Scalable Systems

I know of only two infinitely scalable systems ever built.

One we know, and many people will correctly guess this when asked. It is the internet. The internet started up October 29, 1969 when the first two systems, located in Stanford University and the University of California, connected. Since then, this network has been growing continually:

1. October 1969: 2
2. December 1969: 4
3. December 1971: 23
4. October 1972: ARPANET went public with a demo linking computers in 40 different locations.
5. currently (2006): estimated minimum size of 500,000,000

(source: http://www.isc.org)

It is stated from the communication of the original builders of the first nodes in the Arpanet that of the original components of this system, software and hardware, none is present in the current system. It is a system that has grown, and continues to grow, exponentially. It serves as probably the best example of a complex system available to us, although an important aspect of complex systems, namely feedback and learning, is still very rudimentary.

The second system is one very few people know, and to my amazement even the people who know of this system do not know that it is a system that might be categorised as infinitely scalable.

What happened on the 29th of October in 1969 can be called a seminal event. There is another one, not far distanced in time. October again, but now the 27th, in the year 1972. The location is Xerox PARC (Palo Alto Research Centre) in Los Angeles. The project team sits around a NOVA computer with a screen on which was typed:

3 + 4

With a speed that was described as “glacial” the answer was printed on the screen:

7 (to the delight of everyone it was the correct answer!).

Now this may not seem to be “seminal” at all, but I think it was. What really happened was that this system which was asked to perform this simple computation had just been bootstrapped for the first time from an interpreter written (of all languages) in one page of BASIC code by Alan Kay (but Dan Ingalls really got it working). And from this moment in time that it has bootstrapped it has never (***never***) been down, and all instances of it found today are living morphs of that system.

I have on my laptop a direct descendent of it, called the VisualWorks development environment. Another one is Squeak. They have come to be by sending messages to that system that bootstrapped in 1972, over a period of now almost 35 years. Any application that has been delivered using Smalltalk, and this for example includes all VisualAge products from IBM, has seen the light of day by doing exactly that: sending messages to a living system. The Smalltalk system was and continues to be: a living system.
This living system has been put to sleep often (saved to disk) but that is how it should be regarded: put to sleep. On waking it continued to be doing what it was doing. Any change, including generating a runtime application, was done by sending messages to this wonderful organism.
I do not know any other system built by humans that meets these criteria of scalability and “livingness”. If you do, let me know.

I place this second seminal event in 1972 on my list of the four root seminal events in my other paper on metaphors. I admit there is a certain arbitrariness in this, because the period in which the event itself took place saw many developments that are strongly dependent on each other, such as the internet, advances in transistor technology, telecommunication, etc.