I hesitate to defend Blurb, but I've produced 21 copies of 13 different books with good satisfaction. Certainly the quality of the CMYK printing is not equal to most alternatives. And Blurb is definitely inconsistent. Identical books printed at different times or places (US vs. Europe) often have easily visible differences. And Blurb often creates garbage. Of my 13 books, I've rejected 3 and Blurb reprinted at no charge, with no argument. Two of those had bad color and tone, one had physical damage to several pages.
But consider that you get what you pay for. Blurb is cheap, from 30% to over 200% cheaper than their competitors. That's a big factor many overlook. With the exception of AdoramaPix books that are printed on "real" photo paper (and cost a small fortune) all the others use the same CMYK process. Maybe one has a consistently better CYMK process than the others, but is it twice as good? I don't think so.
After my second Blurb book, I made a "test" book, using several images printed in 3 or 4 versions. Each version had a slightly different curve adjustment applied, trying to match the soft proof. As a result of that test, I now apply a very slight s-curve to images I submit. The curve lowers RGB value 76 to 68 while anchoring the curve above 175. It's not a perfect restoration of blacks and shadows, but it's as close as I can get to my sRGB monitor display.
I also include a small synthetic version of the Macbeth colorchecker as the last page of every book. I can then compare each new book with older ones, and have a good basis for rejection if needed.
And I have to respectfully disagree with Andrew. There is good reason to soft proof. It will give you a reasonably good indication if your image is printable to your satisfaction. If you are not satisfied, there is little or nothing you can do about it. Either live with it, or don't print that image.
In one of my Blurb books I had an image of a Japanese woman in a dark blue kimono. It came out a flat, solid blue. All the folds and texture of the kimono were lost. Really ugly. After the fact, I soft proofed the original image. The out-of-gamut problem was obvious. Nothing I could do about it. I wish I had proofed before printing, because I would not have used that image.