I used to read book reviews on Amazon, but I found that they're often too cursory and sometimes, not fair enough. I read this great entry but I haven't found some truth-worthy book reviews. Jon Skeet's reviews are great, but the number of books reviewed is so small.
Can you suggest me some sources of reviews that I can trust?
I usually depend on Google Books : http://www.google.com/books
They pull reviews from all over the place for a particular book or subject. Give it a try. I agree with you also that Amazon most of the time gives fair and confusing reviews. Also believe that sometimes the review themselves come from the publisher or publishers distributors instead of actual factual reviews.
The other day I bought a book that I actually had to return (first time I ever returned something to amazon) from a HEADS__ series, bad book overall, I couldn't even read one full page. Hope this helps.
One source of good book reviews that I pay attention to is Scott Mitchell's book reviews that usually come towards the end of his Toolbox column in the monthly MSDN magazine. The most recent issue included a review for a book on refactoring SQL that looked really interesting.
What I find Amazon good for is the ratings. The reviews themselves like you say can be suspect but the overall rating I find tends to be pretty good.
I usually don't go beyond that to tell you the truth. If the book is good and covers the subject I'm interested in, well what more do I need to know?
Sometimes reviews can be helpful in clarifying points or elaborating on the synopsis but I just use that as a filter not as a primary source for information.
Beyond that I look at authors (eg Josh Bloch for Java) and buzz. Buzz is the sum total of opinion on the net. From going through dev sites, blogs, sites like this and so on you get a feel for what books currently have developer mindshare. Java Concurrency in Practice and Effective Java both had excellent buzz and that's all I really needed.
It's so hard to take an individual review--even a "fair" review--and take the opinion as gospel. You have a different agenda to the reviewer, come from a different background (probably) and I've seen mediocre reviews for books taht I thought were great.
I've found that the aggregate reviews from threads on here (and similar places) are good indicators. When lots of people use similar terms to describe a book, that seems to be a good indicator.
I find that Amazon gives a fair indication, through the rating, of a book. Although, of course, sometimes the reviews themselves are a little less than useless.
Google books is an excellent resource for recommendations, as are the various blogs I read (principally Jeff's Coding Horror, though books from authors of blogs that I read are usually welcome purchases too, Dave Shea Ccsszengarden.com : The Zen of CSS Design), Dan Cederholm (simplebits.com : Bulletproof Web design) Jeremy Kieth (adactio.com : DOM scripting) and so on.
Also Slashdot has the occasional useful review.
Since a couple of years I do not trust any reviews, whether it's about hardware, software, books or anything else. There is now too much discrepancy between a review and a practical experience. It goes both ways: things that generally have good reviews turn out to be a disaster, and items with poor review often happen to be something marvellous.
Only comments in forums and blogs may shed some light, especially if the pattern is the same throughout international resources.
You definitely can't trust reviews of cell phones any more, doesn't matter who writes them. Everyone has his agenda these days.
As for Amazon, well, the last time I read a guy give one star to a Samsung smartphone because he only saw 6 Mb of available memory while there was a 4 Gb memory card shipped along (numbers can not be precise since it was a few months ago). So he obviously could not tell apart internal memory and memory card. If such monkeys are allowed to write reviews, then what kind of trustworthiness can you expect?
Trust the words of people you know or whose reputation you do not question, not just any guy who runs a review site or blog.
Reviews of books can also be very subjective, the opinion of the reviewer will depend on his background and experience. For instance, I have heard comments from some people that they find the C++ book of Stroustroup to be complicated and for the most part unreadable, while I personally love it and consider it to be a high quality work. So, who is right then?
I trust only a "number of reviewers". The bigger it is, the better the chances you'll get an objective opinion.
I know I'm late to this question but I only just found it by accident (searching for a book review!)
I'm so unhappy about amazon reviews that are clearly done by sock puppets and friends of the author that I set up I Programmer to provide independent book reviews.
Currently most of the reviews come from a print magazine VSJ (UK only) of which I'm the editor and so can get at the reviews but I've got the review team doing more just for the web.
Many of the reviews that are up at the moment are a bit on the short side because they were supposed to fit into limited space on a printed page but web only reviews can be as long as the reviewer likes.
Not all the reviews are perfect I admit but you can believe that they are unbiased and as honest as the reviewer can manage.
mikej