Avalon: Microsoft's microchannel

A couple of years ago, I forecast that XML is to Microsoft as PC was to IBM. A new essay by Joel Spolsky has now fleshed out in marvelous detail How Microsoft Lost the API War, and it turns out that — to return to my analogy — Avalon is to Microsoft as microchannel was to IBM.

You'll understand what I mean if you've been closely involved with PCs for the best part of two decades, as I have. If you haven't then here's a quick recap extracted from my August 2002 item:

"When IBM saw the PC market slipping away from it, the company set to work to create a completely new, proprietary architecture called microchannel, which it launched in 1987 at the heart of the PS/2 family of next-generation PCs. The PS/2 flopped — much to the chagrin of IBM's most loyal enterprise customers, who obediently bought them by the truckload — handing market leadership on a platter to Compaq ... Today Microsoft, suddenly faced with the prospect that its championing of XML web services is about to cannabalize its almost total dominance of the desktop software market, is hard at work developing its own version of IBM's desperate microchannel ploy."

When I wrote that, it was in the context of Microsoft's plans for the latest version of Office. It seems that I got the wrong product, even though I'd correctly nailed the corporate psychology. As Joel explains in his essay with much more panache and erudition than I could ever muster, it's not Office and its XML schema but Avalon, the new graphics system at the heart of the next version of Windows that directly parallels IBM's forlorn microchannel experiment:

"Microsoft ... might have implemented things like Avalon, the new graphics system, as a series of DLLs that can run on any version of Windows and which could be bundled with applications that need them. There's no technical reason not to do this. But Microsoft needs to give you a reason to buy Longhorn, and what they're trying to pull off is a sea change, similar to the sea change that occurred when Windows replaced DOS."

Or, more likely, similar to the sea change that didn't occur when microchannel failed to displace the original PC architecture. A direct result of that failure was that Microsoft stopped backing OS/2, which was IBM's planned successor to DOS, and began working with IBM's hardware rivals to perfect Windows instead. So the sea change became a succession of cumulative events:

• The failure of microchannel fatally undermined IBM's reputation for invincibility
• Continued rapid growth in sales of PCs fueled demand for a next-generation operating system
• IBM's failure to deliver a functional version of OS/2 left the field open for Windows.

Joel's essay goes on to outline that the threat to Microsoft lies in the rapidly growing popularity of web-based applications, despite the less complete user functionality of the browser:

"... the Web user interface is about 80% there, and even without new web browsers we can probably get 95% there. This is Good Enough for most people and it's certainly good enough for developers, who have voted to develop almost every significant new application as a web application. Which means, suddenly, Microsoft's API doesn't matter so much. Web applications don't require Windows."

As Joel points out, web applications run on servers, and even if they're written (as many of them are) to run on Microsoft servers, they can also run "pretty well under Linux using Mono", so not only do they make Microsoft's dominance of the desktop redundant, they don't even guarantee it can retain a presence on the server. What's more, there are plenty of efforts currently under way to add rich-client functionality to the browser environment. This is a big, rapidly growing market that's crying out for a next-generation client system, but Microsoft has shut down any work on developing a better browser, and is instead gambling all on persuading the market to wait for it to deliver its shiny, new, Avalon-based Windows technology.

The demise of Windows would be one hell of a sea change, but Joel's essay is creating a lot of waves because he's made a very strong case that it could happen. I think Avalon sounds a lot like microchannel, and that makes Longhorn equivalent to the PS/2: marvelously engineered, but destined to be late, unpopular and ultimately one of the biggest mistakes in its creator's history.

Nevertheless IBM did recover from its microchannel episode, and even microchannel technology itself was ultimately assimilated into the PCI bus that remains at the heart of modern PC architecture.

In ten years time, when applications run in rich browser clients, Windows will have settled into its legacy platform niche, the much-derided Avalon will nevertheless have made important contributions to subsequent innovations, and Microsoft will be working hard to re-establish the market position of Office having relaunched it as a web-based application. The company will seem remarkably chastened compared to its standing today, but it will still be a force to be reckoned with.

posted by Phil Wainewright 11:11 AM (GMT) | comments | link