Wednesday, February 21, 2007

Peter Naur's Turing Award speech

The great Danish computer scientist, Peter Naur, received the 2005 Turning Award from ACM for his "fundamental contributions to programming language design and the definition of Algol 60, to compiler design, and to the art and practice of computer programming."
It’s very hard to overestimate Naur's contributions to the fields of computer science and programming in general. He really was the first who focused on how to define programming languages and programmes in a coherent way, thus ensuring the languages and programmes can be understood by others. His name is of course the source of the 'Naur' part of 'Backus-Naur Form', which is used to define the syntax of programs.

Yesterday Naur gave his Turing Award acceptance speech at the University of Copenhagen, so those of us who weren’t there at the Award ceremony could get a chance to hear it. Not surprising the entire faculty of Computer Science at the University showed up, and quite a few students and others as well.

The title of Naur's speech was 'Computing versus Human Thinking', and basically was about his attempts to describe how human thinking works, though he dwelt a little upon his past work over the last fifty years.

When listening to Naur, it's important to realize two things:

  • He sees science differently from most scientists. In his eyes science is about giving a "coherent description", and not about the scientific method. This has of course been fundamental to his contributions to the field of computer science (which he thinks is misnamed, and should be called 'datalogy', since he really considers it all about learning about data), but it means that he can be rather dismissive of other peoples' work.

  • He is a bit of a rogue, taking pride in going against the mainstream. He is right often enough for other people to not entirely dismiss him out of hand, but it ruffles some feathers.



I was aware of these things, yet even so, his speech did surprise me.

As I said, most of his speech was devoted to his model of human thinking (or mental life as he said). First he started with describing how his interests in such things lead him to study such fields as philosophy, psychology and linguistics, and how he found them all lacking.
Well, lacking is too mild a word – he dismisses the entire fields, for example saying that psychology is "all questions, no answers", and that there are no such things as language and knowledge.

Having found all these fields lacking, except for a few rare basic works (all published around 1900), he set out to make his own model of how peoples' mental life works. He calls this model a "synapse state model", and makes it very clear that it works entirely different from computers (thus probably making AIs impossible).
It would be impossible for me to go into any great details of Naur’s model, but apparently his speech was published in the January 2007 issue of Communications of the ACM, and can be bought here (membership necessary)

Now, Naur is clearly working outside the mainstream here, and is stepping on quite a few toes, but even when doing so, he still works within the scientific process. He is trying to get his ideas published in peer reviewed journals, and while waiting for that to happen, he continues his work, working together with other scientists (including a psychologist) to ensure that his hypothesis are as good as possible. He expects it will take him up to twenty years to get his ideas published, because that’s what it usually takes for this kind of stuff.

Neo-creationists should take notice of this. This is how science is done, and why scientists have no respect for the neo-creationists’ attempts to circumvent this.

Labels:

3 Comments:

Anonymous Anonymous said...

He was a great scientist in the sixties but he's a kook now.

February 25, 2007 3:58 PM  
Blogger Kristjan Wager said...

Well, yes. I was trying to imply it, without being disrespectful of his earlier work, but that's pretty much my impression as well.

Of course, I have some respect for the fact that he actually tries to work throught the scientific process, and get his works published.

Doesn't make his ideas more valid, but he doesn't stand and cry about having his ideas repressed.

February 26, 2007 3:30 PM  
Blogger Algosome said...

Well, it is a fact that brains are nothing like computers, at just about every descriptive level. Consider the contrasts:

asynchronous : synchronous

massively parallel : small-scale parallelism

basic timescale of nanoseconds : basic timescale of milliseconds

associatively sequenced : numerically sequenced

associative memory referencing : linear memory referencing

probabilistic : deterministic

semi-linear (analog) logic : binary logic

This is bound to have major consequences, though any competent neuropsychologist will recognize that Naur`s model of the mind is unpublishably primitive.

June 24, 2007 8:31 AM  

Post a Comment

<< Home