Some of the most revolutionary new ideas come from spotting something old to leave out rather than thinking of something new to put in. The Sony Walkman, for instance, added nothing significantly new to the cassette player, it just left out the amplifier and speakers, thus creating a whole new way of listening to music and a whole new industry. Sony's new Handycam rather brilliantly leaves out the zoom function on the grounds that all a zoom does is cost money, add a lot of bulk and render every amateur video ever made unwatchable. (They might, while they're following this line of thought, consider marketing a record-only video player, and video companies might consider releasing movies that are actually recorded in fast forward mode.) The RISC chip works by the brilliant, life-enhancing principle of getting on with the easy stuff and leaving out all the difficult bits for someone else to deal with. (I know it's a little more complicated than that, but you have to admit, it's a damned attractive idea). A well-made dry martini works by the brilliant, life-enhancing principle of leaving out the martini.

You also get dramatic advances when you spot that you can leave out part of the problem. Algebra, for instance (and hence the whole of computer programming), derives from the realisation that you can leave out all the messy, intractable numbers. Then there's the new, improved UK directory enquiries service. A couple of years or so ago something radical changed: when you dialled 192 you actually got a civil, helpful answer, usually - and here was the clue - delivered in a Scots accent. The whole operation had been rounded up and moved to Aberdeen, where they had a plentiful supply of civil, helpful people who didn't have to be compensated for living in London. Somebody bright at British Telecom had spotted that the location was immaterial - the problem of distance could simply be left out of the model (something they have yet to comes to terms with in their pricing structures). With a little extra cable laying it seems to me that they could have moved UK Directory Enquiries to St Helena or the Falklands, thus bringing whole new possibilities of employment to areas that were previously limited to the things you could do with sheep. The Falklands could, while they were about it, put in a bid to run Argentina's directory enquiry service as well, which would give the Foreign Offices of both countries something to think about.

Almost everything to do with the net involves spotting the things we can now leave out of the problem, and location - or distance - is one of them. Wandering around the web is like living in a world in which every doorway is actually one of those science fiction devices which deposit you in a completely different part of the world when you walk through them. In fact it isn't like it, it is it. Trying to work out all the implications of this is as difficult as it was for early film makers to work out all the implications of being able to move the camera. What else is going to fall out of the model?

Over the last few years I've regularly been cornered by nervous publishers or broadcasters or journalists or film makers and asked about how I think computers will affect their various industries. For a long time most of them were desperately hoping for an answer that translated roughly into 'not very much'. ('People like the smell of books, they like popcorn, they like to see programmes at exactly the same moment as their neighbours, they like at least to have lots of articles that they've no interest in reading', etc.) But it's a hard question to answer because it's based on a faulty model. It's like trying to explain to the Amazon River, the Mississippi, the Congo and the Nile how the coming of the Atlantic Ocean will affect them. The first thing to understand is that river rules will no longer apply.

Since you're currently holding a magazine, let's think about what might happen when magazine publishing is no longer a river in its own right, but is just a current in the digital ocean. Magazines are starting to appear on the Web, but since they are just a number of interconnected pages in a world of interconnected pages, the boundaries between 'magazine' and 'not-magazine', or indeed between 'magazine A' and 'magazine B' are, from the Web browser's point of view, rather vague. Once we drop the idea of discretely bound and sold sheaves of glossily processed wood pulp from the model, what do we have left? Anything useful?

From the reader's point of view it's useful in much the same way that a paper magazine is: it's a concentration of the sort of stuff she's interested in, in a form that's easy to locate, with the added advantage that it will be able to point seamlessly at all kinds of related material in a way that a paper magazine cannot. All well and good.

But what about the magazine publisher? What does she have to sell? What's she going to do now that she doesn't have stacks of glossy paper that people are going to want to hand over wads of greenies to acquire? Well, it all depends on what sort of business you think she's in. Lots of people are not in the business you think they're in. Xerox, for instance, is in the business of selling toner cartridges. All that mucking about they do developing high-tech copying and printing machines is just creating a commodity market in toner cartridges, which is where their profit lies. Television companies are not in the business of delivering television programs to their audience, they're in the business of delivering audiences to their advertisers. (This is why the BBC has such a schizophrenic time - it's actually in a different business from all its competitors). And magazines are very similar: each actual sale across the newsagent's counter is partly an attempt to defray the ludicrous cost of manufacturing the damn thing but is also, more significantly, a very solid datum point. The full data set represents the size of the audience the publisher can deliver to its advertisers.

Now I regard magazine advertising as a big problem. I really hate it. It overwhelms the copy text, which is usually reduced to a dull, grey little stream trickling its way through enormous glaring billboard-like pages all of which are clamoring to draw your attention to stuff you don't want; and the first thing you have to do when you buy a new magazine is shake it over a bin in order to shed all the coupons, sachets, packets, CDs and free labrador puppies which make them as fat an unwieldy as a grandmother's scrapbook. And then, when you are interested in buying something, you can't find any information about it because it was in last month's issue which you've now thrown away. I bought a new camera last month, and bought loads of camera magazines just to find ads and reviews for the models I was interested in. So I resent about 99% of the advertising I see, but occasionally I want it enough to actually buy the stuff. There's a major mismatch - something is ripe to fall out of the model.

If you browse around an online magazine (HotWired, for instance, springs unbidden to mind) you will find a few discreet little sponsor icons here and there which you choose to click on. You only get to see the proper ad if you're actually interested in it, and that ad will then lead you directly towards solid, helpful information about the product. It is of course much more valuable for advertisers to reach one interested potential customer than it is to irritate the hell out of ninety-nine others. Furthermore, the advertiser gets astonishingly precise feedback. They will know exactly how many people have chosen to look at their ad and for how long, with the result that an unwelcome ad for something no one's interested in will quickly wither away, whereas one which catches people's attention will thrive. The advertisers pay the magazine for the opportunity to put links to their ads on popular pages of the magazine and - well, you see the way it works. It is, I am told by people with seriously raised eyebrows, astonishingly effective. The thing which drops out of the problem is the notion that advertising need be irritating and intrusive.

That's one model of how online magazines work and it is, of course, absolutely free to readers. There's another which will probably arrive as soon as it becomes possible to move virtual cash around the Internet, and that will involve readers being billed tiny amounts of money for the opportunity to read popular Web pages. Much less than you would, for instance, regularly spend on your normal newspapers and magazines because you wouldn't have to be paying for all the trees that have to be pulped, the vans that have to be fuelled and the marketing people whose job it is to tell you how brilliant they are. The reader's money goes straight to the writer, with a proportion to the publisher of the Web site, and all the wood can stay in the forests, the oil can stay in the ground, and all the marketing people can stay out of the Groucho and let decent folk get to the bar.

Why doesn't all the money go to the writer, I hear you (and indeed myself) asking. Well, maybe it will if he's happy just to drop his words into the digital ocean in the hope that someone out there will find it. But like any ocean, the digital one has streams and eddies and currents, and publishers will quickly have a role finding good material to draw into those currents where readers will naturally be streaming through looking for stuff, which is more or less what they do at the moment. The difference will lie in the responsiveness of the market, the speed with which those streams will shift and surge, and the way in which power and control will shift to those who are actually contributing something useful rather than just having lunch.

The thing we leave out of the model is, essentially, just a lot of dead wood.