This is a strange book. Despite the giant word "WEB" on the cover, it's difficult to see this as "best practices for web-based software." Instead, it reads like a guide for designers who've only built desktop software and are being forced against their will to deliver a web-based product.
It's far too long at 658 pages; there are needless sections on general suggestions for designing for the web that are far better written about elsewhere. I understand the authors' desire for completeness, but there's just too much basic HTML here padding out some sections.
And the final *seven* chapters deal with the design of data reports, charts, graphs, and even maps. Now, these are important topics, but they are not such significant parts of most web applications to deserve more than half this book's length. And have these authors never read Edward Tufte? It's hard to imagine a collection of uglier, more garishly colored, visually heavy maps and diagrams than what's presented here.
It's a little annoying, too, that most of the diagram images come from *desktop* applications like Excel or Crystal Reports, not web applications. There's a good reason for that: these kinds of data-intense diagrams tend to be for specialist users committed to spending long hours in an application. In most cases, that's a situation that calls for the more powerful capabilities of a desktop application. When's the last time you looked at a scatter plot on a web site?
But between the dull basics of the first chapters and the mind-bending statistical overkill of the last seven, there are some good and useful sections. For example, there are good rules of thumb for form layouts, handling input validation gracefully, and search filtering. There's nothing adventerous or innovative here, of course. Advice tends toward the conservative and reliable list view-to-object view model (the way your email program works), with a few breaks for product comparison interfaces. As in so many of these kinds of books, the authors also include examples of utterly pointless novelty interfaces (zooming lenses, radial tree navigation schemes, photo "data mountains") that are notable for their near-total absence outside the HCI lab.
The strange thing about this book, and others like it, is the almost willful blindness of what *actually* works in web application design, and what *actual* users vote with their clicks to make successful. Innovative, popular, and usable web applications like Amazon.com, Flickr, Craigslist, eBay, Google's Gmail, or the applications built by 37Signals are nowhere to be found. These applications are successful because they embrace the constraints imposed by the web and HTML (and their strengths), and find ways to support users' tasks that make sense in that environment.