By Siobhan Kennedy
Microsoft Corp’s defense case took its first serious knocking at the antitrust trial in Washington yesterday when the company’s chief technical expert was forced to concede inaccuracies in sworn videotaped evidence presented earlier this week. During cross-examination by the government’s lead attorney, Jim Allchin, Microsoft’s senior VP and technical guru admitted that videotaped evidence presented to the court on Monday did not tally with his own, personal test results, even though Redmond claimed the tapes were supposed to be an exact replica of Allchin’s work. Allchin’s entire testimony hinges on the fact that he claims to have carried out tests that prove the performance of Microsoft’s Windows 98 operating system software is degraded when the browsing function is suppressed using special IE removal software developed by the government expert, Edward Felten. Felten’s software is designed to show that IE can be removed from the core OS so that users can install an alternative browser if they wish. But Allchin’s contention is that uninstalling IE affects the performance of the entire OS, so users should not do it. But using a series of still screen shots taken from the two-hour long videotape which purport to mimic the expert’s tests, government attorney David Boies tried to show that Redmond’s engineers had not actually used Felten’s program at all and that it was in fact the way Windows itself works which slows the system down. As evidence, Boies showed that an un-Feltenized OS would have a window bar across the top of the screen that reads Microsoft Windows Update Microsoft Internet Explorer proving that IE had not yet been removed. Then, turning to the screen shot where Redmond claims to have used the Felten program, he drew the court’s attention to the same window bar. Instead of reading Microsoft Windows Update Microsoft Windows 98 (what a Feltenized OS should look like), the window bar clearly displayed the inscription characteristic of an un-Feltenized OS. Although Boies didn’t actually accuse Microsoft of falsifying evidence, the screenshots were enough to send a ripple of excitement and tension through the Washington courthouse. This video you brought in here and vouched for to the court and testified how much you checked, that’s just wrong isn’t it? asked Boies. To which Allchin nervously replied: They must have filmed the wrong system… I do not think the Felten program had been run but I don’t think that matters, I know that a performance problem exists. Boies bounced back: You know that it matters whether what you say here is right or wrong? You know that matters don’t you? After a long pause Allchin replied: What’s on the screen is the truth. But Boies refused to let up. How in the world could your people run this program, calling it the Felten program, when they didn’t test it? I am going to have to go back and find out what happened, a weary Allchin said, they probably filmed it several times and probably grabbed the wrong screen shot.
Allchin explained that one reason for the inaccuracy could be that the testing machines may or may not have had other applications loaded onto them, which could have had an adverse affect on the system and make it look like the engineers had not run the Felten program. But Boies jumped on his explanation immediately, reminding Allchin that Redmond had sworn the replica video tests were, like Allchin’s laboratory tests, carried out on what they referred to as virgin machines, or computers that only had Windows 98, and no other application software, installed. You testified under oath that the problems and you believed were shown were on virgin machines with just Windows 98. You believed that then. Boies said. It’s all relative to what you say a virgin machine is, Allchin replied, Whether or not [Microsoft] Office is loaded is irrelevant. Speaking at a press conference later that afternoon, a spokesperson for Microsoft told ComputerWire that, in hindsight, Redmond should have been more careful to ensure that the replica tests were carried out under the exact same conditions as Allchin’s original laboratory work. He admitted that the same machines were left sitting around for anyone to use as demonstration models, to test Office for example, and added that inevitably some such tampering had caused the discrepancy. But he dismissed the attorney’s examination as nit-picking over video production techniques. He added Microsoft had been on the telephone to officials in Redmond during the lunch hour and that they had confirmed that the Felten tests had indeed been run. However Allchin, during the redirect, admitted that there was still a question about the Window bar showing the wrong information. My conclusion is that is was a Feltenized machine and that there wasn’t any confusion about the tape being stopped and started. There is a question about the title bar, he said. Officials said a more detailed response could be expected in the next few days. Jim Allchin’s re-direct is expected to continue until lunch time tomorrow, after which Microsoft’s third witness, Michael Devlin, president of Rational Software, will take the stand. Devlin is expected to say that for Rational ISVs (independent software vendors) there is no meaningful way to distinguish between the IE APIs and any other Windows APIs. He will argue that it is in Rational’s interest, and that of other ISVs, to have API access to as much Windows functionality as possible.