But that was only a baby step. We had more ambitious goals.
In our meetings with each side, Jack and I had begun asking the sorts of questions that Spock would ask: What are your company’s interests? Think about the legitimate interests of the other side. How might we together develop some creative options that might serve the interests of both sides better than conventional arbitration?
IBM’s primary interest, clearly, was to protect its intellectual property. To the extent that Fujitsu used its technology, IBM also had an interest in being adequately paid for that use.
Fujitsu’s primary interest was to stay in the compatibility business. It wanted a real shot at being able to develop compatible software through its independent development efforts.
That much was easy to figure out. But our discussions also suggested to me that the parties had a common interest in something they hadn’t yet identified: certainty. Certainty about what materials Fujitsu was entitled to use in its development of IBM-compatible software. That clarity was what the 1983 Agreements had failed to provide. I said, “Look, for this conflict ever to end, we’ve got to have a set of standards. There have got to be more precise rules. Unless clear standards are established, won’t you be stuck fighting for all eternity?” I asked each side, “Wouldn’t both companies be better off if Fujitsu knew in advance what information it can use, rather than fighting these things out after Fujitsu has released a program to its customers?”
We were now deep in a Spockian process of thinking things through—trying to solve a hard problem with all the creativity we could muster. This was more fun than anything we had done so far. In fact, for me it was thrilling.
In my mind, the shape of a possible deal gradually emerged, one that would involve tearing up the old 1983 Agreements and starting from scratch. With the benefit of hindsight, I saw that these agreements had been doomed from the start.
One flaw related to the DP list—which had not been a bad idea in theory. Fujitsu wanted to escape the onerous burden of the continuing semiannual license fees, and the 1983 Agreement had given it an incentive to “clean up” its tainted programs by writing successor programs that somehow didn’t violate IBM’s rights. But without any meeting of the minds about when a successor was “clean enough” or what part of a DP Fujitsu could use in some future program, new fights were inevitable.
In actual operation, the DP scheme struck me as incredibly inefficient. Fujitsu was spending time and money redeveloping old programs to get rid of the taint. IBM was spending time and money monitoring and tracking Fujitsu’s successor programs and fighting over whether the new programs were clean enough. Wouldn’t it make more sense, I thought, to avoid these disputes by simply allowing Fujitsu to use the DPs any way it wanted in its future software development—for a price? I suggested to IBM, what you ought to care about is that Fujitsu pays you adequately for this privilege. Let’s get rid of the ongoing license fee and establish a lump-sum payment that allows Fujitsu to use its own DPs freely.
This, too, took some persuasion. IBM was attached to the ongoing license fees. They liked the income stream. I also sensed that IBM valued the whole setup as a way to make Fujitsu’s life as miserable as possible. I said, “Fujitsu has taken this material from your old programs. Sell them a paid-up license and let them use their DPs anyway they want. Admittedly, a fair price will need to be established for a paid-up license. You and Fujitsu will have to negotiate the amount—or the arbitration panel will have to set it—but that will only happen once. And then you’re done!”
Gradually IBM came around.
A second flaw in the 1983 Agreements was its total lack of guidance with regard to the “external information.” IBM was required to pro-vide some information to Fujitsu, but exactly what was it? At what price? How would Fujitsu developers get access to that information and only that information?
This was the most difficult conundrum of all.
The simple solution, Jack and I thought, would be for IBM to write down the “external information” for each program on a piece of paper and then sell Fujitsu a license to use it. When we asked Fujitsu whether this would be acceptable, they not only rejected the idea but ridiculed it. They actually laughed. IBM would have every incentive to leave something out, they explained—something essential but not obvious. It would be like buying an expensive kit to build an elaborate model airplane. You would spend weeks trying to assemble it, only to find that an essential piece was missing and the plane wouldn’t fly. And you wouldn’t know which piece was missing! No, Fujitsu would only trust its own people to extract the externals.
IBM in turn found that idea ridiculous. Giving Fujitsu access to IBM source code would be like opening up the cookie jar to a child with a ravenous sweet tooth, IBM protested. No matter how strictly the rules were defined, Fujitsu would take more than they were entitled to.
Jack and I eventually found an answer, but it wasn’t simple. We would come to call it, somewhat ponderously, the Secured Facility Regime.
First, we decided, we would create a precise definition of external information. “Instructions” would articulate that definition. Then we would set up a secure work site managed by an independent expert, in a location far away from all of Fujitsu’s operations. “Secure” would be an understatement: the place would be tighter than Fort Knox. A handful of Fujitsu programmers would work there, isolated and under guard. They would be permitted to examine IBM programming materials, yes, but they’d only be allowed to write down (on “survey sheets”) the external information as it had been defined in the Instructions. That’s all they would be allowed to do. They would have absolutely no role in writing software and no contact with their colleagues who did. (I didn’t envy the lonely souls who would end up working in our airless little capsule.) Before the survey sheets left the building, they would be vetted by IBM or an independent expert. If approved, they would be sent to the Fujitsu engineers who were writing new programs. Needless to say, that second group would have no access to anything but the survey sheets.
It was an exhausting process to contemplate, but the parties liked the concept because it offered real security to each.
But how could this scheme become operational? Many details remained to be filled in. Were the parties prepared to negotiate these issues directly with each other? Both parties said they’d try, but neither had any confidence that they would ever reach agreement. Then what?
The parties had the 1983 Agreements to fall back on, but that wasn’t a big comfort. So I made a bold suggestion: Why not give the panel the power to make any decisions necessary to implement this new regime—but only to the extent IBM and Fujitsu were unable to agree?
Barr rather quickly responded that IBM would give the panel this sweeping power. He had only one condition: that going forward, the panel would consist only of Jack and me. He didn’t say why he didn’t want Macdonald to be part of the new regime, and I didn’t ask. Barr did say IBM had a good deal of trust in Jack and me, largely because they knew we understood their interests.
Fujitsu, however, really struggled with this decision, I later learned. Certainty sounded good in theory, but ambiguity, after all, has its advantages. Specificity can be terrifying, especially if it forces you to accept fewer benefits and tighter restrictions than you think you are entitled to. Moreover, Fujitsu had been hauled into arbitration unwillingly, and the idea of giving the panel so much coercive authority was a bitter pill. But what was the alternative? Fighting things out under the 1983 Agreements? If they wanted to continue down the compatibility path, they were stuck. They were going to have conflicts with IBM. The only decision was whether the Jones-Mnookin regime would be better than conventional arbitration.
Fujitsu took the leap. Jack and I were willing to serve without Macdonald, and Macdonald gracefully resigned.37
In February 1987, the parties signed a new accord laying out this framework. (We called it the “Washington Agreement” because it was signed in Washington, D.C.) T
he agreement was only a few pages long and had taken only four months to create.
We gave the parties sixty days to negotiate the details. As expected, they failed.
A new hybrid process was about to begin—one in which Jack and I would each be wearing two hats. We would mediate where possible and rule when necessary. Over the summer, Jonathan Greenberg and I wrote an Opinion and Order describing the framework. (Jonathan became counsel to the panel and would remain closely involved in its work for the next ten years.)
In September, Jack and I held our first big press conference and let the world in on our secret. As you’ve no doubt gathered, we thoroughly enjoyed our fifteen minutes of fame.
Then the hard work really began.
During the next year, our biggest challenge was to create Instructions that defined with real precision just what information from IBM programs Fujitsu would be allowed to extract in the Secured Facility and then later use in software development.
The judgment calls were tough. Before we could define external information, we had to grapple with such metaphysical questions as “What does IBM-compatibility mean?” Or to be more precise, to what extent should Fujitsu’s operating system legitimately aim to be IBM-compatible? We soon realized that compatibility was not a point, but a spectrum. Where did Fujitsu deserve to come out on this spectrum?
I can assure you the two parties fought like crazy over this issue. After briefs and argument, Jack and I decided that Fujitsu was entitled to a reasonable opportunity to maintain certain types of compatibility, but not others.38
There was no way Jack and I could have done this without technical help. Barr and Raven were equally clueless. To assist us, we assembled superb technical teams from both companies who did most of the heavy lifting.
These teams worked surprisingly well together, and we protected them from the litigators as much as possible. (Some of the lawyers were not happy to be sidelined and remained a bit dubious about the whole regime.) The teams were headed by Ron Alepin of Fujitsu and Bill McPhee of IBM, two software geniuses whom I could occasionally understand, and by lawyers Tony Clapes (of IBM in-house) and Mike Jacobs of Morrison & Foerster, who understood programming at a sophisticated level and could articulate the concepts in a way that could be consistently applied.
In less than a year the Instructions were completed. As you may recall, the 1983 Externals Agreement had tried to define “external information” in two sentences. Our Instructions ended up being more than forty single-spaced pages.39 Through a rule-making process, we also ended up promulgating hundreds of pages of detailed rules providing strict and elaborate safeguards governing the operation of the Secured Facility, and ensuring that Fujitsu software programmers outside the facility did not have access to IBM programming materials.40
Finally, we had to resolve two conflicts over money.
The big issue was the price for the new paid-up license: how much Fujitsu would pay for the right to use the DPs in its own software development. Needless to say, the litigators were primed for a fight on this issue. IBM was asking for more than $2 billion. After receiving preliminary briefs, we announced the legal standard to be applied and the key variables we would be estimating in determining an amount.41
Although the financial stakes were high, Jack and I didn’t find the task of determining the lump-sum payment all that daunting. Indeed, Jack had been a top executive at a huge railroad and had made lots of big financial decisions. I found his attitude comforting. “After all,” he told me, “it’s only money! And both companies have plenty of it.” I agreed. In our view, the really important work was creating and implementing the new regime so that both companies could get on with their real business: competing in the marketplace. Setting a price for the paid-up license was just cleaning up their old mess.
But that was easy for us to say. Each company assembled vast teams of lawyers, engineers, and consultants who spent many person-years preparing massive submissions concerning the variables we had identified as relevant to the determination of the price for the paid-up license. I’m sure each party spent millions. At times the strain on the lawyers showed. A few of the younger lawyers got so caught up in the fight that they started to attack each other in letters to the panel. Personal insults were flying back and forth. Jack and I eventually lost patience with this nonsense. We warned them that if they didn’t cut out the personal attacks, we would impose fines on individual lawyers. Fortunately, we never had to carry out the threat.
In the end, we found it neither difficult nor time consuming to establish a fair price for the paid-up license. While some courts might have heard months of live testimony and argument, we limited the actual hearings to one week. The only live testimony consisted of the cross-examination of each party’s experts.42 We gave counsel a chance to make oral arguments, and submit post-hearing briefs. Then we promptly made a decision,43 setting the total price of the paid-up license at $396 million.44
The smaller money issue related to the annual fee that Fujitsu would pay for the “external information.” (As you will recall, the 1983 Externals Agreement was useless in this regard—the parties had neglected to set a price.) After briefs and argument, we decided that for the first year in which the regime was in operation, Fujitsu would pay between $25 million and $52 million, depending on how many IBM programs it wanted access to. This gave Fujitsu an incentive to save money by accessing fewer IBM programs.
It was then again time to go public, with another carefully orchestrated round of press conferences both in New York and Tokyo. In our first round, Jack and I had simply sketched out a framework. Now we had put meat on the bones and we wanted people to see what we and the parties had created. We released an opinion, prepared with Jonathan Greenberg’s able assistance, explaining the details, including the Instructions and the dollar amounts Fujitsu would be required to pay.
Again, there were hundreds of newspaper articles around the world, most very laudatory. As the Wall Street Journal reported, some observers hailed what we had done as “a model for resolving arcane, high-tech lawsuits.” But others considered our role controversial. A Dean Witter analyst harrumphed, “This is a warning to other large companies: don’t throw your fate in the hands of an arbitrator. You never know what they’ll do, and they’ll order you not to say anything about it. They are above the lawyers and above the courts.”45 Ah, those were the days!
In May 1989, the final piece fell into place: the Secured Facility was up and running. Of course, its location was a carefully guarded secret. In our rather quiet announcement of this milestone, we just said it was in Japan.
No expense was spared in creating this hermetically sealed chamber. If we and the parties could have put it on the moon, we would have. Instead we located it in Sakura, an obscure town about forty miles outside Tokyo, on the top two floors of an eight-story office building. Inside the building’s sole elevator, you would never know that floors seven and eight even existed. We had the buttons and numbers for those floors removed from the elevator panel. You could only gain access to these floors with an elevator key, and only the facility manager and the guards had them. No one else reached the seventh or eighth floor unless their names were on a list, and even then they had to be personally escorted. There were guards on both floors twenty-four hours a day, 365 days a year. As far as we knew, no one in Sakura—or even the building—ever knew what was going on in there.
The handful of Fujitsu programmers who worked in the facility were not permitted to bring anything in or out of the inner sanctum—not a cell phone, not a wallet, not a scrap of paper. (Personal items had to be checked in a locker.) They had no access to a phone. Every time they went in or out, their movements were documented. The IBM source code was kept in a locked safe. We used security consultants to design the system, of course, and before the facility opened we hired a different security expert, whose primary client was the CIA, to check things out. He said our facility was more secure than either the Pentagon or the CIA headqu
arters in Langley, Virginia.
For the next seven years the regime operated without a hitch. As an extra safeguard, IBM built its own secured facility in Hampton, New Hampshire, to inspect Fujitsu’s new programs. IBM had the right to bring a claim before the panel if it found evidence that Fujitsu was breaking the rules—and I assure you, IBM looked. But it never brought a claim.
Let me begin my assessment by turning back the clock to June 1985, just before the arbitration began. Both IBM and Fujitsu were awash in strong currents of demonization and tribalism. Was it wise for them to give up on negotiation and go to war? Or did they simply get caught in emotional traps?
To answer this question, let’s suppose each side had been able to consult with Spock and consider his advice dispassionately. In 1985, could they have resolved their conflict through negotiation? What could they have done differently?
With regard to IBM, Spock would have pressed them on whether their hardball strategy in 1984 was wise. As you recall, Fujitsu offered to add many programs to the DP list in 1984, plainly signaling that it was prepared to negotiate a deal in which it would buy immunity for these programs. IBM refused this offer and demanded a draconian remedy: triple fees and the withdrawal of the offending software from Fujitsu’s customers. It was these continued demands—and the threat of arbitration—that triggered the complete breakdown of negotiations in 1985.
Spock would have asked IBM and its lawyers, “How likely is it that an arbitration panel would grant you this harsh remedy?” Spock would have given them slim odds. He would have pointed out that the 1983 Settlement Agreement was a contract, and contract law does not allow punitive damages. He would have noted that an arbitration panel probably wouldn’t order Fujitsu to pull the offending programs off the market because it would disrupt the business operations of innocent bystanders. In fact, I have a hunch that the lawyers on both sides of this case made similar assessments at the time.46
Bargaining with the Devil Page 20