Bargaining with the Devil

by Robert Mnookin

  47. Even moves that might have promoted progress across the table ended up hampering it. In March 1992, de Klerk unexpectedly called for a snap referendum, asking white voters whether they supported his decision to enter into negotiations to create a new South Africa. Although the referendum question did not specify what kind of political structure the new nation would have, de Klerk promised to ensure that the white minority would be protected from the black majority through some sort of constitutional provision. It was a promise that he would not be able to keep, but he won by a two-thirds majority, with more than 80 percent of all whites participating in the referendum. Such solid backing behind the table did not help move the negotiations forward, however. Instead, de Klerk became more inflexible. He had a strong mandate for negotiations, but not for a concession on a key issue: one man, one vote. See Sparks, Tomorrow Is Another Country, p. 134.

  48. The strike immobilized the industrial centers and powerfully showcased the party’s strength in numbers. Sparks, Tomorrow Is Another Country, p. 140.

  49. On June 17, an Inkatha posse raided the township of Boitapong. Mandela, Long Walk to Freedom, p. 603.

  50. Martin Meredith, Nelson Mandela: A Biography (New York: St. Martin’s, 1998), p. 495.

  51. Sparks, Tomorrow Is Another Country, p. 151.

  52. Lodge, Mandela, p. 179.

  53. Sparks, Tomorrow Is Another Country, p. 152.

  54. Waldmeir, Anatomy of a Miracle, p. 231.

  55. Martin Meredith, Nelson Mandela: A Biography (New York: St. Martin’s, 1998), p. 493.

  56. See Robert D. Putnam, “Diplomacy and Domestic Politics: The Logic of Two-Level Games,” International Organization 42:3 (1988): 427.

  57. Waldmeir, Anatomy of a Miracle, p. 252.

  58. Ibid., p. 232.

  59. My colleague Roger Fisher first coined this phrase. See Roger Fisher and William Ury, Getting to Yes (New York: Penguin, 1991).

  60. http://nobelprize.org/nobel_prizes/peace/laureates/1993/press.html

  61. “Nelson Mandela & F. W. De Klerk,” Time, January 3, 1994.

  62. Waldmeir, Anatomy of a Miracle, p. 231.

  7: GIANT SOFTWARE WARS

  1. Some of the IBM programs that IBM claimed Fujitsu copied were not, strictly speaking, operating system software programs but “middleware” programs that are closely related and worked in conjunction with the operating system. For various definitions of the terms, see p. 149.

  2. Dan Evangelista, telephone interview by author, March 4, 2009.

  3. Confidential interview by author, January 30, 2009.

  4. Robert Mnookin, “Spreading the Word about Spreadsheet Software,” California Lawyer 4(4) (1984):5. The first spreadsheet software program, VisiCalc, was introduced in 1979 and ran on the Apple II computer. Lotus 1-2-3 and Excel were successor spreadsheet programs that were aimed at the IBM PC market.

  5. Takuma Yamamoto, Fujitsu: What Mankind Can Dream, Technology Can Achieve, trans. Dick Belcher (Tokyo: Toyo Keizai, 1992), p. 18.

  6. This is usually done by the agreement of the parties or of the arbitrators they have designated. If they can’t agree, the forum institution—here, the American Arbitration Association—is empowered to select.

  7. For anecdotes about IBM’s willingness to bear staggering legal costs, see James B. Stewart, The Partners: Inside America’s Most Powerful Law Firms (New York: Simon & Schuster, 1983), pp. 97–98.

  8. The government surrendered in 1982. As the Justice Department explained, the computer world was changing and the government no longer believed it could prove that IBM was an illegal monopoly.

  9. Stewart, The Partners, p. 54.

  10. “A Celebration of the Life of Thomas D. Barr,” Cravath, Swaine & Moore LLP, April 4, 2008, p. 1.

  11. Under the American Arbitration Association rules at the time, a three-person panel could decide to proceed in one of two ways. One way, which I call the “partisan model,” allowed each party-appointed arbitrator to meet privately for discussions with the party (and counsel) that appointed him. Under this model, the party-appointed arbitrator is simply an advocate for that party and tries to win over the chairman. This process might eventually lead to a coalition between the chair and one of the party arbitrators, or the chair might end up mediating a compromise between the two arbitrator-advocates, but either way the chair essentially decides the outcome. There is very little room for mutual problem-solving or value creation. The other approach is for all three arbitrators to act as independent neutrals and impartially evaluate the case on the merits. See Opinion of Panel at n.3, International Business Machines Corporation v. Fujitsu, Ltd., Case No. 13T-117-0636-85, September 15, 1987 (American Arbitration Association Commercial Arbitration Tribunal) (Unpublished Opinion). Under contemporary AAA rules the “partisan model” applies only if the “parties have specifically agreed … that the party appointed arbitrators are to be non-neutral.” See AAA Commercial Arbitration Rules, Rule R-12 (b). See http://www.adr.org/sp.asp?id=22440.

  12. At that time there were seven other major computer companies, but people would joke about “IBM and the Seven Dwarfs.” Fujitsu was in the data processing business but not even a dwarf. See Paul E. Ceruzzi, A History of Modern Computing (Cambridge, Mass.: MIT Press, 2003), p. 248; Emerson Pugh, Building IBM: Shaping an Industry and Its Technology (Cambridge, Mass.: MIT Press, 1995), p. 296.

  13. Two developments in the 1960s led to the realization that IBM’s mainframe operating system software was of enormous economic importance. IBM’s dominance in the mainframe world was substantially enhanced after the introduction in 1964 of a new family of computers—the IBM System360—all of which used the same operating system. This allowed a business customer to start small and build up without a total reinvestment in software applications and peripherals each time the computer needed to be replaced by a newer and more powerful one. For example, a customer’s application program designed to run on a particular central processing unit (such as an IBM model 30) could, for the first time, run equally well on a more powerful processor (such as an IBM model 40).

  The second change was IBM’s decision to “unbundle” some of its software from its hardware and impose a separate charge or license fee. This was done for some programs in 1969 and for operating system software in the late 1970s. Although IBM had invested hundreds of millions of dollars in software development, the IBM software had previously been included as part of the hardware package. Unbundling allowed IBM to better match software development costs with revenues. Over time, it also made it abundantly apparent to IBM executives that software licensing fees were a potential gold mine.

  14. MITI assigned to Fujitsu the task of developing large and small computers and to Hitachi those that were midsized.

  15. This was the strategy of Gene Amdahl, who left IBM to start his own company to develop IBM compatible mainframe computers. Fujitsu would later acquire Amdahl. See Yamamoto, Fujitsu, pp. 86–89.

  16. Programmers write software in “source code,” which another programmer can read and understand, but which a computer cannot read until it is “compiled” or translated into “object code” consisting only of 0’s and 1’s—bits and bytes. Programmers cannot “read” or modify object code unless it is somehow “decompiled” or reverse-engineered back into source code. In the early 1980s, IBM generally stopped distributing source code of new programs or significant additions.

  17. Yamamoto, Fujitsu, p. 52. He also suggested that the desire to “add our own unique ‘Fujitsu functions’ [created] a veritable mountain of difficulties for our engineers to scale.” Ibid., p. 89.

  18. Ibid.

  19. Eventually Hitachi pleaded guilty in California to criminal espionage relating to hardware. An individual defendant confessed that he had been copying software as well as hardware. For a discussion of the Hitachi criminal case, see James B. Stewart, The Prosecutors (New York: Simon & Schuster, 1987), pp. 87–133. Hitachi later settled IBM’s civil claim and signed a settlement agreement that was very
similar in form to the one that Fujitsu signed with IBM in 1983. (See pp. 151–53 for a description of the IBM-Fujitsu Settlement Agreement.) I surmise from this that Hitachi also had improperly made use of IBM system software in its efforts to be compatible.

  20. Stewart, The Prosecutors, p. 111.

  21. Ibid., pp. 121–22. Because the quote was anonymous, we will never know whether it was accurate. My point is that it was published and highly inflammatory.

  22. Apple Computer v. Franklin Computer Corporation, 714 F.2d 1240 (3rd Circuit 1983), was the first appellate court decision holding that United States copyright law applied to operating system software. This decision was issued on August 30, 1983, about two months after the parties’ Settlement Agreement. Moreover, as discussed later, this holding would not necessarily be recognized in Japan.

  23. To turn up the heat on Fujitsu, IBM also went to MITI, the Japanese government agency, and suggested that Fujitsu might be encouraged to agree to adequate corrective measures to avoid another embarrassing public disclosure.

  24. A one-page “Protective Procedure” prohibited Fujitsu software programmers from having access to IBM materials, with certain exceptions that were not adequately defined. For example, there were exceptions for “research” and “high-level architectural design” but those terms were not defined.

  25. In order to write an application program that works with a particular operating system, a programmer must have certain interface information about the operating system so that the application can obtain services from the operating system. This interface information defines the form in which requests and deliveries for services must be made.

  26. Joseph W. S. Davis, counsel for IBM Japan, interview by author, February 24, 2009. Davis later wrote an insightful book about conflict resolution in Japan. See Joseph W. S. Davis, Dispute Resolution In Japan (Den Haag: Kluwer Law International, 1996).

  27. Philip J. McConnaughay, formerly a Morrison & Foerster partner who now is dean of Penn State Dickinson Law School, makes this point in his interesting article, “The Risks and Virtues of Lawlessness: A ‘Second Look’ at International Commercial Arbitration,” Northwestern University Law Review 453 (1998–99): 458. Phil cites and quotes my now-deceased Harvard Law colleague Professor Arthur T. von Mehren, who some fifty years ago, after a visit to Japan, wrote that “a potential source of misunderstanding in international transactions [with the Japanese is that] the parties may hold different conceptions of ‘contract’ even though both understand clearly the terms of agreement.” Arthur T. von Mehren, “Some Reflections on Japanese Law,” Harvard Law Review (1958): 1494 n. 25.

  28. Under the 1983 Settlement Agreement, a “Final Report” detailing alleged violations was a prerequisite to initiating arbitration.

  29. Masanobu Katoh, interview by author, January 26, 2009.

  30. I recognized Nicholas Katzenbach, sitting at the back of the room. By that time IBM’s general counsel, he was a revered figure to many in my generation. When I was an undergraduate in the 1960s, Katzenbach played an instrumental role in one of the most famous incidents of the civil rights era. As deputy attorney general in the Kennedy administration, he confronted Governor George Wallace, who was trying to prevent two black students from enrolling at the University of Alabama. This incident was later dubbed the “Stand in the Schoolhouse Door.” Katzenbach also served as attorney general and deputy secretary of state in the Johnson administration.

  31. Fujitsu maintained that all of its uses of IBM material were permissible because they: (1) dated to uncopyrighted versions of IBM programs and were therefore in the public domain; (2) did not constitute protected expression under the copyright laws of the United States or Japan; (3) constituted use of “External Information” within the intent of the 1983 Settlement Agreement and Externals Agreement; (4) were subject to immunity under the 1983 Settlement Agreement; (5) constituted research or high level architectural design authorized under the 1983 Settlement Protective Procedure; or (6) constituted standard programming techniques of a sort that were not subject to copyright. In other words, Fujitsu was arguing, “Between the DP list and the Externals Agreement, we can do anything we want.”

  32. The panel granted Fujitsu’s motion for summary judgment on two issues. We ruled that the provision of the Settlement Agreement relating to triple license fees was unenforceable because it was punitive. We also rejected IBM’s claim that, because Fujitsu had violated the Settlement Agreement, IBM had no further obligations to Fujitsu under the Externals Agreement.

  33. Deciding each individual claim would set a very narrow precedent that would be of little help in evaluating what “substantial similarity” meant in the context of another set of facts.

  34. For further reading on a model of mediation that favors the parties meeting together, see Gary Friedman and Jack Himmelstein, Challenging Conflict: Mediation Through Understanding (Chicago: ABA, 2008). I followed this approach as a mediator in chapter 10 herein.

  35. Macdonald, a Canadian, might have thought of himself as a North American rather than an American.

  36. Over the years, Fujitsu would develop deep ties to Morrison & Foerster. In appreciation of Raven’s contribution to the company, Mr. Yamamoto, Fujitsu’s president, would plant a tree in the special plot at Fujitsu’s Namazu facility to honor him. Today, Mr. Katoh is president of Fujitsu America.

  37. What would happen if Jones and Mnookin disagreed? We never did, but the Agreement for Further Arbitration provided for the appointment of a third arbitrator if necessary.

  38. Jack and I decided rather easily that Fujitsu was entitled to try to maintain application program compatibility—that is, that application programs written for IBM’s operating system should be able to run on top of Fujitsu’s operating system. A much more complicated and demanding form of compatibility involved “multi-vendor interoperability.” This mouthful related to the degree to which an operating system allowed customers to connect IBM and Fujitsu mainframe hardware. We decided to give Fujitsu only a limited opportunity in this regard.

  39. The process was an interesting one. When the technical team got stuck, we would offer informal “guidance,” or sometimes issue a ruling. Occasionally our rulings on these highly technical matters would make little sense to either side. The technical team would negotiate a revision we would gladly accept. Indeed, I suspect the in terrorem effect of having the two of us rule on highly technical questions created a powerful incentive for the technical teams to agree.

  40. We ended up promulgating literally hundreds of pages of rules and procedures in order to ensure that Fujitsu programmers writing the Fujitsu programs had access only to the information on the Survey Sheets, not to restricted IBM programming materials. As noted earlier, the 1983 Agreement contained a “protective procedure” that was about a page long and totally inadequate. It allowed Fujitsu programmers to have access to IBM materials only for “research and high level design,” but those terms were left undefined.

  41. We announced that IBM would be entitled to the lump-sum equivalent of the present value of its expected semiannual license fees under the 1983 Settlement Agreement for all of Fujitsu’s DPs, old and new. In making this calculation, we would take into account the fact that the programs would not remain on the DP list forever; Fujitsu had the ability over time to rewrite them to remove the taint. So there were lots of variables. What interest rate should be used to compute present value? How many Fujitsu customers would use each DP? How tainted was each DP and how long would it take Fujitsu to clean it up?

  42. We required each party to submit the testimony of its experts in written, affidavit form. The only live testimony was the cross-examination of these experts.

  43. As part of each party’s submissions we asked for spreadsheets into which we could plug our own estimates for each variable. Jack and I were comforted by the fact that when we plugged our estimates into each side’s spreadsheet, the IBM spreadsheet produced a lower dollar amount than the Fujitsu spreadsheet
. Moreover, the two spreadsheets produced numbers in the same neighborhood. Then we created our own spreadsheet and plugged in our estimates for different variables.

  44. I later learned that on each side some of the lawyers had a pool going. The social scientist in me would love to know what those estimates were.

  45. Michael Miller, “Fujitsu Payments to IBM to End Software Fight to Be $833.2 Million,” Wall Street Journal, November 30, 1988.

  46. Many years later, after the Secured Facility Regime was up and running, Barr offhandedly told me that he had never thought it likely that an arbitration panel would force Fujitsu to remove the offending software from its customers.

  47. When the Externals Agreement was signed in 1983, I suspect that no one within IBM thought it might be used to create the sort of affirmative obligation that the panel ultimately (and in my mind reasonably) imposed. Even if Spock had pointed out this risk, I suspect such a reading of the Externals Agreement was so far from IBM’s original intention that nothing similar to the Instructions—with its standards and rules—could have been negotiated. IBM might well have rejected a Secured Facility solution for strategic reasons as well. I could imagine someone within IBM thinking, “While the regime provides some safeguards for us, it lets Fujitsu know in advance what it can use. Why should we give them that kind of certainty and security? Aren’t we better off tormenting them after the fact?”