This is a series on the book Gödel, Escher, Bach: An eternal golden braid by Douglas Hofstadter.
Earlier diaries are here
Today we will discuss Levels of description and computer systems pages 285-309.
If there are computer science experts out there, especially artificial intelligence experts, you can help. Musicians can also help with some of this.
From the overview
Various levels of seeing pictures, chessboards and computer systems are discussed. The last of these is then examined in detail. This involves describing machine languages, assembly languages, operating systems and so forth. Then the discussion turns to composite systems of other types, such as sports teams, nuclei, atoms, the weather and so forth. The question arises as to how many intermediate levels exist - or, indeed, whether any exist.
On p. 285, Hofstadter discusses seeing Shirley MacLaine on TV and ignoring the fact that the image is really pixels. But if Shirley is acting that's yet another level. Are pixels laughing? Is Shirley laughing? Or is the character she is playing laughing? Maybe the author of the script is laughing? That last sounds silly, but if we quote from a play, we credit the author, not the actors.
Later on 285-286, he notes that the best computer chess player isn't that good. Times have changed! The best computer programs are as good or better than any human - and they don't operate purely on look-ahead.
Later on 286, he notes that good human players don't look at every move - two things come to mind. One is a quote from Jose Capablanca - a world chess champion and perhaps the best player ever. He was asked how many moves he looked at before deciding, and he said "only one move - the right move". But it can't be as simple as Hofstadter makes out, because sometimes the best move violates the principles. No one who has played for a while looks at illegal moves .... but the best players sometimes find moves that seem totally wrong, but are brilliant.
There follows a long section on computers - can any computer expert out there comment on how this has changed?
On p 290, Hofstadter describes the excitement felt by the first computer programmers, who had to hard wire each problem. This is brilliantly described in the novel Cryptonomicon which I recommend.
On p. 294, he describes briefly how kids learn language. When my nephew was learning to speak, he would wander around, pointing at things, and saying "Ot's dis, mommy?". For a long time, he used only 6 words: Mommy, Daddy, up, food, and "ot's dis". Then he went to full sentences.
On p. 295, he talks about error messages. These are still very confusing!
On p. 297, Hofstadter mentions how, in human speech, our message usually gets through, even with improper syntax; this doesn't often happen with computers! But another thing that happens with humans and computers is that the message gets part-way through. That is, the person you are talking to seems to understand, but doesn't. Or, the computer program runs with no errors, but does the wrong thing.
I'm curious as to the current state of AI research, discussed in lots of places in this chapter, e.g. pages 298-299.
And I'd like a musician's input on the ideas expressed on the bottom of p 299 - that different scales have different emotional content, and that enharmonic keys can feel differently. What about transposition? If you transpose a piece, or a song, into a different key, do the emotions change?
On p. 301 ... I wonder what Hofstadter thinks of modern word processing and publishing? Especially a tool like LaTeX?
On p. 302 he talks about the flexibility of our minds. But can we make ourselves think differently? If so, how come we aren't all really smart?