A better backup can restore online freedoms

There is only one thing more boring than making backups, and that's finding that your backups don't work. A friend of impeccable discipline was recently rewarded for years of patient CD burning by discovering that of twenty discs covering a two year period, three — branded HP — were readable. The rest, a cut-price bulk brand, were riddled with errors. Furthermore, he had around 120GB of painstakingly saved data from his Windows 98 days in a format that could only be read by the selfsame operating system.

The savage truth about backups is that they are only guaranteed at the moment you've made a successful restore from them. Until you've done that you can't know that they're safe, and after that they begin on a path of decay with an unknown half-life. And that's assuming you've made a good backup in the first place; for such an essential task, the lack of decent basic tools in Windows is unforgivable.

One idea has the potential to solve all these problems. Imagine a disk drive on your desktop that was guaranteed never to forget, that couldn't crash or be lost in a disaster. Furthermore, imagine that disk drive could be transferred painlessly and instantly to any computer you owned, from anywhere in the world. Drag your documents into the drive, and they'll stay there until you want to retrieve them — and any time you need more room, you can just add more.

Such a magic device is feasible. Distributed backup, where groups of people store chunks of each other's data over a peer to peer network, is not a new idea, but in the age of broadband and cheap hard disks it is an eminently good idea. By encrypting your data, chopping it up into easily manipulated chunks and storing duplicates across the Net, you can be confident that when you need it, a copy will be available from somewhere.

The software can monitor the reliability of each location, moving stuff around as necessary if the total number of available duplicates falls below a safety trigger level. All you have to do is volunteer to set aside the same amount of local disk space as you plan on using remotely, and the sums all add up — if you want to store 1GB of data eight times over, then allocate 8GB of disk. I lack the maths to model what the most effective mix of duplication, chunk size, monitoring rates and so on would be, but there's no reason not to make the software self-tuning. And the best user interface is no user interface — if the system presents itself as just another storage device, then everyone can use it.

You'll end up with a backup system that is superior to any alternative for reasonable amounts of data, while being effectively free. Because the client software can follow any trends in hardware or software, the system should remain viable for as long as the Internet carries on working, silently taking advantage of increases in bandwidth and storage.

Such a proposal is not dissimilar to various anonymous peer-to-peer publishing systems such as Freenet, although the purpose is very different. And here lies the second great advantage of distributed back-up; it provides a very useful and entirely benign use of technologies that risk being outlawed. Freenet causes conniptions among our moral guardians, because it offers a way to anonymously publish lots of data; as part of its process, each user agrees to store encrypted data on their hard disk.

With Freenet, the users don't know what's in the data, nor do they know who provided it or where it's going, which makes it impossible to prosecute someone for wilfully hosting or distributing illicit material. One potential solution is being discussed whereby it would be illegal to have encrypted material on your system to which you had no key; the argument goes that if you do, you must be up to no good. It's an extension of the old "if you're innocent, you have nothing to hide" argument for police surveillance; morally and logically null though it is, we live in times where such ideas are routinely turned into law.

Yet the existence and widespread use of distributed backup would be a sovereign cure to such nonsense. It would demonstrate just how much potential for good exists if we're allowed to create and experiment with the tools we've developed, and not limited by moral panic or the wish of established interests to preserve the status quo. And my friend would no longer be stuck with a large collection of drinks coasters that once held his life's work.