A good book. The first chapter explains how a priori isn't synonymous with analytic or necessary. The second and third chapters are a critique of moderate and radical empiricism, showing that the first leads to contradiction by assuming what it rejects tacitly and that the other, not only leads to contradiction, but also to skepticism. And in the fourth chapter he defines moderate rationalism as foundationalist fallible rationalism. All these four chapters were masterfully written.
But starting from the fifth chapter, things take a wrong way. He downplays the amount of disagreements that aren't resolved through a priori reflection. Then he tries to respond to some the epistemological objections to rationalism, and due to his approach being near intuitionstic, he doesn't say much in the end.
He ends the book with chapter 7, in which he tries to sketch a rationalistic justification of induction which turns out to be the foundational reverse of the pragmatic solution, so instead of saying that induction is highly likely to be true because it has the best consequences a posteriori, he says that induction has the best consequences because it is highly likely to be true a priori, or in his words "The claim, in other words, is that a chaotic world, though perfectly possible prior to the consideration of empirical evidence, is rendered extremely unlikely (in the respect in question) by the occurrence of standard inductive evidence and that it is an a priori fact that this is so".