Future tense

As a major player in the growth of the web, Tim O'Reilly has a vision of how the internet will develop - and wants to tell everyone about it. Charles Arthur meets him

Tim O'Reilly is a busy man. Move any faster and his feet would not be touching the ground. As it is, he's taking a rapid dinner at a Heathrow hotel when we meet, having spent a day briefing companies including BT and Microsoft about the things that he thinks are important - the futures that are coming to meet them whether they like it or not.

O'Reilly is the head of the West Coast publishing outfit that bears his name (and no relation to the proprietor of this newspaper). But what he's really good at, in his own words, is being an "arbitrageur of the future". His favourite quote comes from the science fiction author William Gibson: "the future's already here: it just isn't evenly distributed." O'Reilly knows where it is. And through his familiarity with that, he's on first-name terms with most of the big names in the internet - he mentions in passing Jeff [Bezos, the founder of Amazon] and Steve [Case, who brought AOL to glory].

And it's true that a conversation with this man will leave you gasping for breath, with so many unusual takes on the future. This article could have been two or three times longer if I included everything that was discussed.

Why does a publisher follow this evangelical path? He grins. "My business always does well when we do what Jim Barksdale [of Netscape] said: 'Find a parade and get in front of it'. Our approach is that by helping people understand what's important they'll buy books about it to understand how to do it."

So, Amazon and Google. The key here is the mantra that "platforms beat applications". Think of all the great applications there have been - WordPerfect, Lotus 1-2-3, Netscape. How were they beaten? By Microsoft, which offered a platform that many developers could work with and produce for. An application is a closed thing unto itself. A platform offers all sorts of possibilities.

"Jeff [Bezos] heard me give a talk about this, where I said, 'If you don't go to a platform, Microsoft will kill you', and he made me come and give it again to Amazon's people. But I didn't want to scare them, so I left off the last bit about Microsoft killing them. They're really clear at Amazon that they have to be a platform; that's a key part of their thinking." The result is that you can get software hooks (APIs) that let you tap Amazon's database of CDs: rankings and prices and availability. You'd think that would all be downside for Amazon: more load on its servers, but for what?

"Not at all! They're making money hand over fist from their API. It has led to an explosion of people building interfaces to Amazon. I know one guy who has built a business worth something like $5m per year from a tool that builds associate sites [for people who direct business to Amazon]. The key is that every fourth transaction through each site gets his associate ID, so he gets the commission." Google too knows that Microsoft is gunning for it, and has developed APIs to let people pull data from it.

By contrast AOL (which helped make O'Reilly a millionaire: he built GNN, the first web portal, and AOL bought it early on and paid him in stock which he sold for about $20m - £12.5m) doesn't get it, he says. It owns MapQuest, which as its name suggests has huge amounts of map data, but it's just a web page. He's almost disconsolate at the dullness of it. He can see a vista of possibilities: location-based services for Web phones... GPS hookups... but no, AOL just has a website. Similarly, eBay is fiercely protective of its data, with no software hooks for external use. Not clever, he thinks - although it might be saved by its ownership of PayPal, which he thinks has the possibility of being so essential to web use that it could save its parent from itself.

Meanwhile, he perceives a huge change under way in hardware and software, towards the commoditisation of both. Hardware is now an easy choice: you get an x86 processor (it need not even be from Intel) in the volume you need and stick on what O'Reilly calls "Lamp" - the Linux operating system, the Apache web server, the MySQL database, and any (or all) of the scripting languages Perl, PHP and Python. What unites those software components is that they are free, and open-source. "That drives value up to another layer," he says. "There's still room for niches - Intel, Cisco can be very profitable. But the internet demonstrates the one successful way of competing against Microsoft: don't try to be proprietary. Build software and deliver the services over the net, and be profitable with those services." Again, Amazon and Google fit that description. Google is lots of cheap Intel processors wired together and, O'Reilly points out, already fits the description by companies such as HP and IBM of "on-demand computing" that is as available as electricity. "Google is on-demand searching," he says. "What more do they need to do?"

Lamp and hardware commoditisation will bring an era in which the availability of web services such as Amazon's and Google's will explode, he believes. And nothing much can stop that - except, he fears, a couple of things: digital rights management (DRM) and software patents. Both are means of interfering with inventiveness at the rate that computing needs to evolve; both get lawyers into discussions that should really be between the "alpha geeks", the ones who exploit the interesting new foibles of software and interfaces.

"DRM, as Larry Lessig says, is the past trying to interfere with the future, and it's terrible for a government to help it happen. But what that tells me is that the innovation will still come - but in countries which don't sign up to Wipo [the World Intellectual Property Organisation, which lays down decisions on infringements]. Innovation will happen anyway."

So he must like innovations such as Apple's online Music Store? Not at all. "Charging $1 per song is the most destructive thing for the music industry since Napster." How so? "There are obvious cost savings in this new medium that consumers expect. But they're not getting it in this pricing. We've taken out the manufacturing and distribution costs and it's still $10 per album?

"Only once the competition comes in will we find out what the price should be per track and per album. I suspect it's going to be a lot less. The music industry is going to have to adjust rapidly. At the moment it's not making itself better than being free."

But doesn't the web at least offer users the ability to pick and choose from a huge variety of sources like never before?

"I keep hearing the rhetoric that the Web will make everybody equal," he says. "But it just changes who the publishers are, not the need for publishers who chose from a big range and put out what is special. If you have millions of items to read through, you've got to have somebody to go through them."

Tim O'Reilly happens to be just that man.