This web page is licensed under a
Creative Commons License.
(Copied from an old review originally posted elsewhere.)
Executive Summary: A solid, non-dark, hard-SF exploration of some of the things that may happen when we get computers powerful enough to accurately simulate intelligent systems. Raises some fascinating philosophical questions, and explores the edges of them, but introduces some untenable magic near the end for the sake of the action. Recommended to the general SF reader, and especially to cellular-automaton and artificial life fans.
Background: Western civilization, circa 2050. Not much has changed in fifty-odd years; not really enough to be convincing. Egan had to set his story far enough in the future to get really really powerful computers, but he has no particular interest in the politics, sociology, or otherwise of this future world. That's OK.
Computers are ubiquitous and highly networked, and computer time is available in massive quantities on the QIPS ("quadrillion instructions per second") market. A programmer will casually add a few billion computing elements to a computation on the chance it'll be useful, quasi-intelligent programs are available to filter your incoming mail by how you'd feel after reading it, automatic program-verification finally seems to be working. That sort of thing. It's possible to create a Copy of yourself in the global computing net, if you can afford it.
These Copies have no legal status at the moment, although since some of them are Copies of very powerful individuals whose originals are no longer around, they are beginning to have considerable influence. Controversy and upheaval are expected, but they form no part of the plot, and we never find out whether or not they actually happen.
Story and Storytelling: Good unobtrusive prose (except for the constant misspelling of "bail out" as "bale out"), plausible background tech, non-flat characters. A few subplots that never get tied up (Maria's mother, the missing four weeks of Peer, and even the long Riemann thread) suggest some hasty editing, or perhaps just some subtlty of resolution that went beyond me.
The first 230 pages or so are the best. We are introduced to the world and the idea of Copies, Egan plays with some of the implications of having ghosts in the machine, some of what it might feel like to find yourself a Copy, some of the things Copies can and can't do. There are some very nice touches here; the wonderfully oxymoronic "Solipsist Nation", for instance, is a meme to watch, as is the idea of people who can directly manipulate their own states of mind, and even their basic personality traits; we can only wish that Egan had gone even more deeply into this.
We meet an artificial life / artificial universe devotee who spends computer time she can't really afford playing with a toy universe modelled as a complex 3-D cellular automaton. Anyone (like me) who has spent guilty hours building and being mesmerized by CAs and artificial worlds on today's puny computers will strongly identify with, and strongly envy, her.
The last section of the book is somewhat less satisfying; some of the characters have managed to get themselves into a pocket universe with virtually unlimited computational power, and after seven thousand subjective years or so, a Terrible Problem suddenly develops. Even if we're willing to believe in this universe, the use Egan makes of it is narrow and unfulfilling. Some thousands of people have had some thousands of years of living essentially as dieties in an environment that they can control down to the level of physical law, if they so choose; any author introducing such a thing is obliged to either show us some of the wild transhuman things that will have happened, or at least to convince us that they exist, but we wouldn't be able to understand them if he were to show them to us. Egan does neither; he is heading so determinedly towards the action in the last few pages that he doesn't seem to have noticed the magnitude of the engine he has created to get us there. Which is a pity.
Philosophy: Egan uses the novel to address by example some of the questions that philosophers like Dennett and Hofstadter like to play with: the relationship between brain and mind, and between brain and consciousness, the location of mind, and so on. If we are conscious because of the information flows in our brains, and if a sufficiently-detailed model of a brain running in a computer would also be conscious (which is a background assumption here), what happens to that consciousness if the brain-model is run at strange time-scales, or on two machines at once, or in finely-divided subcalculations on machines randomly scattered across the globe? That sort of thing. The questions are fascinating and classic, and Egan draws some straightforward hooks for us to hang our thinking on.
The main problems of the book, as with all too many such books, occur when Egan feels obliged to have the philosophical questions lead to some action. In the end, the links he makes between the two are questionable and not terribly convincing: the "dust theory" that underlies the last part of the book is either an extreme panpsychism (in which not only does every object in the universe have consciousess associated with it, but so does every possible way of dividing up the universe into objects, across all of time), or the tautological observation that a fiction writer can choose to tell any story e wants, from any viewpoint.
I have no large complaints about the theory itself, but it should have no power to motivate action; the things the characters do to try to ensure that some model of themselves continues to run on the universal computer are pointless, in that those models are already running, and have always been running, anyway, as have all possible variants of them. Egan may have noticed this at some point; one character mentions the fact in passing, but they all continue inexplicably working on the project regardless. And the literal immortality that some of the characters eventually achieve (or have the opportunity to achieve) is not in fact possible; in casually assuming that it's possible to map an infinitely long life onto the state-space of a finite universe, Egan is ignoring something that one of his characters points out: the difference between a very long time and forever is a difference of kind, not of degree.
And finally, the Problem that the characters face at the end of the book seems to be motivated by a desire to have some Tom-Clancy-like action, and a climax. But the Problem stems from a category mistake; having incorporated the strong AI hypothesis (correctly-connected matter gives rise to, or just is, mind) as an axiom from the start, Egan slips at the end and gives consciousness some magic powers, as if the universe notices conscious systems as special, and allows them to, for instance, have effects one level up, in the universe containing the computer that is running the universe they are in. Egan certainly isn't alone in this; it's no doubt based on some muzzy (mis)understanding of the uncertainty principle and wave-function collapse, and others in his position have done the same sort of thing. But the book would have been better without it.
Moaning about the philosophical imperfections aside, this is a strong, enjoyable book that offers some well-worked-out handles on some of the most interesting questions that hard SF can address. If you've ever stared at a cellular automaton chugging along, and dreamed what it'd be like to live in there, or wondered what sorts of religions the inhabitants might develop, or even if you never have but think it sounds neat, I recommend "Permutation City" to you.
(I also recommend it for the almost-coherent 20-line poem in the frontispiece, consisting entirely of anagrams of the title; looks like Egan has discovered the Anagram Server!)
%A Egan, Greg %T Permutation City %I HarperCollins / HarperPrism %C New York %D 1994 %G ISBN 0-06-105481-X %P 341 pp. %O paperback, US$4.99
This web page is licensed under a
Creative Commons License.