I’m not sure if this book paints a plausible version of the future, but the path it takes //is// followed mostly to its conclusion. The premise is that moving hundreds of years into the future, the majority of conscious life implements itself in far more efficient media than our very slow neuron-based brains. Life then flourishes in such an environment and begins to disassemble the planets to use as raw matter, to reassemble into a substance something called “computronium” which is supposedly the most efficient configuration of matter for performing computation (and so for implementing structures that themselves implement brains, consciousness, //life//). The author doesn’t elaborate on this, though, and I wonder: aren’t there many kinds of computation for which different structures are best suited? There are many, many components to even the simplest computer nowadays, and one of our computers would be downright primitive (hopefully in terms of structure as well as simple miniaturization and speeding up components) in ten years, let alone 100 years. There’s memory, a whole slew of different arithmetic and logic units, registers/caches (which are different kind of memory), buses for data transfer, specialized processing for integer math, floating point math, 3-D graphics, and so on. Maybe the idea behind computronium is that we’re only referring to a substance for the efficient implementation of neural networks, but that still seems rather limiting. There are some things neural nets are very good at, and other things we just want to implement as specific circuitry. So it seems to me a little more structure is involved. Even our brains, which use only neurons, use many different kinds of neurons organized in different ways to organize different structures. For example, brain structures for processing visual information, auditory information, language, and so on. I don’t even want to claim that, given our brains nowadays being only neuron-based, there’s a reason we’d want to do complicated numerical mathematics using neural nets, even though incredibly some rare people //can// take cube-roots of 20-digit numbers in 10 seconds or so. I’d say that takes a tremendous amount of practice and abnormal type of focus. Still, a dedicated circuit could do that sort of thing in a millionth of a second, with no practice or learning or trial-and-error to get the method down, and so relying on neural nets is wasteful. Don’t take the planets apart for //that//, please.
Consider that you’re a conscious, intelligent entity implemented in some computer. You have a neural network a thousand times bigger than a standard human brain. It’s implemented electronically, so it runs ten thousand times as fast as a human brain. What does that mean? How can we on the outside conceive of what your brain is like on the inside, of what it’s like to be you? Even if you had only the same number of neurons as we did, they could operate much faster. Instead of chemical signals being released from vesicles into gaps, you’d be all electronic. Let’s say that would make you 10,000 times faster. So you’d be able to, say, read a book in 1/10,000 the time I could (assuming you’re not even optimized for reading or taking in data some way other than visually using your brain’s “OCR” ability, which you probably would be). Four hours of my reading time would take you 1.4 seconds. Those 1.4 seconds would probably still seem to you like 4 hours, just that the speed of your thinking would mean the world around you would be at a virtual standstill. You could take your time and do years-long analyses of simple problems, then act before a couple hours had passed. Your brain’s mood-affecting chemicals could easily be altered (as would have to be the case) so that you could focus your attention seemingly for days at a time (two full 24-hour periods, subjectively, would just take up 17 seconds of our time), with no distractions (like all the maintenance-work we have to do here to keep our bodies fed, bathed, exercised, and otherwise taken care of) and there’s no reason you’d need sleep or even get tired. You could get a whole lot done (whatever kind of work is meaningful to you), and that’s with just the same number of brain cells as we have, and limited memory and mental computation abilities. No reason to stick with that number, to not have some reliable mental RAM, and a calculator with which you’d just //think// numbers at and immediately //know// the answer.
That’s just a singularly narrow attempt at trying to understand what it would be like to expand in a small way beyond the limits that our cell-based brain hardware imposes. I’m making all sorts of assumptions about those limits, too. (But that’s a discussion for another time: For example, why can some people seem to read extremely fast, supposedly without skimming? I can skim material quickly, but don’t gain much out of it save for a general idea of what the writer was communicating; I prefer to read every word. If skimming is adequate for extracting information from a piece of writing, then we should create a form of writing which allows an author to write material in a form of shorthand which, when read at “read-every-word” speed, conveys just as much information as skimmers reading ordinary English (and therefore skipping most of the words on the page) would absorb.)
Back to the book. There’s a wealth of ideas here, but the imagination and realization (conveyed by the author’s descriptions of the technologies, devices, environments, etc.) seem halfway there. Rough sketches; inadequately described. Like an image the author himself doesn’t see clearly or completely in his mind’s eye, but wants to invoke anyway. So I’m left without a vivid scene in my own mind’s eye with which to appreciate the story, and that’s probably what it is about the style, the information between the words, that bothered me. It’s possibly less so the words themselves which are at issue than the un-clarity of what’s being described: the lack of a unified vision. Since many s.f.-ers seem to regard Stross’s books highly, perhaps it’s just that he doesn’t speak my exact language, doesn’t match my protocol for painting images with words. Again that’s not to knock the interesting ideas discussed, but just that this all seemed like an agglomeration of things I’d read before, and the plot wasn’t compelling enough to make up for that.