Analysis: How MS used the WISE Trojan Horse against Unix
The jury said it wasn't illegal - but it certainly wasn't nice
MS on Trial Bristol may yet appeal after its loss against Microsoft (as of Friday, it hadn't come to a decision on this), but as the case stands it serves as an awful warning for companies doing business with Microsoft. A nice little business took it into an alliance with Microsoft, then it got sucked in deeper, and then the carrots started to disappear, replaced by the stick. For Bristol, and for small developers in general, it's a depressing tale. Since its formation in 1991 Bristol has produced Wind/U, a set of libraries and utilities which allow software developers to implement their Windows applications on Unix, OpenVMS and S/390 platforms. This was originally an independent initiative, but after various Microsoft approaches Bristol signed a deal with the company in 1994. This licensed Bristol Windows source code under the Windows Interface Source Environment (WISE) programme. The hook The good news for Bristol and its customers was that by having access to source Bristol could improve the performance and compatibility of Wind/U. The bad news, as it transpired, was that it made the company dependent on continuing access to the source code. This access was vulnerable to changes in pricing, restrictions on the code Microsoft was willing to give out, and changes in Microsoft policy. All of these factors were clearly identified in the case Bristol made. Microsoft had originally identified WISE as a mechanism which would help it get NT established in corporate networks. In the early days of the OS Microsoft needed to accept that there would be co-existence (NT's market share was then vanishingly small), so WISE was useful. It also acted as a mechanism for controlling that coexistence. Microsoft then saw Sun-backed efforts such as WABI and PWI as real threats that could wrest control of Windows APIs from it, so it favoured WISE as an 'official,' controllable version. Once WABI and PWI had bitten the dust, Microsoft started using WISE as a weapon against Java. Documents authored by Microsoft's Bob Kruger and produced in court support this. One is headed "WABI death knell," while "An Open letter re PWI" says: "another virtual API bites the dust. Good riddance." Kruger's then boss, Dan Neault, meanwhile explains Microsoft's long-term objective: "[I]f customers will only develop supporting code near the services this code supports, Msft should make available on UNIX APIs in those technology areas where new development is going to happen, so that it can easily be moved to NT when customers will accept NT services for those areas now deemed unacceptable." That is, Microsoft saw WISE as a mechanism for positioning itself so that Unix to NT migration would be eased for the time when customers would accept NT. On the Java front, the email traffic covers Neault, Paul Maritz and Bill Gates himself. Gates urges Maritz to co-opt Unix and "avoid the Java interfaces becoming the primary ones the ISVs use," and Maritz responds that Neault is doing this through "Windows APIs on Unix" (i.e. WISE). Marketing material from the time makes it clear that Microsoft was positioning WISE as a long-term initiative that licensees and their customers could have confidence in. That meant continued access to the source at reasonable prices, and access to it on future versions of Windows. But the honesty of this pitch became somewhat dubious. But Bill starts to wonder By 1996 Gates is wondering if "by creating cross-platform solutions we risk weakening the Windows franchise - hurting ourselves rather than let Netscape hurt us." Neault's staff are writing that they want "the WIN32 layer to be fairly mediocre in performance and feature coverage. We want it to be just good/cheap/timely enough to get a lot of people to use it," and that "we don't want it to work too well. A non-objective is total redeployment of Windows on Unix." The result of this was a reduction of the technologies Microsoft was prepared to ship, to "an apple core of technologies." This hit the server side particularly, where Bristol was strongest. Neault actually devised a list of technologies to "jerk" from the source, describing it as a "beautiful" list which removed "the technologies that third-parties are going to try their damnedest to put on Unix." Some more emails show where Microsoft was at. Says our old friend Jim Allchin: "What we're trying to do is to get Netops to add NT services to their existing Unix networks and to migrate over time all of their services to NT." Maritz tells Gates that Microsoft can now switch from the carrot to the stick, while then WISE manager Ramesh Parameswaran suggests developing "COM+ stub APIs on Unix and `receiving' service on NT Server. This approach will help ISVs who want COM+ services on Unix and will start migration to NTs since NT Server will be required to provide the COM+ services." Neault's loyalty code Pricing strategy also changed, increasing substantially but with the addition of a royalty waiver programme which introduced further carrots and sticks. (MS enemies list)Here's Mr Neault on the subject: "We need to consider approval of [waiver requests]in light of criteria along the lines of the following: 1.Do we want to help these specific applications in the marketplace? 2.Do we want these ISVs to flourish (e.g. are they avoiding things that are to the Windows detriment)? 3.Even if 1 and 2 aren't the case, is it important/is this waiver needed to move these ISVs to a single Win32 codebase? 4.Do we want these ISVs to change their behaviour (quid-pro-quo - possibly requiring a press release attesting their loyalty)? [ah, a signed confession] 5.Does the application under consideration exceed criteria above beyond the extent they build demand for a platform that is competitive with Windows?" The fact that Bristol's rival, Mainsoft, accepted the restrictions and prices of Microsoft's new licence agreements was an important plank of Microsoft's defence. If Mainsoft accepted, then what was wrong with Bristol? Wasn't the company just - as Microsoft argued in its closing statement - being greedy? Furthermore - and this is why the failure to convince the jury a monopoly existed was important - doesn't Microsoft have a right to decide on the technology it licenses, and how it is priced? Bristol argued that the changes to the licence agreements were an unfair use of monopoly power, and that by agreeing them Mainsoft became "an active aider and abetter of Microsoft's deliberate scheme to restrict competition." The contract with Mainsoft, said Bristol, was unlawful. The terms and conditions Microsoft was willing to offer Mainsoft and Bristol certainly did deteriorate over the years, and this process was certainly driven by top level Microsoft executives' desire to damage potential competition (WABI, PWI, Java) while using WISE as a Trojan Horse to allow NT to infiltrate Unix markets. It also seems clear that, having been hooked on the source code, Bristol and Mainsoft had little room for manoeuvre. Mainsoft had first been promoted as a lever against Bristol, but later Microsoft deemed the ploy had been sufficiently successful for there to be no need to "further anoint" the company. Carrot and stick again. Sue or co-operate? Mainsoft chose to co-operate and play by Microsoft's rules, while Bristol chose to sue, and - so far - Bristol has lost. Nevertheless, although it lost the antitrust action, its summary of its original complaint remains eminently valid. Said the company: "As applied here, the antitrust laws prohibit a monopolist from using its monopoly power in one market as a lever to gain a monopoly in a second market. The violation consists of creating a Trojan Horse by marketing the WISE program as the guarantee of an open standard for ISVs, and then hobbling that horse to assure that ISVs who accepted the invitation to 'innovate' using the Win32 APIs would not be able to run their programs on Unix servers. Microsoft achieved this by restricting WISE licensees to an 'apple core,' and specifically by excluding technology necessary to support server applications." Bristol, with the aid of Microsoft's internal emails, made it clear that all this happened, but failed to convince the jury it was illegal. Bristol now, with justification, argues that Microsoft did engage in anticompetitive activities (despite Microsoft's triumphant claims of vindication). But what, in light of the verdict, can the remedy be? ® Complete Register Trial coverage
Sponsored: RAID: End of an era?