Wired on Friday: A fortnight ago, a small group in San Francisco drank to the tenth anniversary of the first web banner advert. A month ago a group of London web veterans celebrated the (somewhat arbitrarily chosen) birthdate of the British net industry.

As I write, a flurry of tenth anniversaries are upon us.

In December, the W3C organising consortium of the web will throw a party in Boston to celebrate its tenth year. In February, they'll be celebrating a decade of web design in Amsterdam.

But for all the instant global nostalgia over the early days of the popular net, there will be plenty of tech anniversaries this year that no-one will recall. Who raises a toast to a decade of the Virtual Reality Markup Language, a way of creating a three-dimensional cyberspatial net, whose hype was as great ten years ago as that of the web? Does anyone mark the birthday of WAP, the web-equivalent for mobile phones, which saw its fifth birthday last spring?

What makes a technology establish itself and last a decade? And how do we know how far it will live into the future?

That's a question that plays on the minds of those who have to gamble on such time frames. For businesses, it's the stuff of every capital asset purchasing decision. When do you switch out from under an apparently dying software technology, like Novell Netware, which has been dying for as long as the web has been booming, but somehow still keeps going? When do you adopt a bright new system - like "ultra-thin clients", cheap PCs connected to a multi-tasking central server - a business solution advertised as the next big thing for a decade, but which never quite working as planned?

While it may seem that its those of us who consume technology products that bear the brunt, there's parallel pain for those who develop the new tech.

The cost for a technology company like chip manufacturer Intel of keeping ahead is terrifying: each generation of Intel's cutting-edge chip tech requires the invention of new techniques, both in design and fabrication. The company writes off brand-new manufacturing plants to obsolescence almost as fast as they commission and build them.

It's an incredible pace for such a large corporation to maintain. And it's a dice roll they've consistently managed to call correctly, with very few stumbles. When those gambles don't pay off (as with Intel's recent problems creating a design that would move their new 64-bit processor, the IA64, into the consumer market) the company takes a considerable hit. Bad publicity, wavering stockholders, and the certainty that competitors like AMD will take advantage of the stumble, make every new advance a risk. But for the very same reasons, Intel can't afford to stop innovating.

What's peculiar, though, is despite the frantic pace of investment and change underlying computing, how static it seems to be from the consumer's point of view. The first computer of mine to download a web page was a 90MHz Pentium.

Now I use a 2.8GHz machine to do the same thing. In broad terms, that's thirty-two times the power, and yet, on the surface, it does the same thing.

The technology that drives the modern Intel processor really is quantitatively different from the computers of ten years ago. But as the celebrations attest, they both still run the web.

The real keepers in the tech world aren't technologies at all, but niches. The web is a niche - it's an activity that computers do, which technology supports. Intel's chips technologies obsolete themselves because the niches they support - word processing, database handling - grow more complex, even as they appear to stay the same. No-one would be sacked for predicting that spreadsheets will be around in a decade. But a particular version of Microsoft Excel? No.

The trick, of course, when planning for the future, is working out whether what you are looking at is a niche or a technology that merely supports a niche. It grows even more confusing when most tech companies will happily conflate the two in an attempt to convince you that their unique technology is in fact a vital niche.

Is Microsoft's new .Net a technology, or a niche? Microsoft promises you it's an entirely new way of looking at, developing and shipping applications.

Sceptics would say that's just the same old way of selling you Microsoft's latest brainstorm, soon-to-be- obsolescent tech.

My own suspicion is that it is, actually, a new niche, but truthfully, if telling niche from tech was easy, much of the current IT industry would vanish overnight.

There are some gut-level, questions you can ask to tell them apart: what would happen if the company selling you this disappeared overnight, for example? Would I still end up using something like it?

But if you really want to bet on what will be around in ten years time, bet on the niches, not the technologies. New technologies come and go, but niches stay comfortable for decades.