This isn't a new book, but I only read it recently. After I wrote a post about Stephen Few, Alberto Cairo, and David McCandless, Alberto Cairo recommended I read this one. I did, and think it's great, but I do have a few reservations.
This book covers different types of charts (maps, trees, timelines, networks, etc.) in an encyclopaedic manner. Each chapter is about one of the types of charts. There are small sections on theory, but the chapters are mostly examples—lots of examples.
I found the theory sections quite light. I don't think I'd like to learn data viz from this book; I would definitely choose one of the other two in this post.
I enjoyed seeing all the examples. They are diverse and beautiful. However, my biggest problem with the book is that there is no critique about any of the visualisations. Many of them sacrifice functionality in pursuit of beauty. There is nothing wrong with beauty if your objective is to make artwork. However, if people read this book and think that these examples are appropriate for fast communication of facts, they could easily be led down a wrong path.
I would have much preferred to see more commentary on the examples explaining when it would be appropriate to use each example.
I do recommend this book, but I would recommend getting familiar with the guidelines behind data viz first. Also, if you want to take inspiration from this book, be very clear about what the purpose of your own work is. If it is to communicate facts efficiently, tread carefully with the examples in this book.
What great data-viz books have you read recently? Share your reviews in the comments below.