% fortune -ae paul murphy

MacTel: "the real story"

As I discussed yesterday, Robert Cringley has spent a lot of PBS Pulpit time trying to find the gold in Apple's MacTel decision. Now, personally, I rather think people, even people like Steve Jobs, make mistakes - but, lets consider an alternative theory.

This one starts with Microsoft - because Microsoft's biggest problem isn't Apple, it's Microsoft. It took them six calendar years and probably more than 35,000 man years to produce a new point release for iVMS - and along the way they announced and then abandoned a whole new kernel, a new network architecture, new security methods, and hundreds of interesting technical ideas.

Look at Microsoft's Itanium support and you see another side of the same organisational problem: they're about eight years into programming for VLIW hardware and have yet to produce anything more than some fairly basic, largely hand coded, OS and application ports that almost totally fail to make use of the hardware - in effect, Microsoft has been doing to Itanium what it did to Alpha, SPARC, and PPC: producing inferior performance by treating the machine as a kind of inferior x86.

Ask yourself - if you, or a good programmer you know, had the SQL-Server source, and Intel's Itanium compiler suite, could you do worse in a week than Microsoft has in eight years?

I don't think so - and the reason is that Microsoft is absolutely hobbled by backwards compatibility; and not so much in their code, as in their culture. Basically they're crippled by their own internal processes: having successfully evolved their development machine into an x86 ghetto by combining hiring and promotion selectivity with organisational structures that work against innovation.

If that's the anvil, the sledge -Intel's inability to advance the x86 architecture - is coming down fast. Right now they're significantly behind SPARC and PPC on both performance and power use - and that's only going to get worse.

So what to do?

Microsoft converted to the PPC architecture for the X360 to provide itself with a logical continuation strategy: sell it into homes as an entertainment hub, bring the major personal applications to the X360, meet the resulting demand for comparable performance at the office with a new line of Microsoft PPC based PC gear, and then phase out support for the remaining x86 customers over a nice long period like 15 or 20 years so Dell, HP, and Lenova can't move in time to threaten Microsoft's takeover of their hardware business.

As strategies go, this is very cool stuff - and doable except for Microsoft's proven inability to provide a new OS or make use of PPC facilities like the Altivec in the new software. Right now, for example, the X360 games inventory is less than exciting, and largely for the same reason the Itanium failed: great hardware, crippled, and crippling, software.

So how to fix that?

Well, suppose Microsoft recognises the obvious: the X360 is the fastest Macintosh money can buy - good graphics, six concurrent threads, three cores, Altivec onboard, 3.2Ghz - and just buys a MacOS X co-development deal for it?

At perhaps $10 billion in cash and Microsoft shares it's a bargain - because it gives Microsoft's management what they so desperately need: a way to bypass their own organisational inertia to get a proven, network oriented, OS for PPC; an installed base; significant new application sales; and, a network of smart, committed, PPC developers.

Now if Apple had stayed with PPC a deal like this would just turn Apple into a Dell/HP variant and you'd expect the Apple board to oppose the licensing for that reason. With MacTel, however, an Apple decision to license "Leopard" to Microsoft for co-development and deployment only on PPC just uses Microsoft's money to extend Apple's short term market, and Apple's board couldn't possibly turn it down without triggering significant shareholder action.

Microsoft, of course, would really see something else here: to them MacTel would be just another lane on its bridge to the future because short term MacTel sales increases make it easier to sell home use applications for MacOS X on X360 and ultimately to migrate office use to MacOS X on PPC.

And what about the dominoes? Apple either goes SPARC or dies with x86; Intel becomes a Microsoft assembler; HP takes it in the neck, and the IBM/Sony/Toshiba partnership takes the Asian market with Cell but is trapped into selling Linux against MacOS X in other markets.

As in: come on, Bob - join me in saying "Merry Christmas!" to Apple shareholders and admit this idea is just a no brainer.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.