Karol Ostrovsky. The poor guy. It seems his opponent and grading committee wasn't properly informed of the standard procedures here in Sweden. Normally a defense take around two hours. Karol's must gone on for like four hours or so. And since it started at 10 am and going well past lunch there weren't many people left in the auditorium the last hours. I didn't stay either, I simply had to get some lunch. But eventually Karol also became a doctor.

The Most Nerve Wracking Defense was obvious that of my own. But as I said, it all went very well in the end.