Gareth Rees takes on that perennial question, how to judge Greg Egan's fiction:

Judged by the standards of the literary novel, Egan’s works fall far short: his prose is dry and impersonal; his characters carry out their function in the story but no more; his plots are often episodic and lack dramatic conflict or resolution; he has a tin ear when it comes to satire. But all of that can be forgiven because he brings to his work a unique interest in the character of physical law. Many science fiction writers pay homage to this subject, of course, but for most the laws of nature are there to serve the story: a discursion on the physics of a wormhole, say, would be for most writers an adjunct to a fantastic voyage therein, but Egan has the chutzpah to imagine that the reader will delight in the physics for its own sake.

It’s this unique quality that makes Egan’s deficits excusable: a reader who seeks a novel that aims to provide insight into the human condition has ten thousand exemplars to choose from, but a reader who seeks a novel that aims to provide insight into the laws of general relativity has only Egan’s Incandescence.

I'm rather sympathetic to this line of argument. The most common response to the Egan defense is that scientific fidelity doesn't excuse poor characterisation, but this strikes me as a shell game; there's plenty of sf with poor naturalistic characterisation that shouldn't be sheltered by the Egan defense (some of it written by Egan, such as Zendegi, the novel that came between Incandescence and The Clockwork Rocket), but as Gareth says, Egan is uniquely committed to this approach, and skilled in its execution.