Thursday, December 14, 2006

Advice for PhD. Students by Tony Hoare - Marktdoberdorf Summer School on Security


Me he tomado el atrevimiento de copiar un post de www.kierberg.com una excelente conferencia de Sir Tony Hoare (padre de CSP y Hoare Logics) con consejos para los estudiantes de doctorado de la escuela de verano en Marktdoberdorf. Particularmente, me parecen muy imporantes sus comentarios sobre la creatividad y sobre la capacidad de reconocer ( y usar si es posible) el trabajo de otras corrientes rivales, de manera integradora, claro está.



Transcript
… so perhaps I ought to live up to my reputation and pass on some advice about your research, right now and perhaps in your subsequent career. I learned a lot from a distinguished scientist of the past, and particular I would recommend Peter Medovar, a British biologist who was quite famous and he wrote two lovely books, one of them is called “Letters to a young scientist”. And the other one is more auto-biographical describing his own life and he called himself “a thinking radish” — a vegetable of some kind I never liked very much.

(…) He gave advice about your relationships with other scientists, your colleagues and your rivals, and how you should publish (…) your own work and compare it with that of others. And I pass this advice on to many of my own students and is quite surprisingly seldom taken in the world it is today. He said, when you described your own work, you should always emphasise its limitations. Say what is does not do. This is a very good precaution, because if you don’t find out the limitations you can be pretty sure that somebody else will. (…)

Always emphasise the merits of the work of your predecessors and your rivals, because this can always reflect further credit on the small extensions and improvements that you may have made to their work. So never claim to have remedied some defects or limitations of this work. Point out how good somebody’s work is, and say “I have just made a small improvement of this particular aspect of it.” Then, he will be on your side. Despite the fact: if you say the other way, he or she will be not be on your side.

The reason why this is good advice is that it furthers the progress of science by building up an unbiased and objective understanding of the theories as a whole and how its very parts fit together. You don’t get individual parts of the theory constantly claiming that other parts of the theory of your merit.

And objectivity is the most important attribute of science. This piece of advice: don’t claim that your scientific results are useful. Don’t claim that they are more useful than somebody else’s scientific results. You are not the right person to judge that! That other people who use it decide whether it’s useful or not. Leave that question open!

This is a piece of advice that I got from Richard Doll, another famous British medical scientist, the man who discovered the link between smoking and cancer. Since that time, it is (…) cancer-producing effects of radiation which have recently been revised: the minimum dose has been decreased as a result of re-analysis of the results of the Hiroshima bomb. He gave a lecture about this at Oxford which I attended. At the end of the lecture, a member of the audience asked him: “What advice would you give to the government — or anybody — as a result of your research that you are now doing?” And he said: “I refuse to answer that question.” And then he explained: “I don’t give advice about my current research. If I give advice about a current research I will loose by scientific objectivity, because my research will suddenly change its focus, and its focus will now be directed towards making that advice correct.” And that is a terrible terribly unscientific thing to do. You must have no focus, no personal focus in your scientific research. Advance the science, do not advance your own career! (…) that the world will realize that that is the best way of advancing your career.

He said: “I’m willing to give advice now about research I did 20 years ago. Twenty years ago, I discovered the link between smoking and cancer. But while I was working on that research I never advised anybody to take any action on the basis of that discovery. I’m willing to give advice now, because the research is 20 years old, and I’m no longer working on this area, I don’t need my scientific objectivity. I’m willing to give you advice about the usefulness of other people’s research, but not about my own.”

So don’t make claims about the usefulness of your research! Do not criticize other people’s research are not being useful enough! What you can claim is that you are working in an area which is useful. You can say that you are working in computer science, and computers are of great commercial importance to the world, and therefore I’m asking you to pay me to do this.

This is advice for your Ph.D., of course. It’s not advice if go to work in industry or for a client. When you finish your Ph.D. you must forget this advice, because you are no longer working to fall to advance the course of science — you’re working to advance the economic profit of your company. You’re work all (…) You’re working to serve a client, and you must do everything you can to do the best for that client, and usually the advancement of knowledge is not one of those things.

I could speak for a long time about the difference between working for science and working for a client. Let me just concentrate on one thing: when you’re working as a scientist it is compulsory to be original. You must do something that has never been done before. (…) If you can’t do that you can’t do science. If you’re working for a client on a project to deliver a product, originality is your greatest crime. Anything that can be done with known techniques — published techniques — you must do with published techniques. Originality is the last resort of an engineer, because that is what makes everything so risky.

And there are many other differences between scientists and engineers. But remember: while you’re a scientist don’t claim to be useful! When you stop being a scientist be useful! And it’s much easier to be useful for a particular client. Once you know your client, and you know the date on which you have to deliver your product then you can do anything to get that product out on time on cost to that client to satisfy his needs.

Well I said you should respect the work of your rivals, your contemporaries and above of all your predecessors. But the (…) 1000 persons you should not respect and that is the old fogies who started the whole thing of. So you may be as disrespectful as you like to me and to any of the lecturers in this school, and we shouldn’t complain.

(…) The reason you should disrespect us the most for is that we are the ones who created the rivalries and the wars between various research schools in our subject. In many cases the research schools are named after us: Hoare logic, Petri nets (…) are named after some theoretical systems: CCS, CSP. All these things separate our subject into research schools, and people define their research areas by initials of the school to which they own their allegiance. This determines what conferences they attend, what journals they submit to, what papers they read, and it provides them with a cosy home in which they know how to survive, they now how to write the next paper in a series, they know how to judge the papers in a series, they can referee other papers all about the same subject, and it is all very comfortable. And, of course, most people like to be comfortable. We all like to be comfortable. We all like to work with friends and know what our colleagues are thinking.

But just sometimes at least we should not respect the boundaries established between our disciplines. There’s a lovely book by Andreski called “Social Sciences as Sorcery” which was published by (…) a long time ago and he has some lovely phrases about the (…) of truth rarely lie along the battle lines of entrenched schools of thought. Of course, to begin with, if you invent a lovely new theory and notation like Petri nets, you are very keen to explore the boundaries of this theory. You know it is useful for some things, and you like to be useful for everything. And so your goal is to make this theory a universal theory that is going to apply well — you hope conveniently — to everything.

All scientists are looking for universal theories, so that is an excellent scientific motivation. But there are other schools who believe in some other notation and theory are going to explain everything. And then there are more (…) more or less now, they all existed for a long time. But in the end, it doesn’t really pay to go on fighting. As you try to extend the borders of you own territory of it further, you want to explore go into more mountainous regions, you have to cross raging rivers. And the whole thing becomes more and more cumbersome, and then you have to fight your battles on the border at the tops of the mountains with your neighbours and of course that’s very uncomfortable and often people get hurt.

After a while history tells us neighbouring countries do decide on — if possible –natural borders between their territories. So it is possible to understand in what areas each theory is capable of giving its best and indeed construct roads and links and bridges over the rivers and through the paths of between the theories.

And I think computer science is right for that kind of (…) and coexistence between rival schools. So when you see the opportunity to understand and you somebody else’s theory from some neighbouring school of thought, please do so! And if you see even more beneficially some opportunity of building a bridge not (…) constructing a syntactic translation between theories that usually doesn’t convey any really deep insight. But if you see a way of merging the semantics of the theories, perhaps (…) them in some more general and therefore less useful theory, then take that opportunity, too!

This is my other Grand Challegne which I’ve been putting forward to some time. It’s a challenge for unifying theories. Not a grand unifying theory — that is perhaps to difficult for us. But hopefully unified theories where two theories can be brought together to be applied together on a certain range of problems.

And that’s the end of my general advice about science and I wish to give my last concellation about needing this summer school. One of the consolations of life and death is that you will leave descendants: you will leave your children and grandchildren carrying the good work after you. So we urge you — those of you who stay in academic life, particularly — should take care to send you own students to Marktoberdorf summer schools of the future.

1 comment:

ix said...

Very nice talk, Hugo! It was good to see/read it :)