I State of the Art
In order to understand the organisation and structure of a network like the Internet it seems necessary to develop a new vocabulary of description and explanation.
Normally metaphors are used to describe the Internet like the concepts "path", "infobahn", "datahighway", "architecture". But these metaphors, unless understood in a new and extended meaning, suggest that the Internet is like car driving on a highway. But what for instance are the tires of that vehicle? Or what is the "gas station" of the Internet? The term "architecture" of the net suggests a three dimensional building with an entrance door, windows and certain levels. But what is the roof of the Internet and what is the staircase to its cellar? No one knows.
Some speak of the Internet as an organism. But what is the "brain" of the Net and what its "great gut"? You see that a metaphorical use of terms to describe the organisation of such a complex phenomenon like the Internet is not very well suited for an appropriate understandig of the specific processes which are happening in the Internet.
So what we have to do is to develop an new theory and new concepts for the description and understanding of a complex structure like the Internet which are no metaphorical uses and which are already developed for the description of large, complex and interrelated phenomena. I mean the theory of social systems, the second order cybernetics as developed in the writings of Ranulph Glanville and Heinz von Foerster, the theory of knots (topology) and the theory of social networks.
We can describe the Internet as a system operating in a certain environment. The Internet can be described as a communication system, consisting of certain components between which certain interactions are observable to an observer. The environment provides a structural coupling between the Net and its support from the environment. The environment "holds" the system and provides the relevant ressources and supports to maintain the systems functions.
For instance, the supply of current, the computer hardware, the material cables of the telephon networks, the buildings in which the terminals are housed and of course the user is strictly part of the environment of the Internet. They are no components of the system itself.
All communication systems are operationally closed systems. That simply means that no system can function outside the system. All communication takes place in the system and the only thing which can take place in the system is communication and nothing else.
Therefore the Internet can be described as an autopoietic system which produces the components out of which it exists out of the components of which it consists. The components are single communication units and they produce further communications (replies, answers, e-mail, discussions, news groups, etc.).
In this mechanism of continuous production and reproduction of the various components of the system selective interactions take place. Certain links between certain components are used more often than others. Through that process, which is a temporary, time-based mechanism, a certain actual structure is established in the Internet.
The principial problem one has to face is the unobservability of the processes going on in the Net. The only way to observe interactions with a certain home page or server is the amount of electronic mail, the count of log-ins into a certain page or the numer of subscribers to a e-magazine or a discussion group. So we can only talk in a highly selective perspective which depends on where we are located ,when we look into the Net and how, about the actual structure of the Internet as accessable to a certain observer at a certain moment.1
But we can talk about its organisation as opposed to its structure. Humberto M. Maturana and Francisco Varela have tried to define the difference between organisation and structure. The organisation of a system consists in the necessary relations which define the system. The structure of a system consists in the actual relations between the components which integrate the systems as such.
Therefore the organisation of a system like the Internet can remain invariant while it maintains its identity without disintegration. Structures can vary provided they satify the organizational constraints of the system. (Varela 1984, S. 25.)
Varela, Francisco: Two Principles for Self-Organization; in: H. Ulrich/ G. J. B. Probst (ed.): Self-Organization and Management of Social Systems, Berlin e. a. 1984, p. 25-32.
1If there ist anyone who can help me further on in that question ist asked to.