Wednesday, 3 November 2010

Rethinking the Desktop

WARNING: This post is extremely geeky and most like boring. I've only written it here because it's the best place I could think to show it! Don't read if operating heavy machinery...

Recently I've been reading a lot of (geeky) stuff about the state of our operating systems. In the first ( KellaByte (not her real name... at least I don't think so) talks about how boring the Desktop OS is these days and how incompatible it is with touch. The second (, a response to the Apple "Back to the Mac" event, speaking about the pains of backward compatibility.

The problem is, we're still using the interface paradigm set out by PARC back in the early seventies. We're still using the basic hardware ideas as set out in the 70s/80s. What happens when you ignore everything that went before and start all over again?

GPUs can perform some calculations considerably faster and better than an x86 processor - people are just starting to get to grips with using them for non-graphic processing. What happens if you build a processor with 4 gpu cores and 4 ARM-based cores, for instance.... what could that machine do?

Could we redesign the bus on a motherboard, could we do it better? What about memory? Could we use mixed memory types to squeeze effectively 32GB of RAM into the space taken by a single stick of 2GB? If we were to speed up the bus and the primary storage, would we still need as much RAM at all?

So, brand new hardware... what about the OS? These days, it needs to be able to fit on everything from a tablet/netbook all the way up to a 24" widescreen monitor. It needs to be built for touch, but still be mouse/keyboard compatible. It needs to be clean, simple and light (on resources). It needs to be able to be simple to use for the average user and for tablets/netbooks, but it needs to be able to knuckle down for high-end stuff (developing, multimedia etc).

The way to go is probably the Linux route - 1 underbelly with a slightly different Desktop Manager for each of the form factors. You have SomeOS Lite for the tablet/netbook/novice user, SomeOS Mobile for the upper-end of the Laptop market and SomeOS Max for the high-end desktop user.

In terms of user experience, you have an App Store - including third party pay-for apps, community developed free apps and OS updates (like an overlap of a mobile app store and the ubuntu package repositories). That way you can fairly quickly have a large collection of applications available for your new OS, despite the fact that nothing ever created before will work without some kind of emulation.

I wish I had the money to put together a team of experts and create something totally brand new. I think we'd give Apple a run for their money...

No comments: