It is addressed in the book, and it distinguishes between the right kind of iterative improvement and the wrong kind. Norman writes:”It is the rare organization that is content to let a good product stand or to let natural evolution perfect it slowly. No, each year a “new, improved” model must come out, usually incorporating new features that do not use the old as a starting point. In far too many instances, the results spell disaster for the consumer.”

Finding the right people willing to risk the kinks of a new design is one of the toughest jobs at a startup or when launching a new product.

Maybe this is addressed in the book — I haven’t read it — but one problem with ground up design is that it lacks all the iterations the existing product has gone through. First versions of products often lack important design aspects because at some point you run out of money and have to ship. It usually takes many iterations before even the designers are satisfied with the product.

Also new designs tend to lack soul. What I mean by this is that users tend to find all sorts of uses for products that the designers can’t anticipate. Whether it’s using the product in an alternative way to achieve it’s primary function or to solve an entirely different function altogether. New designs are often much more uni-purpose than classic designs; this can be true or just a perception because the design is unfamiliar.

There is a huge risk and expense in brand new designs and what can often happen is that people don’t want them even if they are considerably better. This can be because better is defined too narrowly, or because it is missing key functions for that individual, or for a variety of psychological reasons.