Keeping Networks Stupid

Referring to David Weinberger's weblog on "The Telecommunications Story" , I'd thought I'd throw my comments in as well, being an ex-telecom guy.

Just before I left Telecom Australia (in 1987!), I remember sitting in an internal briefing meeting where someone in marketing had just come back from a trip to the USA, and considering the effects of deregulation on the future of Telecom. We were talking about various "new" technologies (ISDN, Intelligent Networks), and also the merits of circuit switched technologies in general. These were heralded as the key infrastructure of the future, upon which which "value added services" would be built.

I have never forgotten the declaration made at that meeting that:

"we in Telecom must not be reduced to just carrying the bits - that will make us just a commodity!".

So it appears that the implications of the "stupid network" idea have been understood by Telcos for a long time, even the government owned one here in Australia, and they have all been fighting it ever since.

But that is the paradox, as you have noted - this kind of network is much harder to make money out of, being a commodity. But someone still has to invest in it and build it - where will the money come from?

Of course, I've referred to the infrastructure of the Internet, which despite being called stupid, is really quite clever. All those routers have CPUs and memory and an operating system, they do a certain amount of error correction, they decide where the data has got to go when there is a problem somewhere in the network etc. When we talk about the Internet being kept simple & stupid, we are really referring to the Internetworking Protocol, which hides the underlying networking, allows systems to establish communications sessions with one another and leaves the rest to other end-to-end protocols like TCP etc.

Networks evolved in very purpose-built ways because of the need to manage scarce and expensive resources, and because networks were very prone to add a lot of errors or noise. It needed to be designed for efficiency. Technology has now got to the point where data can be transmitted with very low error rates and very high bandwidth - allowing simplification of what the network must do, and lower cost per volume of information transmitted. So we can relax the efficiency principles now, and gain some flexibility.

Maybe the clue can be taken from the way the brain has evolved - lots of inefficiency and redundancy, and massive interconnectivity (but don't take that analogy too far, I'm not one who thinks the internet will develop any kind of conscious experiences any time in the next million years!)

We will be watching carefully what happens in the USA, since Australia will broadly follow the US lead - although there are some important differences, not the least being we've only got 18 million people in a land mass about the same size as the USA.

BTW, David commented that he was "nowhere near as up on this area as just about anyone you'll meet, but it's good to know that ignorance doesn't deter me from making large-scale pronouncements." But with friends and acquaintances like David Isenberg and David Reed, I'd say all the Davids have it exactly right!