Herb Caen once wrote that the sprawling night fog that fills the San Francisco Bay could be precisely mapped in the mind of the listener from the sound of a dozen foghorns, each one of which was tuned to a different pitch. You can build up a similar picture of the mind of Bill Gates from his epic written deposition. For here, in no less than 43,000 words is Mappi Bill. Or 464 legal points, many of which are an anxious toot or blast each tuned to a slightly differing pitch of paranoia.
Some you've heard before, and some you haven't - such as the threat to withdraw Windows from market within six months [par 183], or the reclassification of Embedded Windows XP as "a set of tools", not an operating system. Or the claim that Internet Explorer browser development costs Microsoft $100 million a year[par 234]. The latter we calculate as enough to employ over 1,000 engineers at current market salaries.
Anyone who's watched IE innovation slow to a crawl since 1999 - since when we've had the addition of a print preview feature, and little else of note - may well ask what they're all doing. Much as we love the 'scrapbook' feature of IE for the Macintosh, it doesn't add up to three thousand man years' worth of innovation. But we digress.
Perhaps the most extraordinary part of his written performance is its context. Remember: this is the plea for clemency from a convicted criminal. Microsoft was found guilty two years ago, and large parts of the conviction were upheld on appeal. The current hearings are being held to resolve the punishment.
Now when murderers appeal for their sentences to be commuted, it's usually on the grounds of diminished responsibility. They don't argue that the sentence should be rejected because it deprives them of patronizing their favorite eateries. Perhaps our take on the Judaic-Christian tradition of justice is a little naive, but we've always kinda assumed that when you've been found to have done wrong, the punishment is supposed to hurt. Murderers - either civic or corporate - don't get to choose their own sentence.
Although the inkies this morning paint Gates' testimony as a return to sanity and coherence, his regal, novella-length, deposition suggests otherwise. Those 43,000 words explain how badly the proposed remedy will harm the convicted company. Which rather misses the point, and suggests that his grasp on reality - and contempt for the judicial system - is just as shaky as it was in his video deposition first time round, when he quibbled over the meaning of the word "the", and, when challenged with authoring a particularly incriminating memo, bleated that "the computer wrote it!"
In fact Gates inadvertently, and surely unconsciously, suggests a picture of thriving innovation in a post-settlement Windows world. One where OEMs, spurred by the freedom to differentiate their OS offerings, add their own innovations, the winners and losers determined by the Darwinian competition that free markets are supposed to stage.
Of course Bill doesn't quite see it like that.
He also admits - and again it's subliminal - that clones of Windows such as KDE's Desktop remove the incentive for Microsoft to profit from ubiquitous elements such as the "Start" button. Now imagine if a company, let's call it HighwayCorp., owned a monopoly on using the word "STOP' on signs at road intersections, forcing competitors to devise ingenious ways to portray the same message. We'd fairly soon decide that 'STOP' wasn't the sole property of any single company, and force HighwayCorp to ensure its signs compete on other merits, such as the reflectiveness of the paint, or their durability against attack by trigger-happy NRA members. (This metaphor isn't completely accidental, as attentive readers will know).
Commodification has always been the enemy of Microsoft, and many of those four hundred legal points convey the same fear. Here's the most emblematic.
Feel Bill's pain
217... if Microsoft were obligated to allow ISVs to clone all the functions of all the "Microsoft Middleware Products" in Windows, Microsoft's ability to improve Windows would be hampered because the interfaces between modules would necessarily be "frozen" so that third parties could write to them. Given the large number of "Microsoft Middleware Products" in Windows under Section 22.x, the effect would be to freeze large parts of Windows.
As John Lettice points out in his summary of Bill's live testimony, Windows clones such as Sun's WABI truly did frighten the bejesus out of Microsoft. They always have: in 1993 Brad Silverberg explained to Andrew "Undocumented DOS" Schulman that Microsoft must keep changing the APIs to prevent Windows being cloned and commodified.
But the argument falls down when you remember the noble software tradition of backward compatibility. Microsoft itself is as aware of this as anyone else, carrying the burden of years of legacy code: it's careful to extend API functions with extra parameters, ensuring that they fail gracefully if the most modern library isn't present underneath. Microsoft wisely prefers this approach to obsoleting whole regiments of API functions, and in fact hasn't carried out a serious purge since the move to Win32 almost a decade ago.
Bill isn't finished, though.
272. Over the long-term, modifications to Windows by individual OEMs acting in their short-term self interest would present a classic tragedy of the commons problem. Just as a lake that is fished too heavily soon will support no one, the PC ecosystem as a whole will suffer if the stability and consistency of Windows is not maintained...
If you're looking for the most startling example of doublethink, look no further. It's a godsend, in that it confirms Gates innate fear of competition, and suspicion of free and open markets. For the Tragedy of the Commons is the classic knock-down-Ginger that trumps many classical libertarian approaches to supply and demand, as Hayek acknowledged.
But what commons, and what ecosystem? Ask any venture capitalist who's been asked for funding for a PC software startup in the past decade. In the name of "integration", most opportunities for thriving utility markets have been extinguished by Microsoft's desire to have a piece of the pay. Usually at the expense of the "competition". The biggest overgrazer on these commons has been Microsoft itself.
Which brings us to the nub of the case: the lofty, imperious tone of Gates' deposition - he really believes he single-handedly brought about a horizontal computer industry, saving the world in the process - and the contempt and fear in which Microsoft is held by almost everyone else in the business. This can be rationalized by entertaining the idea that while Microsoft is conceptually A Good Thing, the actual implementation is awful.
Gary Kildall's Digital Research had the idea first, was once in pole position, and on the evidence of some rather arcane and very old technical documentation we saw recently, was very much planning for the 1990s at the moment Gates thought he was buying a clone (a clone, mark you) of DRI's CP/M operating system. Forget the whimsical folklore about flying planes, Digital Research was simply considered too powerful at the time to make star billing when IBM launched its PC, and was handed an effectively worthless bit-part as the expensive "third choice" when the PC launched in 1981. You could hardly buy it, for love nor money. But that's another story.
Therapists typically base the nuttiness of a patient on the strength of their convictions, on which basis this 43,000 word opus alone stands as a kind of testament to Bill's madness.
But in the context of an enforceable remedy, and in pointing out ambiguities in the prosecution's arguments, the Chairman actually makes a reasonable legal if not morally-justifiable case. The weakness of the "middleware" remedy we'll explore in our next instalment, dear readers. ®