Here's a long review of the book and Robin Hanson's replies to various reviews.
It seems to me that a world in which brain emulations are possible would probably look quite different from what Hanson imagines.
1. One of Hanson's key premises is that only very few people will be emulated, namely those that are best suited to be productive and obedient workers. But why would the technology be restricted to emulating only a few people? Being emulated sounds like one possible path to immortality, so I suspect many would want to be emulated and be willing to pay for it. Also, especially given that you could have many copies of you, it could be easy to get a loan to be emulated (which would be repaid by the work done by ems later on). So, income restrictions shouldn't be a massive stumbling block. This in itself would push the em scenario in a very different direction from what Hanson imagines.
Corollary 1: Would identity theft be a problem? Once a person has been emulated, they could be further illegitimately copied at low cost, and the pirated copies used as slaves. Current difficulties to enforce copyright laws seem to suggest this might be difficult to prevent.
Corollary 2: The em scenario looks somewhat similar to David Brin's Kiln People. What did Brin get wrong? Why wouldn't the em scenario look closer to the Kiln People rather than The Age of Em?
2. Neglected topic: Why not colonize the galaxy with ems? Currently it's virtually impossible for actual humans to colonize even Mars. With ems it would be possible to colonize the entire galaxy in about 1 million years (which is long by human history standards, but very short by astronomical time frames).
(a) How would a society of ems spanning the galaxy look like? (There's that fun Krugman paper for a start.)
(b) This connects the discussion about ems with Fermi's paradox and Hanson's points about the great filter. Why haven't aliens (who presumably emerged millions if not billions of years before us) already emulated their brains and colonized the galaxy?