Rupert Goodwins' Diary

Monday 16/10/2006

There's no exact day to mark it, and plenty of arguments to be had about the details. Nonetheless, 25 years ago TCP/IP was born — and with it, the beginning of the Internet. As the Internet Society says:

"Two of the core protocols that define how data is transported over the Internet are now 25 years old. The Internet Protocol (IP) and the Transmission Control Protocol (TCP), together known as TCP/IP, were formally standardised in September 1981 by the publication of RFC 791 and RFC 793. "

If nations are languages with armies, then networks were protocols with marketing departments — until TCP/IP came along. For the first time, the people behind the standard had set out to be as general as possible, to make no assumptions about hardware or operating systems, but instead to create a lingua franca that had no other purpose than to link together whatever it could. The other networking standards around at the time — IBM's SNA, the ITU's X.25, the deep and mysterious SS7 — were tied to commercial interests, were expensive and complex solutions to badly defined problems, or were plain wrong.

TCP/IP was both utilitarian and visionary. It was also hotly opposed: when as a green young hack in the early 1990s I first sat in on presentations with networking companies, I soon found three approaches to TCP/IP. One was to ignore it altogether, and present whatever the company was flogging as the only such thing in existence. (This is still the preferred mode of most companies pitching their wares, and always, always wrong.) One was to dismiss it as transparently unsuitable for serious work; an experimental, flaky and largely unsupported flight of fancy that had no business pretending to be in the same world as the acronym explosion of the real stuff. The final way was to say yes, it's interesting, but it'll never get critical mass. Why, everyone uses the ISO standards — or why, IBM's got its networking all sewn up. Why would they change? Especially for something that didn't even have a real committee behind it, and no commercial imperatives protecting it?

Like a judo master, TCP/IP took those very weaknesses and revealed them as strengths. There were no government committees, seeking consensus from industry and abroad. There was no marketing department, double-checking every proposal in the light of existing and foreseen revenue streams. There were simply engineers and academics, dedicated to designing and creating the universal network, and all they asked was to be left alone to do it.

With nobody important or well-monied caring, the same people were free to give away everything they did. Which they did. The only things that mattered: rough consensus and running code.

And that won. I once tried to write a science-fiction story where everything the proponents of strong intellectual property law enforcement claim was actually true, and the smallest act of creation resulted in something that was rigorously protected: the only society I could come up with where that worked was stultifyingly feudal. As far as I could work out, the burden of enforcing those rights would utterly stifle most of what in the West are considered essential human freedoms, including the abilty to create anything at all of value. It made for a very dull story indeed.

The application of copyright and patent law would certainly have killed TCP/IP. It is a resounding irony that the value of nearly everything the big media companies are trying to control exists largely because of the Internet, which exists only because TCP and IP exist, which exist because they are free.

Man too is born free, but everywhere is subject to the terms and conditions included hereunto in the End User Licensing Agreement. Change the record!