The pattern server is a collaborative hypertext knowledge management system built on the WWW.
The purpose of knowledge management is organizational memory. Those who don't have the knowledge need it, quickly. Information technologies, such as books, are meant to minimize the time and distance between the user and the desired information. Consider a reference book, like a dictionary. Everyone knows the basic structure of a dictionary and the kind of information which is stored therein: words, organized alphabetically, and their definitions. We learned by rote when we were very young to know the alphabet and we can therefore utilize a dictionary book efficiently, combined with the fact that the structure of the information is very simple.
Engineers' handbooks are more complex. The various fields - chemical, mechanical, electrical, etc have relatively standard categories of information, tables, naming conventions, etc., which are all taught in school, so it is not too difficult for an engineer to look up the desired information. The many indices help significantly.
So how is the information accessible? The structure of the book is, in a sense, externalized in the form of the table of contents, the various indices, etc., though it is a static structure because it has to be printed. The externalized structure helps us - the searcher - understand the information categories and their relative placements in the book. Hopefully, the idea we have in our head when we go looking matches at least one of the index entries.
So if we don't understand the structure of the reference volume, we can't know where to start. We are subject to the author's interpretation, naming and categorization of the contents. This is effective only when it has been around long enough to be standardized, like the alphabet, like the concept of a table of contents.
But today is about information overload. Computers and hypertext don't help manage information, they somehow compound the problem. It is undeniable that we have access to more information than ever before, and that this is good. This information is. It doesn't have an alphabet for orderly reference and it doesn't have a standard presentation. It's ephemeral and moving at light speed. We want to find what we're looking for in this chaos, just as we want to find the definition of a word or find information about the property of a material, but at the same time, this new information is of a different nature.
Printed books are therefore more appropriate for narratives or information whose structure is already reasonably well known or can be learned quickly.
Imagine if the book or the system could personalize itself to each individual user. By personalize I mean not user preferences but the user's hardwired perspectives on the world. Everyone sees the world differently, and just as "one man's heaven is another's hell", one categorization scheme may make total sense to one person but be meaningless to another.