WIRED:Dominance by a handful of services poses a threat to the internet's resilience and flexibility

LAST WEEK I spent an afternoon with a group of writers and thinkers. Like most sensible people, they were neither technophobic nor technophiliac – they used the latest gadgets and software when it was useful to them and not a moment before. But many of them had brought laptops, and the conversation turned to the internet tools they used.

They talked about the social niceties of Facebook; using Google to monitor critical reviews of their work; which one of them Twittered and how they sometimes had to defend Twitter from cynics; the writers who used LiveJournal to stay in touch with their fans; how they would upload pictures of the event to Flickr afterwards.

What struck me, talking to these tech users (as opposed to workers in the tech industry itself), was how filled their conversation was with trademarks and company names. Ten years ago, the talk would have been the same in many ways but far more about generic technologies: e-mail and the web, perhaps IM and even Usenet.

Somehow our use of the net has narrowed into becoming the customers of a few companies, and the specific tools they offer.

Those who use one product are frequently locked out of using another company’s features and there’s plenty of social pressure for each of us to walk lockstep with our friends when we choose what social software to use.

There’s a reason why, in Ireland, Bebo and Facebook gained the upper hand, and it’s not just about who has the better software. It’s about who has already captured your friends.

Is this a bad thing, to move from open protocols like e-mail, to closed websites like Twitter? It’s not as bad as it could be.

All of these products have gained much of their success by being more open than their competitors. Flickr and Picasa have no problem granting other companies and independent developers access to the codes and commands that let them automate or enhance the photo-uploading experience; Facebook has whole ecosystems of smaller firms providing web applications, from Zombie Hunts to music-sharing applications.

If these household names have triumphed, it’s because they have gone a long way to imitating the interoperability of the original standards. But is that enough? E-mail, the web, Usenet and the internet itself were all designed to be radically decentralised. There are very few ways you could disrupt or control any of them by simply taking over one part of the net.

That’s not true of these single companies. Their owners want to be in control of their product. If they decentralised their service, there wouldn’t be a firm or a website. There would just be an unowned, universally used, free system like the internet: a place where one makes money not a thing that one can make money from.

And that’s what makes this transition so worrying. To give an example: last week, Amazon was accused of removing gay and lesbian literature from its sales rankings and search listings.

You could still buy the books, you just couldn’t find them. Amazon says it was a mistake, and that the mistake applied to more than just gay books. But, for a while, a whole class of writing was erased from one of the world’s largest catalogues of books.

The omission seriously damaged Amazon’s reputation, and Amazon has strong business reasons not to indulge in such clumsy censorship, deliberately or not, again. But what of other, more pernicious acts of control? In China, a Google search for Tianamen Square returns only positive stories and images, with the 1989 protests erased at the request of the Chinese government.

Twitter, Facebook and LiveJournal store not just public comments, but private confessions. Revealing that data to third parties would damage their reputation, but what if governments required it to be handed over under conditions of absolute secrecy?

Even without such conspiracies, the centralised nature of the new net has other problems. They claimed that a nuclear bomb could not break the decentralised internet. But just one bankruptcy or hostile takeover or even strategic decision by any of these companies could destroy millions of people’s access to their own data overnight. A catastrophic hard-drive failure or a disgruntled employee could do the same.

These problems have not gone unnoticed. Public-spirited coders and companies are already working on the standards and software that can decentralise some of these services. Software called Laconica decentralises Twitters’ micro-blogging model. Coders work to extract data from LiveJournal and Facebook so it might be kept safely elsewhere.

An internet pockmarked with trademarks isn’t by itself a problem. Private companies have contributed much of the innovation of the net, but a centralised internet poses a greater danger. Unless alternatives to these services arise, we risk sacrificing the flexibility and resilience of the net to another kind of infrastructure: one of walled gardens and central control, with neither independence nor true privacy for its users.