Halloween Document VI
The Fatal Anniversary
It is now just about exactly a year since the Halloween Documents were first released upon an unsuspecting world. Many things have changed; Linux 2.2 has gone from promise to memory, the Halloween Documents' author has quit Microsoft to go to work for a Linux-based startup, mainstream market-research outfits like IDC now predict a 30% server-market share for Linux in the near future, and the first Linux IPO rocketed Red Hat Software to a six-billion-dollar market capitalization. A few things have remained the same; Windows is still buggy and insecure and crash-prone, Windows 2000 still isn't shipping, and Microsoft is still making excuses.
But perhaps the most dramatic development in the Halloween saga has been a change in the axis of Microsoft spin. Ed "Sheriff of Nottingham" Muth has quietly disappeared from public view, perhaps having publicly tickled his tonsils with his toenails once too often for King Billy's tolerance. Instead, Microsoft has been trying to sandbag Linux with supposedly "objective" studies by third parties that turn out to have been bought and paid for by the boys in Redmond.
Fortunately, these tactics have been pursued with the same slap-happy level of incompetence that made Sheriff Ed's antics so amusing. The Mindcraft fiasco in March set a pattern continued by its sequels; Microsoft got the benchmark results it wanted only to be embarrassed when it came out that Mindcraft had apparently run them on Microsoft-supplied machines, at a Microsoft site, with the benevolent assistance of Microsoft technicians tuning both Windows and (even more helpfully) Linux and then neglected to mention in its press release that Microsoft had paid for and hosted the whole exercise. Mindcraft's credibility was, of course, utterly destroyed.
Mindcraft was just small fry, though one more tiny company who learned too late that when Microsoft brings you flowers, they're likely to end up decorating your grave one way or another. As the fatal anniversary of Halloween I loomed, Microsoft appears to have rerun the same scam on a much larger scale. This time, its date for the dance was a respected name in IT forecasting, the Gartner Group.
Cue the ominous background music ...
Sometime before 6 October, the Gartner Group published on its central corporate website, www.gartner.com, a series of five reports slamming Linux and predicting that its appeal would fade once the inevitable Service Pack 1 for Windows 2000 came out. These reports quickly spawned Linux-is-doomed articles like this example from 15 Oct on the IDG Australia website, which promoted them as objective studies by independent Gartner.
The articles, however, included the following small print at the end:
suggesting (though not proving) that this "Gartner report" was actually written and published by Microsoft on Gartner's own website.
We can't tell you the exact date of the reports, because they aren't there any more. On 19 October Gartner changed the copyright on the reports to no longer mention Microsoft, while publicly insisting that the research had not been funded by Microsoft. Tellingly, however, the URLs of the reports still began with this prefix:
and the same copyright notice, with the same attribution to Microsoft, can still be found on articles in the webletter/microsoft/ directory as this is written. Sometime between 19 and 27 October yanked the reports off its site entirely.
Fortunately, Rick Moen, a well-known Linux activist, managed to recover them from his Netscape cache and found the copyright notice reproduced above. In view of Gartner's protestations that Microsoft had nothing to do with funding the research, the next lines in the notice are especially interesting:
So. At first blush, these "Gartner Group" reports seem to have been a warm personal gesture from Microsoft Corporation to itself. That theory would fit both the copyright and the contents, which exactly repeats Microsoft marketing cant that we've seen before including such trademarks of the genre as the (unsupported and, according to IDC, false) assertion that Linux is taking seats mainly from other Unix shops.
However, the Gartner Group insists that Microsoft did not write the reports. A spokesman explained through the same compliant IDG Australia reporter responsible for the 15 Oct Linux-is-doomed story. Chase the link to get a load of this spin-doctoring; no paraphrase by us could do justice to the way Microsoft and the reporter deftly weave a fog of soothing confusion around the contradictions in their story. We're supposed to believe that:
Rick's comment on their explanation is more eloquent than my own:
Quite. Far be it from us to suggest that the Gartner Group has fraudulently colluded with Microsoft, however suggestive the evidence. Could it all have been some horrible mistake? Could Gartner Group have been innocently victimized by a Microsoft partisan in its own ranks? Perhaps one of these scenarios is why the reports quietly disappeared but not before taking much of the Gartner Group's credibility with them, alas. Poor Gartner Group; the folks at Mindcraft, I'm sure, would weep for them would deeply sympathize.
Ah well. Whatever occurred, I'm sure the large amounts of money that Gartner admits to having received from Microsoft before and after this incident have done much to soothe their upset at looking like patsies.
While the Gartner Group was busily issuing denials, Microsoft was publishing yet another attack document, the Linux Myths page. The Myths page has been well debunked elsewhere, but its technical weight is less relevant here than the fact that it cited the Gartner Group reports as a supporting third-party analysis. The references are still there, even though the reports are gone. Perhaps this was the point of the exercise for whoever actually wrote the reports?
And while all this was going on, a Linux-vs.-NT study that never had a Microsoft copyright on it was being done by a firm that doesn't take a retainer from Microsoft; Bloor Research. The results in their 19 October announcement are consistent with the results of many other third-party studies such as John Kirch's Microsoft Windows NT Server 4.0 versus UNIX. They said "The verdict is clearly in favour of Linux."
(First published 31 Oct 1999. Some material in this article was written by Rick Moen and is used with his permission.)
Eric S. Raymond <firstname.lastname@example.org>
Table of Contents
Before emailing or phoning me with a question about these documents, please read the Halloween Documents Frequently-Asked Questions.
Links to press coverage