Rupert Goodwins' Diary

Thursday 23/02/2006 Who's Asus? Make motherboards, don't they?

Thursday 23/02/2006

Who's Asus? Make motherboards, don't they? Or is it laptops? Yes — and monitors, mobile phones, PDAs, routers, wireless access points: it's a big company with a low profile. It's also keen to be seen as an innovator, which is why it's come up with a futuristic design study for a modular, environmentally appropriate PC. You put the modules on a special shelf; they have no traditional connections but pick up power by magnetic induction and communicate between themselves through high speed wireless. Looks snazzy, and you get to send the modules back to the maker when they've become obsolete — I guess you could even print the address and a stamp on the outside of each, so you could just pop them in the post.

But really! It's fun to think of building computers by clipping together modules like so much Lego, but the physics says no. Induction works, a little, but you can't get much power through it — and if you could, how would you get the heat out? And while lots of people have looked at wireless interconnectivity for chips and modules: it doesn't work. Or rather, there are lots of better ways of cooking that particular frog: optical looks good, as do contactless proximity systems, but you can't just lean modules against each other.

Then there's the traditional failure of any 'future proof' computer systems. You get tied into one particular manufacturer, who invariably doesn't have the resources to develop innovations as fast as the rest of the market — which is changing so fast you're quite likely to want to buy a complete new system by the time one particular bit is outdated anyway. And the rest of the market enjoys far greater economy of scale than your particular unique platform, so you quickly get to the situation where it costs more to upgrade one component than it would to chuck the lot and change — something particularly apparent to anyone who uses a video projector and is faced with a £300 bill for a new bulb that could buy a fairly tasty new TV in toto.

So the system as envisaged by Acer, even if it worked, would be slower, take more power, be less flexible and far more expensive than any comparable system built out of more mainstream components. Style can only get you so far. As for the environmental impact: a much more plausible model is for a very highly integrated thin client and very high bandwidth communications to a central server. Then there's almost nothing to distribute and retrieve — all of the heavy work goes on in central locations, which is much easier to manage and control.

Asus is dead wrong. But I'm glad it had a go — and hope it'll carry on. Like concept cars, the role of such experiments isn't to accurately predict the future even if the company can't resist the temptation to spin it thus. Instead, it's to make us think about things we take for granted, things that prevent us from even imagining there may be alternatives. That's the deadliest trap.