Expert and MS argue over definition of OS
Ah, the old ones are the best ones...
Some interesting subtleties have come to light in the transcript of the testimony by Professor David Farber of the University of Pennsylvania, who was allowed to interrupt testimony from James Gosling because he was about to go on a lecture tour. It was claimed by Microsoft that his examination would take one day, but Judge Jackson suggested that was optimistic, as indeed proved to be the case. Farber's written testimony is a general account of the software development process; the relationship between the operating systems and applications software; the inefficiencies in designing "so-called" operating systems which include inappropriate functions, such as applications like Web browsers; and the negative consequences of permitting Microsoft to add what are now applications to create an ever-larger, monolithic software package with Microsoft calls its "operating system". Farber presented a bare 12 pages of analysis and a 19-page CV that showed he had great experience with operating systems development and the Internet, but not with Windows. Since most of the first day of his examination was taken up enquiring about his knowledge of the Windows 98 operating system, he did not prove to be a very useful witness. It transpired that he had been recruited as a witness through a personal friend at the DoJ, and not on merit for his knowledge of the relevant matters for the case. Nor did his current practical knowledge of software seem very sophisticated. It would have been better to have had somebody who was familiar with Windows to testify. Farber is also an active Democrat, it was revealed in court, but the significance of this McCarthy-like question was not immediately clear - except that Judge Jackson is a Reagan appointee and probably therefore a Republican. Farber said he had declined opportunities to look at Windows 98 source code because he found the required non-disclosure agreement conflicted with his teaching duties, so far as revelation was concerned. Although Farber was able to give an authoritative view on the design of operating systems in general, it was an easy matter for Steven Holley, cross-examining for Microsoft, to make him look like a fool for his lack of knowledge about Windows. But if Farber was made to look the fool, Holley acted the fool by introducing textbook references that were grossly out of date ("ancient history" Farber said). One book was from 1990, a second was out of print, and a third was in its fifth edition in 1997, although it was criticised in amazon.com reviews for being out-of-date. The clearly highly-selective references were used by Holley in an attempt to show that what Microsoft was doing with the integration of IE and Windows was approved by theorists from a generous interpretation of the texts quoted. Farber had no specific knowledge of any of these books (presumably nobody had paid him $300/hour to read them), which did tend to suggest that they were not mainstream. At one point Farber suggested that the (Netscape) author of an email that said that "a kernel cannot sensibly be regarded as an operating system" should go back to school. Farber tried to establish the difference between an operating system and an operating system environment, but had only limited success. It was clear however that using definitions to define an operating system would not effective, and that Microsoft was correct in defining Windows 98 as an operating system, even if it was badly designed, faulty and bug-ridden. Notions of the definition of an operating system from the earliest days of Unix are just not relevant today. What is important is Microsoft's direct actions in making life difficult for a competitor, but these were not addressed in the first day of Farber's examination. Microsoft should be castigated for the implied claim that it was Microsoft that made TCP/IP available "free of charge" when it previously cost $50 in one commercial packaging. It was of course readily available for downloading free of charge. Farber was scornful of DLLs that were carried along as baggage in the operating system and were only used for IE, suggesting that "dead bodies hanging around code that's not used by somebody eventually causes you trouble". Another subject of Farber's scorn was OS/2 version 4 ("I have had the misfortune to use it"). Holley kept repeating the futile exercise of asking Farber if he knew the function of a particular file in Windows 98 until he was stopped by Judge Jackson and told he had made the point. The examples used by Holley were of course of DLLs that were used in IE and elsewhere in Windows 98. What was not clarified is which DLLs are used just by IE. The judge asked who would be the counterpart expert to Farber, and was told Jim Allchin. It will be easy for Allchin to say that he does not know the answer to this important question about DLLs, because he did not have hands-on responsibility for the design of Windows 98. Perhaps some hacker will make known exactly which files are used just by IE only, in time for Allchin's cross-examination. Nevertheless, the issue as to whether IE should be integrated into Windows is really a matter for Microsoft, but any steps that Microsoft has taken to make it difficult for other browsers to function with Windows would be anticompetitive and therefore illegal. Perhaps David Boies for the DoJ will take this up in his redirect. Farber had reservations about the Court of Appeals ruling that talked of it being "absurdly inefficient" to make end users reprogram their products to get additional features, but was not willing to criticise the court. Judge Jackson encouraged him to do so, to much laughter in court. The most worrying outcome for Microsoft may well be that after Farber had completed the installation of Microsoft software on a PC to test something, the machine was immediately reclaimed by his students - for running Linux. ® Complete Register trial coverage
Sponsored: RAID: End of an era?