IBM witness: the inside poop on MS and IBM killing OS/2
The DoJ has had considerable difficulty finding the right person and company to give hard evidence on how Microsoft deals with OEMs, because of the fear of retaliation by Microsoft. IBMer John Soyring was the wrong person, as it turned out (we've been saying that about him for five years - Ed), but at its second attempt, the DoJ seems to have struck gold with Garry Norris.
As far as possible, we have chronologically restructured Microsoft counsel Richard Pepperman's arachnoidally-conducted examination, but there are some areas where it is not possible to be sure of what happened because of ambiguity in the transcript record that was insufficiently probed by Pepperman. We have indicated these areas, and sometimes suggested an explanation. The cross-examination is likely to provide more illumination.
In 1994 Microsoft proposed to IBM a frontline partnership, along the lines of the one it had with Compaq. IBM had proposed development projects with Microsoft, but in each case Microsoft declined on the grounds that IBM was a competitor, with the result that IBM found itself at a disadvantage in the marketplace. IBM would receive a lower price for Microsoft products, with better technical, marketing and development support - but providing IBM agreed to drop OS/2.
A meeting between IBM and Microsoft was held in Chicago in July 1994, when IBM VP Tony Santelli met Joachim Kempin, the Microsoft VP in charge of OEM sales. The subject was the improvement of relations between IBM and Microsoft. It seems that Dean Dubinsky (previously IBM's relationship manager, and based in Seattle) for IBM, and Mark Beber for Microsoft, were then charged with the preparation of a framework document entitled "Alliance Proposal - IBM-Microsoft relationship overview".
IBM considers dropping OS/2
Pepperman introduced what was apparently a later draft of this as a confidential exhibit (dated 5 April 1996, but it is more likely that this is a production date for legal document discovery). The document was subsequently reviewed by the IBM Corporate Executive Committee (CEC), which today is composed of the 13 most senior IBM executives. What was decided is something of a conundrum, since Norris' evidence is ambiguous on the point, and Pepperman did not clarify it. The best interpretation of the evidence available at the moment is that the CEC may have changed its mind as to whether to drop OS/2, which Microsoft was of course pushing for, in order to get a better price for Windows 95.
Alternatively, the PC Company may have acted alone, or with PSP (IBM personal software products). Dubinsky drafted a letter that Thoman sent to IBM CEO Lou Gerstner, produced as an exhibit by Pepperman, that stated that the CEC had not approved what Thoman called the "Alliance". This letter also says: "IBM PC company to reload [sic] OS/2 Warp on all systems." It is the word "reload" that suggests that the CEC may have reversed a previous policy.
It is also worth recalling that Microsoft's original problem with the FTC in 1989, before the DoJ took over, was over a collaboration between IBM and Microsoft whereby OS/2 was designated for the business market, and Windows for the consumer market, this being suspected as being anti-competitive market splitting. This may have been one reason why the CEC decided to compete with Microsoft, if that was in fact what happened. When Norris took over, the relationship between IBM and Microsoft was poor "with little trust", according to a document produced by Pepperman but not identified precisely. Microsoft's principal reason for its attitude was fear of OS/2. At a briefing around March 1995 when he took over his new job, Norris was told that Microsoft had been told that IBM would not stop shipping OS/2, and that IBM would not promote Microsoft products exclusively or jointly. Norris confirmed that when he took over the Microsoft relations job, IBM paid no royalty on MS-DOS (or of course on PC-DOS), that the royalty on MS-DOS tools was $2, and on Windows 3.x it was $9 (believed to be the lowest in the industry, and valid until September 1997).
Gates loses face
There then appears to have been another meeting involving Rick Thoman, Tony Santelli, and Bruce Claflin for IBM, and Gates and Kempin for Microsoft, at Comdex in November 1994. IBM apparently informed Microsoft that it would be loading OS/2 on every capable PC (i.e. with at least 8MB), with Windows 3.11 also loaded as a dual boot alternative. Microsoft was also told that IBM would follow an 'IBM First' policy, whereby sales reps would first try to sell IBM products. If the potential customer wanted Windows, IBM would support the product to ensure that the customer enjoyed "a good end-user experience".
The events at this meeting (which will not be considered by the court, since it is hearsay evidence) were described by Norris in his deposition, although he learnt about them second hand during a briefing four or five months later, when he was being briefed on his new appointment: "I was told that Gates was surprised; that Kempin was expecting agreement for us to jointly and exclusively promote Microsoft products, reduce shipments of OS/2; that he was embarrassed because he didn't expect that to happen. I was then told that over the ensuing months, up until the time I got there, that Microsoft had taken a number of retaliatory actions against IBM, reducing coverage of the IBM account from three people to one. They notified us that we would be treated like any other OEM. They became non-responsive to phone calls, slow in getting beta code.
"We would ship products up for capability testing. The test took 60 to 90 days and should have taken one to two weeks. We were missing systems on the compatibility lists, when our competitors' same systems, similar-configured systems, were on those compatibility lists. We were cut off and uninvited to several OEM events. And some other things that I don't recall at this time."
It appears likely that many of these retaliatory actions were initiated by Kempin, who had clearly lost face before Gates at the Compaq meeting. Microsoft was particularly worried by OS/2 at this time, because IBM was marketing it very hard, and sales were picking up. Norris saw his first job as being to renegotiate the market development agreement (MDA) that had been signed in February with Microsoft, to obtain additional better terms. Norris achieved this later in 1995. Microsoft wanted $75 from IBM for Windows 95 - a very high price - but was willing to offer some discounts in an MDA. Microsoft offered an $8 reduction if IBM agreed to "restrict, limit or delay shipping OS/2".
MS puts the screws on
Additional reductions were offered if IBM ramped-up Windows 95 shipments to 50 per cent of PC shipments in 60 days, but IBM was unable to meet the requirements and thought they were over-optimistic on Microsoft's part. Close analysis of the Norris' deposition transcript shows uncertainty as to whether a decision was in fact made and implemented to squash OS/2, in order to take advantage of an $8 discount for Windows 95. No doubt there will be further evidence about this when Norris is examined and cross-examined. Many OS/2 developers and users have felt betrayed at the sudden change in IBM policy towards OS/2. Many feel that it was a moment of infamy: but we now know that eight pieces of silver were offered, and probably taken.
Norris negotiated the agreement for a Windows 95 licence, with a further $7 of discounts agreed in mid-1995. The first draft of a Windows 95 licensing agreement was received by IBM around the time Norris took up his new appointment. IBM did not commit wholeheartedly to Windows 95 at the time - probably as a negotiating tactic, although some IBMers might have thought that OS/2 could compete with Windows 95. The agreement was finally signed 15 minutes before Windows 95 was launched, on 24 August 1995.
Four people from IBM attended the Windows 95 launch event in Redmond. Microsoft's reason for wanting a high price for Windows 95 was that IBM competed with Microsoft, whereas Compaq did not (and is believed to pay $25 for Windows 9x as part of its frontline partnership). IBM's concern was that in 1995 it had paid $40 million in Windows royalties, whereas in 1996 the anticipated royalty level would be $220 million, with the price of Windows having increased from $9 to $45.90. This should provide strong evidence of monopoly leverage, and negate evidence from Microsoft's economist Richard Schmalensee.
A key Microsoft negotiator was Mark Beber, who previously had worked for IBM, as did his sister, and his grandfather and father. He called Norris around 19/20 July 1995 and said that he had been instructed to cut off negotiations about a Windows 95 licence until the audit of IBM's Windows 3.x sales had been settled. Norris consulted Tony Santelli, who was two levels senior to him in the IBM hierarchy. It appears that Microsoft wanted to negotiate at a higher level in IBM, probably because Norris was proving to be tough. An original list of 38 contentious issues had been reduced to less than 10 by Norris before Santelli became involved in the negotiations. Norris did the briefing of IBM's management, and drafted communications.
Chairman Bill demands respect
Rick Thoman subsequently wrote to the IBM CEC: "A general point is their perception of IBM's non-respect for Microsoft. Gates was irate because of the lack of respect he feels IBM has for Microsoft. He cited Lou Gerstner's quote in Business Week that Microsoft was 'a great marketing company, but not a great technology company' as an example. ... He [Gates] also cited 'smear campaigns' planned by Dan Lautenbach and others against the Windows 95 product." Pepperman also revealed in his questions that IBM alone was granted permission for third-party installers to load Windows 3.11, but at first this was not included in the Windows 95 agreement, although it was later allowed again. It was also revealed that IBM negotiated the period between the release of a bug fix (OSR) and the beginning of IBM installing this to be increased from 90 days to 120 days, since IBM said it needed the time for testing. Other OEMs had only 90 days.
Norris said that an IBM concern was that Microsoft might make ten releases a year, and that frequent quick releases would be burdensome. (The provision was included in the Windows Family Desktop Agreement, which was signed in 1996.) IBM was pressed by Microsoft to agree to installing 300,000 copies/month of Windows 95 in the first five months after release, failing which IBM would be liable to pay an additional 20 per cent. IBM was sure this was impossible, but signed in desperation to get the code, it seems. In the event, IBM shipped less than 100,000 copies/month, but perhaps because of its earlier protests, Microsoft did not seek the 20 per cent uplift, although IBM did agree to an increase from $45.90 to $46.60 per copy, with an agreement for 5 million copies a year. We now know that Microsoft was engaged in momentum marketing, and did not wish it to leak out that real sales were running at less than a third of what Microsoft hoped. This also accounts for the unbelievably large sales that Microsoft was claiming shortly after the release on Windows 95. The same pattern has been seen for claims about NT sales, although these have been mainly confined to percentage increases rather than hard numbers.
Pepperman probed to find out whether retailers were saying that they did not want OS/2, but Norris said there were dual boot OS/2-Windows 95 PCs being sold. From September 1995, IBM introduced the 'IBM First' policy. In a document Pepperman produced dated September 1995, and also used in a presentation prepared by Norris for the IBM CEC, a bullet point read, "'IBM Only' solutions have caused client/server customers to lose confidence." Norris said that at the time, IBM-Microsoft relations were still poor.
Breakup hits IBM sales
Because of the delay with the licensing, IBM had missed market opportunities and was late shipping PCs with Windows 95. The document also said that "PC Company needs to repair its Microsoft relationship in order to respond to customer needs and avoid market share erosion". Norris said that: "As a result of that license agreement discussion and missing those advantages that our competitors had, we attempted to get in some of Microsoft's enabling program, the solution programs, the authorised technical centre program, the authorised support center program, and the education centre programs. But Microsoft rejected us on every attempt. Our competitors were a part of those program. So we were selling IBM Solutions. In some cases we were leading with OS/2 by selling IBM Only Solutions. And what the chart was meant to convey was just that -- that without the Microsoft solutions, without the Microsoft certifications, without the Microsoft assistance, that we would begin to see market share erosion. And in fact, Compaq and Microsoft and others had begun telling our customers this."
Norris added that there were customers who said that they were losing confidence in IBM, and that as a consequence IBM decided to repair its relationship with Microsoft in the fall of 1995. Norris claimed a loss of $175 million at the time because of its inability to support customers. IBM proposed co-marketing to Microsoft, but it was always rejected while IBM was "shipping those competitive products".
IBM tried to license Microsoft Office, but Kempin said in a conference call that Microsoft wanted $250 for it, the same price as for OEMs that ship their machines through retail channels. As a consequence, IBM continued to ship SmartSuite.
It is going to be fascinating to see what happens in court when Norris is examined and cross-examined. Since there will be no written direct testimony, the Microsoft legal team will find itself at a disadvantage of not being able to spend days researching matters disclosed in Norris' primary examination.
The information in this chronology has been derived from the Norris deposition, but there are gaps and inconsistencies that cannot be resolved based on what Norris said. Microsoft provided a substantial part of the new information in documents it produced during the deposition, and in leads it provided in questioning the witness. We welcome comments and information from readers to clarify issues in The Register's Bulletin Board (see right-hand column), where we shall post clarifications and extensions, or to Graham_Lea@compuserve.com ®
Last update: 01.12.2000