Monday, January 26, 2004

Telirati Newsletter #38

This isn't one for the ages, but it is an example of giving kudos where due. Microsoft often set the standard for handling problems. It will be interesting to see if content protection and other priorities can be accomodated without negatively affecting that customer focus.

Telirati Newsletter #38: How it’s done

How it’s done: When Microsoft got caught in a privacy problem, using unique ID numbers generated on the fly (these are the backbone of every software service that has to identify things across networks) to identify Windows customers across sessions with Microsoft Web sites, and for other purposes, Microsoft immediately issued a statement to the effect: “We were wrong, and here’s a fix to get rid of it.” Immediately, as in hours after the first press reports. On the weekend.

How it’s not done: Intel could have just stopped programming unique IDs into some Pentuim III CPU chips and allowed customers to make up their minds whether to buy chips with or without the numbers. Instead, Intel shot another hole in their foot claiming that ID numbers found in some Pentium II chips were a production error. That this is a breathtakingly unbelievable excuse is the kindest thing I can think to say about it.

How it’s done: Microsoft has released an internal memo, widely leaked to the press, explaining in clear terms that having one’s witnesses savaged by an extremely talented litigator is not the same thing as being on the wrong side of the law. While a strategy of being obstinate bastards is probably the right one for Microsoft, it does open witnesses to ridicule. Information explaining why such cross examination virtuoso turns don’t amount to much is the right antidote.

How it’s not done: While Microsoft, in it’s memo, points out that cases are decided on law, not PR, Microsoft have not done well against the government which wants the case decided on PR, and not law, or at least not a conventional reading of the law. By fumbling the ball in the creation of illustrative videos on the one hand, and by not consistently painting the government’s case as novel, radical, and anti-capitalist on the other hand, Microsoft has allowed the press to lazily focus on it’s executives getting reamed over relatively minor points, peripheral to the core issues.

How it’s done: A clock, built in 1746, housed in the museum at Versailles, that indicates the day, month, year, phase of the moon, and position of the planets, will tick over into the year 2000 without fault. Perhaps the maker knew that the new millennium begins in 2001, a number that should cause no difficulty to a competent clockmaker. Perhaps people thought differently before they were so artificially influenced by digital computers to think that first things are numbered zero. Is your firstborn child number 0 in your family? When you count a dozen eggs, do you start at zero?

How it’s not done: Intel, in containing the damage from the public reaction to processor IDs, has described the public as confused over the difference between privacy and anonymity. This and other attempts to “educate” the public in this matter are doomed to fail. There is nothing to educate about: either you want a processor serial number, or you don’t. All this, however, does not get at the underlying question: Do you want stuff in your computer that you don’t know about, and that cannot be subjected to independent review to determine if it causes security or privacy problems? It’s a simple question.

How it’s done: Sun, SGI, and now Apple are trying, with varying degrees of success at this point, to incorporate the “open software” model into their licensing regimes. This is more than just a fad. It is obviously useful to ISVs who have to struggle with incomplete documentation, and it is critical to creating and maintaining confidence in precisely the place that Intel has squandered confidence. Customers want to know what they are getting, at least to the extent that they have confidence that others who may be more interested in security will review open source systems and publicly report problems. It is a great shame that the continuing DoJ litigation against Microsft has effectively blocked Microsoft from making dramatic changes to its licensing models, since drama must be reserved for the settlement.

How it’s not done: They’re from the government, and they’re here to help. The new Cyber Citizen initiative of the US Department of Justice includes “personnel exchanges” between government and private industry to help thwart cyber-crime. Ick. How do you spell “warrant-less search?” If you want a heads up on the next pool of toxic public relations poison, here it is. Stay far, far away.

How it’s done: Microsoft has partially admitted to itself that there are more market segments than there are Windows operating systems to fill them, and has backed off a probably doomed effort at pounding a big, sharp-edged square peg called NT into the round consumer hole.

How it’s not done: Public statements by Jim Allchin and Bill Gates indicate that they are still true believers in NT for everything. Make no mistake, NT is a great product, and the next version will be an enormous step forward. Further, the NT development organization has successfully organized development efforts in several significant areas, among which networking and browser development, and forged them into a compartmentalized, yet coherent whole. No small accomplishment. But this accomplishment should not obscure the fact that it appears that Microsoft fails to grasp its own success with Windows 95. It was a magical thing to create a product so thoroughly accepted by individual consumers making individual choices. NT does not capture that magic. And, now that Microsoft wants new versions of Windows 98, there is, in effect, no Windows 98 team in place. The way way OS development is organized will serve Microsoft well in that large modules, like the browser, Universal Plug & Play, TAPI, and Multimedia enhancements can be rolled together into a major update, but the opportunity for a major kernel upgrade will probably have to be let go. A shame, since focusing on the x86 platform, and on a personal OS, to the explicit exclusion of NT’s portability and security, could yield significant performance advantages. The question: What will Microsoft learn from this?

Copyright 1999 Zigurd Mednieks. May be reproduced and redistributed with attribution and this notice intact.