Network Theory

Network Theory has been the holy grail of the activity described in this blog. I coined the term NET8 (pronounced Net Theta) to designate this research
around 1998.

Meta-artificial is the name of this blog-site coined as soon as I became aware that my research about Net modeling was getting somewhere.

I picked the word "meta" to indicate the vision that the Net (as we say the Computer of Alan Turing) gives rise to a new kind of
Artificial Intelligence. I thought so being influenced by the work of V.F.Turchin under the term Metacomputing since the Turing/Von Neumann Computer gave rise to the vision of A.I.



Surely, there is a confusion with these terms. Also, I coined the term Internet Systematics which became the name of this blog-site because I thought of it as my first research result. It signaled the discovery of a systems approach to modeling the Net, a goal I set to solve being confronted in late 80s by two competing network technologies the OSI and the TCP/IP (I was the national rep in EU's COSINE and RARE activities in Networking).



I began the research problem to look for principles very similar to what John Day calls Patterns in Network Architecture - a return to fundamentals but took a different path because my initial conditions were very different as I lacked networking experience. My quest for fundamentals did not have the solid ground he was standing upon. My statement of the problem was fixed as a by product of my work in Declarative Programming (or Functional or Applicative or ..) otherwise it would be a total nonsense. I said to myself I used DP to understand what
a Programming Language is (via Denotational Semantics) why not pursue the same course to understand
the Net (this is a very linear version of the thinking process that took place over a long period).

But networking experience gradually accumulated with me as I was involved in large scale Network building projects and joined the global community of Internet developers (for example the RIPE community, the CERN community, the EBONE community etc) participating in network working groups.

After several years of participatory observation I managed to build a Model of Networking which I describe by the term Virtual Von Neumann Recursive Architecture - VVNRA (the term recursive replaced my initial term "sequence" or "chained" under Day's influence) motivated by the need to understand the "whole beast" I was handling.

The VVNRA result was a kind of dead-end did not know how to discuss it with the community, which community ? the engineers, the computer scientists ? the general systems people ? the power users ?.

I did try with a professor, talked some researchers in my work but had no luck (perhaps did not
pursue it properly).

I concluded that the results I had was some kind of pattern, an evolution pattern of the Net. One day googling with the terms "pattern" "architecture" "internet" I found
John Day's RINA (note initially his term was PNA - Patterns in Net Architecture) it provided legitimization for my usage of the concept of recursion, I re-called my 1999 result on the issue of Net modeling.

The same can be said about the concept of Virtualization a key feature of my Model above. In recent times we all know the importance this concept has for Network developments. At the time again I locate it as something vital that stemmed from the way the Net worked. I did not move any further than the assurance that I had constructed a Picture for the Net (see posts of earlier years here).

A theory needs predictions and I made some no doubt (for example the early detection of the importance of multimedia, the early WWW success, the information collection as killer Net application.

An important result was the mapping of Network's evolution phases (see older posting here)
detecting the Strowger Switch
as the initial phase along with the principle of the Quantum of Automation (embodying automation, see older postings in this blog).

In fact the model Virtual Von Neumann Recursive Architecture resulted as an abstraction over the above evolution path.

John Day in his analysis about the fundamentals cames to IPC - Inter-Process Communication (*)
in fact he talks about Layered IPC where the lowest layer is the wires.















In the diagram above APPL 1 is one application wishing to communicate with
another one APPL 2 using a facility IPC. In fact the Layer APPL
is very like the Layer IPC, just processes talking to each other.
So one recursive layer that unfolds with respect to scope, bandwidth, QoS:

IPC n over IPC n-1 over IPC over n-2 over ....

This is very similar to mine:

VVNRA n embedded over VVNRA n-1 embedded over VVNRA n-2 embedded over ....

I thought of this in 1999, the Net at phase N-1 becomes the
vBUS for the Net at phase N.

How come ?

Think the Net prior to DNS with manual naming with ftp HOSTS.txt,
from NIC and so on (see older posting).

So DNS is the introduction of some automation to the Net,

and what is automation ? well the Turing machine thing ?

Can I model onto it ?

Difficult. So move on to next near thing the "Neumann/Turing design".

Let me make it an abstract concept, as the diagram CPU-BUS-MEM shows.

Can I map ? Well let's see:

A Quantum of Automation occurs with DNS, which seems to form a Virtual Von Neumann Architecture where running the bind code etc is the vCPU loaded with this code,
the ZONE servers are its v MEM) and its vBUS is the existing Net used to
propagate the new phase.

Most Net advances fit this pattern going back to the manual switching
rooms automated by Mr. Strownger for the specific purpose of throwing out
the intermediaries, namely the phone ladies that stole his business away.

BGP Autonomous systems did a similar task for core routing.

MIME offered Multimedia to Email,

ARCHIE offered Networked Resources (via anon FTP)

WWW improved the "FTP + multimedia" manual facility and so on.

My diagram recurses over the BUS of the VN - architecture:

















(*)Bob Metcalf 1972 : Networking is IPC,