Original URL: http://www.theregister.co.uk/2011/01/11/windows_on_arm/
Windows on ARM: leading from the rear
Should you be excited yet?
Analysis Bill Gates may have long departed, but he's still in charge at Microsoft, in spirit at least. Every year Chairman Bill would return from a "reading week" – I prefer to think that Bill got lost on his annual trip to the shopping mall – having noticed that there were lots of newfangled TVs / watches / phones / pacemakers about, and none of them were running Microsoft software.
When he returned to HQ, the cry would go round that something must be done. The result was invariably a new OS project. This time, it's different – and Microsoft has decided to let someone else do the hard work. Microsoft will port full-blown Windows to ARM's processors. The OEMs therefore have the headache of providing driver support.
A non-Intellic Windows is not exactly earth-shattering news – Microsoft designed NT to be portable back in the days when it wanted people to buy it because it was a less anarchic and unruly version of Open Systems (aka Unix). Microsoft then spent much of the 1990s developing and supporting Windows on non-Intel hardware, just to prove how portable it was. The big vendors who signed up to a native non-Intel Windows and tried (with varying degrees of persistence) to make a business out of this included the biggest names in the industry: SGI, HP, IBM, and DEC (later Compaq).
DEC held out the longest, before chucking in the towel a decade ago. By then, potential customers had long since stopped caring about the portability of NT. It was a headache enough supporting two Intel versions of Windows, plus whatever really did the grunt work in the back office – a Unix box, or an old mini, or something even larger.
I find this announcement interesting for two bets that Microsoft is making about the market.
The first is that the WinARM handhelds will burgeon, so they more closely resemble the specifications of a laptop. They'll absolutely have to, Intel insisted in a most amusing press conference last week, or they'll stay unsold.
The other bet is that somehow people will prefer using the full heft of the Windows UI – cruft and all – on a portable device. These devices will be tablets and something vaguely called "cloudbooks" – which are cheap notebooks or netbooks by any other name.
I would say one of these punts is better than the other – but I really have my doubts about them both. You can spot the obvious common factor – it's what prompted the move.
"Gee," said the Redmond executives. "Apple has cut down its full-blown desktop OS and nobody noticed. Why can't we do the same?"
So let's examine the first assumption: that people need hods of computing muscle, only in a form factor that's not really a laptop. We've been round this block before – it's really the Ghost of Chairman Gates again, insisting that Tablet PCs are where the market is heading. Unveiling the Tablet in 2001, Gates insisted it would become "the most popular form of PC within five years." They were expensive, and brought all the inconvenience of a laptop to something that (at times) pretended not to be a laptop. No wonder that found such a very small niche.
Intel insists that high-power chips will be needed because we all do "content creation" on the move. Again, wouldn't a successful Tablet PC niche have proved that there's a demand for that? Wouldn't the iPad, which is primarily a content viewer, have flopped for the same reason?
(Having used Windows 7 on an old-ish Atom netbook recently – albeit one with 2GB of RAM – I can't quite see what the fuss is about. It works pretty well.)
Bespoke UI, or redesign of Win?
The UI issue is the bigger uncertainty. You'll note that while Apple retained much of the Mac OS X API for its iGadgets, it developed a completely new shell and UI framework. Microsoft hasn't said whether it too will develop a new shell and programming API specifically for these ARM mobile devices, or whether it will go for compatibility, and allow developers to port their apps more easily – perhaps kludging the UI elements – so you get the "old" style Win UI but with bigger buttons, for example. The lack of a definitive statement may be because Microsoft can't answer the first question – what we'll do with them – with any certainty.
Without doubt, Microsoft is much better at designing UIs than it was five or 10 years ago – competition from Apple and the Vista shambles has forced it to become more focused. Its Media Center and Windows Phone 7 are really quite well done. Microsoft appears to be working on something called Jupiter, which, according to Mary Jo Foley, appears to fit the bill.
It's a framework on top of Windows for "immersive apps", using an easy-to-learn declarative language XAML. But Jupiter appears to be some way from completion – and worse, since it's locked into the Intel Windows release schedule, it might be as much as two years away.
Apple's design approach is to go great lengths to disguise the fact that the iGadget is not a computer – it emphasises reliability and simplicity over sophistication and features. The iPad turns on and off reliably, and it does what it says on the tin. It doesn't care if it's a "cloud device" or not, in fact, you can't even get to the local file system without breaking the warranty. Yet the UI is a far more significant factor than the chip architecture or compatibility. That's why last week's news is a bit underwhelming.
Last week Microsoft should have led its announcements with the UI, and let everyone know that it has decoupled it from the Windows release schedule. And then let everyone know that it's not cutting the dependency on Intel's mobile chips. Where would the ARM be in that? ®
Bootnote: Note that I've tried very hard not to echo the most over-used phrase of the week: "full-blown" – as in "full-blown" Windows. This is because the phrase originates with the bloom of roses, and describes the ripeness of the flower. Whether Windows is ripe is one question, whether it's a rose is another. Safer avoided, we reckon.