All- thanks for the timely and sharp comments. Will try to address some of the points made.
OpenID and single-sign on systems of many flavors are also great ways to avoid sign-up forms. They aren’t mentioned in this excerpt of the book but are discussed in the book itself. So I agree they can alleviate the multiple sign up forms problem.
As Jeffrey has eloquently pointed out the primary use case for this best practice is allowing new customers to get in and try a web application or service. Usually the only option people have is sign up or not. I think that’s broken. It leaves you with one choice -you are in or out. Most times you haven’t even seen the application yet and can’t really make that choice.
Other contexts might very well require a sign-up form which as others have pointed out benefits from being streamlined and well-designed. That’s what the rest of the book is about :)
Lastly, as the last example (or counter-example to be precise) of myplan points out, I’m not advocating turning forms into wizards. I’m advocating experiences that get people engaged in products first, then turn them into record sets in a database second. This does not mean making each question in your form a separate page. It means allowing people engage with and use your product before so they can decide whether or not to sign up as a member/customer.