on 17 May 2013
This book has formidable shortcomings.
At the beginning of Chapter 4 Hofstadter takes the flush toilet as `probably the most familiar and the easiest to understand' example of `mechanical systems with feedback in them'. He describes how the interaction of its main components - tank, water, float, rod and valve - maintains a constant level of water in the tank. Then he goes on to talk about a `little rubber gizmo' which can cause the feedback system to malfunction. He doesn't explain what this `little rubber gizmo' is and how it interacts with the main components already described. So this is an explanation suitable for the reader who needs to know (or at least to be reminded of) the interaction of the main components of the flush toilet but who already knows about - and so needs no explanation of - the `little rubber gizmo'. Surely the author can't believe that any such reader exists; more likely, he just didn't think carefully about what he was writing.
The text continues with an anecdote about how a malfunction of the `little rubber gizmo' once cost Douglas Hofstadter an extra $300 water bill. What is the point of this whole one-page passage that contains both the feedback description and the anecdote? Is the author's purpose to describe the normal, effective working of a flush toilet, as one small but relevant step in the exposition of his theory of mind? (In that case, the stuff about the `little rubber gizmo' and the $300 anecdote are irrelevant and distracting.) Or is the purpose to present a simple example of how a mechanical system with feedback in it can malfunction, thus making its malfunctioning the main point of the example? (In that case, surely the text ought to have started off with a sentence such as `Here is a simple example of how a mechanical system with feedback can malfunction') Or did the author just not bother to think too hard about what role this example served in the book? (That is what I suspect.)
Next in Chapter 4 there is a passage about anthropomorphic language, ie saying that such a feedback system has `desires' and `goals'. The text seems to start by saying that this a bad thing but ends up by saying that it is `indispensable'. The reader gets the feeling that this is a mere hint of a theme that the book will probably cover in more depth later on (and indeed in a later chapter questions are asked such as whether a machine can perceive, be creative, have opinions etc). However, the author doesn't link up this short passage explicitly to any other part of the chapter or to the rest of the book; he just stops after two pages and goes on with something else.
Third comes description in about three pages of two different situations: where a microphone amplifies its own sound coming from a nearby loudspeaker, and where a videocamera is pointed at a screen that shows images of the camera. The reader might reasonably assume that the anthropomorphic passage just gone was an intermezzo and that this third passage was the next step after the opening passage about the flush toilet in developing the analysis of feedback that is necessary to explain the author's theory of mind. However, Hofstadter doesn't say that; he doesn't relate the flush toilet to the microphone/loudspeaker combination in any way, except that he uses the word `feedback' for both. But they are quite different cases: the first is one machine so designed that, if functioning properly, it ensures a constant level of something (in this case water); the second involves two quite distinct machines, not primarily concerned with maintaining a constant level of something, which, if by chance juxtaposed in a certain way, as occasionally happens, can produce interesting results. True the word `feedback' can reasonably be used of both cases, but that is nothing like enough linkage to be meaningful without further explanation: you might as well write a passage about coal mines and then one about explosive mines, and assume that the connection was obvious because both were about `mines'.
It may well be that in order to understand Hofstadter's theory of mind you have to understand the microphone/loudspeaker case and in order to understand that you must first understand the flush-toilet case; maybe, but if so, he should have explained that linkage. And if not, he should not have forced the reader to study the case of the flush-toilet - including malfunctions - for, as it turned out, no purpose at all.
These six pages (pp51-56) provide examples of several generic clarity-hindering faults that can crop up in any text that explains something complicated:
- Failure to explain something from a consistent level;
- Inclusion of material(such as personal anecdote, example from another field, analogy, parable, imaginary dialogue,thought experiment, or some adjacent but different topic) that is irrelevant to explanation of the main topic. The reader may waste a lot of mental energy before concluding that the material is irrelevant.
- Failure to explain specifically how some piece of material (anecdote, example etc) that plausibly does have some relevance fits into the explanation of the main topic. The reader struggles to the tantalising position of concluding that the material probably is relevant in some way, but being unsure exactly how.
- Failure to organise a chapter in a form that is both coherent and clear to the reader.
In the chapters beyond Chapter 4, ie most of the book, Hofstadter puts forward his ideas that `we ourselves .. are strange loops' and that `the quintessential example of this phenomenon', ie strange loops, was discovered by the logician Gödel (pp103-4). He then describes Gödel's work in a chunk of some 70 pages, and constantly refers to it in the 180 pages after that, while never missing a chance to be contemptuous about Russell's Principia Mathematica.
So then, if Hofstadter's explanation of the flush toilet is so poor, how does he do at using advanced concepts of mathematical logic to present his original ideas on the slippery subject of the nature of the mind? Not well. Going through the heavy part of the book, I don't find a clear explanation of a theory of mind, and I do find plenty of the generic clarity-hindering faults given above. There are some whoppers: for example, Hofstadter gives us his personal ethical views on such things as vegetarianism and Albert Schweitzer, as if that material will somehow make his theory of mind more clear and more convincing. But no, life is too short for any more of this.
The book attempts to explain the author's theory of mind. It fails to do that competently. Therefore it can't deserve any more than two stars.