Gödel, Escher, Bach - an Eternal Golden Braid by Douglas R. Hofstadter

The first time I came in contact with Gödel, Escher, Bach (GEB) by Douglas Hofstadter I was seventeen or eighteen and we were supposed to do a presentation or was it a term paper in philosophy class. In any case a friend had heard about this book and we decided to read it. We made the project, but we did not make it trough the book. Although I was enough of a geek already back then, I just had not considered the questions raised by Hofstadter, and was not mature enough knowledge wise. In any case, I read some of it, and most importantly got familiar with the text.

A few years later we were presented by a couple of the dialogs in one of the Artificial Intelligence (AI) courses at Uppsala University. Thus, I saw the book in its natural element. For although it raises quite a few philosophical questions, GEB is above all an important work in the history of AI.

Now, about ten years later, I have finally gotten around to reading the whole book. I made sure it was on my Christmas list a couple of years ago, and in September this year I finally got to it in my list of must-reads.

For those that have not heard about GEB, it can may be be described as a quite important book (was awarded the Pulitzer price in 1980) where the author, Douglas Hofstadter, discuss views on the mind, what intelligence is and how it could be implemented in machines. The book is far from a techie-book though. It is more a journey in philosophy, music, literature and art, as well as mathematics, where the author tries to argue for his views on consciousness.

At the core of the book is a phenomenon that Hofstadter call strange loops. This is a form of self reference, one can say, and the book use the work by the mathematician Gödel, the composer Bach and the artist Escher to illustrate the concept. True to this concept with the three masters from different disciplines, the author weaves together quite a lot of knowledge, and the text is mostly a joy to read.

One can of course note that the knowledge has gone forward in some areas, the book was after all written almost 30 years ago. However, these things are of less importance, because the ideas that the book describe are still of high interest. What is the mind, and can machines think?

Hofstadter introduces the themes of the book slowly and tries to explain the questions before he attempts and answer. The chapters are also interleaved with dialogs inspired by Lewis Carroll, featuring some notable personalities. What I especially like about the book is how full of references and word play the dialogs actually are. It is a joy, at least for people like me who always like to see patterns and clues, to try to spot the theme of the following chapter by the structure, references and style of the dialog. I liked that.

So, what about the message of the book then? Is it possible for machines like computers to think? Is the key in the complexity of the algorithm, the self reference of the strange loop the key?

Some days I agree, some days I do not. I have thought some about it for quite some years now, but I have not read as much as I should have. I do not know. Hofstadter makes some good points, and I think I know what he tries to say. When we learned Lisp for the AI course I was referring to above, we were all awe struck by the elegance and the power of the language. I remember that I was almost hypnotized by the feeling of what could be done, and it was more than once that I ran to the computer lab to just implement a small thing that in my head was just… divine . But, well, that last step was always missing, and it was not until I realized what it was that I knew why the program would not work. It was a interesting experience. The missing part was almost always the thing that Hofstadter describes as a strange loop I think. Now, I find it noteworthy that a lambda-calculus language like Lisp can give such an enormous feeling of almost to know how to do it, to almost have it. At least for me. I loved it. In any case, the "strange loop" was missing. So, is there a natural law that make it impossible for us to implement an intelligence on a computer, or is it just that someone has to figure out how? Evn though it is impossible for a system to describe itself (of course), can unconsciousness or intelligence adhere from the complexity? As a by-product risen out of a complex feedback.

I honestly do not know, but I look forward to read some more books on this matter. In any case, GEB is a classic and I am glad to have spent the time it took reading it. While I do not know what I think on the matters presented, it is still an excellent book and a good introduction. It opens your eyes.