I liked this book when I first read it, but I didn't like it enough. The review I wrote that follows below is ... dumb. This is a brilliant book -- it just needs to be read slowly. Probably more than once.
Iain M Banks is well represented in my mind-expanding books collection. So I expected there would be more to The Algebraist than meets the eye.
And so there is.
Yes, it's not the equal of Feersum Endjinn. Yes, it can be read as a well done variant of the standard space opera; even the the little twist in the epilogue won't surprise Banks fans. And yes, I must admit,
Only Banks, however, would embed an extended, serious and satirical, reply to Bostrum's simulation thesis in the midst of a space opera (see also a NYT article from last summer). 
First, a bit of background. Briefly, Bostrum uses routine statistical reasoning to assert that it is overwhelmingly likely that "we" (meaning at least you and I) exist in a form of computer simulation. David Brin has argued that the improbable success of George Bush suggests he's the alpha and omega of the simulation, but this theory is not universally accepted.
It's fun stuff. Variants of this thesis have been well explored by several authors in the mind-expanding books collection, but Banks has the most explicitly philosophical exposition.
Banks imagines that "the Simulation" thesis has become the basis for a pan-Galactic "faith", called The Truth. It's a relition with some resemblance to various millenialist cults and low brow Buddhist sects seeking salvation by chanting the name of the Buddah... (Emphases mine, the text below may not be completely accurate )
The Algebraist Personally my experience with, and indirect knowledge of, mortal life makes the "punishment" thesis particularly plausible. On the other hand maybe we're just contaminants in the culture dish, or a forgotten version 0.7a of the simulation that's been left to to run on some obsolete hardware.
...The Truth was the presumptuous name of the religion, the faith that lay behind reality. It arose from the belief that what appeared to be real life must in fact - according to some piously invoked statistical certitudes - be a simulation being run within some prodigious computational substrate in a greater and more encompassing reality beyond. This was a thought that had, in some form, crossed the minds of most people and all civilisations. However, everybody quickly or eventually came round to the idea that a difference that made no difference wasn't a difference to be much bothered about, and one might as well get on with (what appeared to be) life.
The Truth went a stage further, holding that this was difference that could be made to make a difference. What was necessary was for people truly to believe in their hearts, in their souls, in their minds, that they really were in a vast simulation. They had to reflect upon this, to keep it at the forefront of their thoughts at all times and they had to gather together on occasion, with all due ceremony and solemnity, to express this belief. And they must evangelise, they must convert everybody they possibly could to this view, because - and this was the whole point - once a sufficient proportion of people within the simulation came to acknowledge that it was a simulation, the value of the simulation to those who had set it up would disappear and the whole thing would collapse.
If they were all part of some vast experiment, then the fact that those on whom the experiment was being conducted had guessed the truth would mean that its value would be lost. If they were some plaything, then again, that they had guessed this meant they ought to be acknowledged, even - perhaps - rewarded. If they were being tested in some way, then this was the test being passed, this was a positive result, again possibly deserving a reward. If they had been undergoing punishment for some transgression in the greater world, then this ought to constitute cause for rehabilitation.
It was not possible to know what proportion of the simulated population would be required to bring things to a halt (it might be fifty percent, it might be rather smaller or greater), but as long as the numbers of the enlightened kept increasing, the universe would be constantly coming closer to the epiphany, and the revelation could come at any point.
The Truth claimed with some degree of justification to be the ultimate religion, the final faith, the last of all churches...
...It could also claim a degree of universality that the others could not. All other major religions were either specific to their originating species, could be traced back to a single species - often a single subset of that species - or were consciously developed amalgams, syntheses, of a group of sufficiently similar religions of disparate origin...
... The Truth could even claim to be not a religion at all, where such a claim might endear it to those not naturally religious by nature. It could be seen more as a philosophy, even as a scientific postulate backed by unshakeably firm statistical likelihood.
There were some potentially unfortunate consequences implicit in a profound belief in the Truth. One was that there was a possibility that when the simulation ended, all the people being simulated would cease to exist entirely. The sim might be turned off and everybody within the substrate running it would die. There might be no promotion, no release, no return to a bigger and better and finer outside: there might just be the ultimate mass extinction...
It's good fun to imagine variations of the theme of "what's the simulation being run for", though by now I think the topic has been pretty well explored. 
Oh, I'd be remiss if I didn't also mention that the simulation theory is one answer to the Fermi Paradox (see also); we are alone because the "purpose" of the simulation requires it. Not coincidentally Deism has the same answer to the Fermi Pardox (God only wanted us); an inexplicable omnipotent deity or an alien uber-geek are but two sides of the same coin. Indeed, one might even speculate that the Fermi Paradox is a bit of circumstantial evidence for the aforementioned coin.
Thanks Iain, please do keep up the good work.
 The ninth page of "Four: Events during Wartime" in my paperback edition.
 Perhaps you imagine I typed in that long excerpt. Of course not. I Googled on some key words and found it had been typed for me. Hmm. Seems a bit too easy. What other clues could be on that site .... (cue music).
 One of my favorite variations came in a book from, I think, Greg Egan (also on the list). In that book Egan pummeled the meme from several directions. In one exercise a simulation is created with an intentional inconsistency; the laws of physics of the simulation are so absurd that it is truly impossible to create a self-consistent "theory of everything". The inhabitants will be crushed by absurdity, and perhaps forced to recognize their universe cannot be "real". Alas, the simulants are smarter than expected, and through staggering brilliance they resolve the paradox. Their breakthrough makes their simulation self-consistent, severs the newly independent universe from the (recursive) simulation that it was hosted within, and condemns some of their uber-geek deities to eternal damnation. [Update 6/16/09: The book is Greg Egan's Permutation City.]
PS. The Amazon reviews say this book is outside of Banks "Culture" universe, but it could be read as the pre-history of something that might become a kin to "the Culture".
Update 1/18/07: On first posting I wrote that the plot didn't seem to hold together all that well. I was particularly thinking of certain aspects of the ending. On reflection, I think that's still true of the resolution of one subplot. On the main plot, however, I now think I'd underestimated Mr. Banks. I should have remembered from his prior work that there's always a hidden agenda to be uncovered. The peculiar course of Fassin Taak's condition does make sense in the context of the schemes that operate between the pages.