United States Court of Appeals
For the First Circuit
No. 97-1807
ADDAMAX CORPORATION,
Plaintiff, Appellant,
v.
OPEN SOFTWARE FOUNDATION, INC., DIGITAL EQUIPMENT CORPORATION
and HEWLETT-PACKARD COMPANY,
Defendants, Appellees.
APPEAL FROM THE UNITED STATES DISTRICT COURT
FOR THE DISTRICT OF MASSACHUSETTS
[Hon. Joseph L. Tauro, U.S. District Judge]
Before
Boudin, Circuit Judge,
Coffin, Senior Circuit Judge,
and Shadur, Senior District Judge.
Alan R. Kusinitz with whom Samuel Adams, Keith C. Long, and
Warner & Stackpole LLP were on brief for appellant.
James C. Burling with whom Michelle D. Miller, Hale and Dorr
LLP, William L. Patton, Jane E. Willis, Ropes & Gray, Kevin P.
Light and Choate, Hall & Stewart were on brief for appellees.
September 4, 1998
BOUDIN, Circuit Judge. Addamax Corporation brought a
federal antitrust suit against Open Software Foundation ("OSF"),
Hewlett-Packard Company and Digital Equipment Corporation. After
a bench trial limited to the issues of causation and damages, the
district court found that antitrust violations, even if they were
assumed to have occurred, were not a material cause of Addamax's
failure in the line of business at issue. Addamax now appeals and
we affirm.
We begin with a statement of those background facts that
are more or less undisputed. Addamax was created by Dr. Peter A.
Alsberg in 1986 and, in 1987, began to focus on developing security
software for Unix operating systems. Unix is a very popular
operating system for larger computers, and security software is a
component that can be used with the operating system to restrict
outside access to sensitive information and to restrict a
particular user to information consistent with that user's security
classification.
During this period, the National Computer Security
Center, a division of the federal government's National Security
Agency, rated security software, giving ratings (ranging from the
most to the least secure) of A, B-3, B-2, B-1, C-2, C-1 and D.
Addamax decided to produce B-1 software for Unix operating systems,
a level of security demanded primarily by government users. During
the years 1988-89, Addamax did develop B-1 security software for at
least two different versions of Unix.
While Addamax was trying to produce its security
software, a different struggle was developing between AT&T--the
inventor of Unix--and a number of major computer manufacturers.
Although originally Unix had been freely licensed by AT&T, it
appears that in the late 1980s AT&T began restricting its licenses
in the face of various software modifications being introduced by
individual licensees; and at the same time, AT&T began to develop
a close working relationship with Sun Microsystems, a major
microprocessor manufacturer. Other hardware manufacturers
professed to fear that AT&T was trying to establish a single
dominant version of Unix, intending to exclude the proprietary Unix
variations from the market.
Accordingly, in May 1988, a number of important computer
manufacturers--including defendants Hewlett-Packard and Digital
Equipment Corp.--formed the Open Software Foundation as a non-
profit joint research and development venture. OSF registered
under the National Cooperative Research Act of 1984, 15 U.S.C.
4301, although that status has no direct importance for the issues
before us. At least one of OSF's professed objectives was to
develop an alternative Unix operating system, denominated OSF-1, as
a competitor to the Unix system being developed jointly by AT&T and
Sun Microsystems.
In 1989, while OSF-1 was still being developed, OSF
decided that it should include security software at the B-1 level.
At that time, only three companies--AT&T, Addamax and SecureWare,
Inc.--were producing security software for the Unix system. On
November 1, 1989, OSF sent a "request for technology" to Addamax
and SecureWare, soliciting bids for a B-1 security component for
the new OSF-1 system. Bids were submitted on November 27, 1989,
and OSF selected SecureWare on December 22, 1989. There is some
indication that the Addamax security software was more
sophisticated--one witness agreed that the contrast was between a
Cadillac and a Chevette--but the Addamax price may also have
appeared more substantial. In any event, OSF-1 itself was never a
very successful product.
Addamax continued to sell its own B-1 software for some
period after losing the bid. Nothing prevented OSF "sponsors" (the
founding members of OSF) or "members" (a great many other
companies) from using Addamax security software for their own
programs; and OSF sponsors and members were not the only potential
buyers of Addamax's program. However, by 1991, Addamax began to
phase out its B-1 security software, turning away new buyers so
that it could devote its resources to the development of a new
security software product, in which it appears that the company was
successful.
In April 1991, Addamax filed a complaint in the district
court against OSF, Hewlett-Packard and Digital, alleging various
violations of federal and state antitrust law. As later amended,
the complaint charged the defendants, together with other companies
associated with OSF, with horizontal price fixing, boycott, and
otherwise unlawful joint venture behavior in violation of the
Sherman and the Clayton Acts, 15 U.S.C. 1-2, 18. A central
theme, although not the only one, was that the defendants had
conspired to force down the price for security software below the
free-market level and otherwise to limit or impair the ability of
Addamax to compete as a supplier of security software.
Considerable discovery was conducted, and in due course
the defendants moved for summary judgment. In a thoughtful
decision in May 1995, the district judge dismissed Addamax's per se
claims on the ground that the alleged conduct of the defendants did
not fit within the narrow categories for which per se treatment was
appropriate; but the court declined to dismiss Addamax's rule of
reason claims, saying that the factual issues involving market
power and anticompetitive effect were unsuitable for disposition on
summary judgment. Addamax Corporation v. Open Software Foundation,
Inc., 888 F. Supp. 274 (D. Mass. 1995).
Thereafter, the parties entered into a stipulation that
the damage phase of the case would be tried first, on a jury waived
basis, to determine "whether the defendants' conduct was a material
cause of injury in fact to the plaintiff and, if so, the amount of
damages." The stipulation further provided:
Solely for purposes of this stipulation, the
Court will assume that the defendants' conduct
as alleged in the Amended Complaint and
described in the non-damages portions of the
expert reports of Drs. Comanor and Howe
occurred and violates the federal and state
law accounts. However, the Court will not
assume, but will hear and take evidence on,
whether there was injury in fact to the
plaintiff as a result of that conduct and, if
so, the amount of damages, which is the
subject of this phase of the trial.
Trial was conducted over 12 days between November 18 and
December 16, 1996. Addamax presented as live witnesses Dr.
Alsberg, three experts, and a single Addamax customer. The
defendants did not present live witnesses but did cross-examine
extensively and relied on documentary submissions and deposition
testimony.
In May 1997, the district court issued a decision
concluding that the defendants' conduct was "not a material cause"
of Addamax's losses. Addamax Corporation v. Open Software
Foundation, Inc., 964 F. Supp. 549, 555 (D. Mass 1997). The court
found that the B-1 software market was a highly risky business,
that Addamax's belatedly-offered product was "too expensive and too
complex, and . . . actually exceeded B-1 requirements." Id. at
553. The court said that Addamax faced severe competition from
AT&T in the B-1 market and that SecureWare's product was "a cheaper
and simpler" one. Id. Accordingly, the court held on the merits
that Addamax was not entitled to any damages.
Addamax has now appealed and argues what in substance are
three different points: that the district court erred in its
factual determination that defendants' conduct was not a material
cause of Addamax's losses in the B-1 security software market; that
the stipulation, pertinent case law or both required the district
court to find that the defendants' conduct had caused damage to
Addamax; and that the court erred in dismissing the per se claims
against the defendants. Because the legal issues set the frame for
the factual ones, it is more convenient to take these claims in
reverse order.
1. As we explained in an earlier case, per se rules
under section 1 of the Sherman Act have left only a couple of
"serious candidates" for per se treatment: these include price or
output fixing agreements (horizontal market division agreements are
of essentially the same character) and "certain group boycotts or
concerted refusals to deal." U.S. Healthcare, Inc. v.
Healthsource, Inc., 986 F.2d 589, 593 (1st Cir. 1993). Since those
words were written, the categories have been narrowed even further
by the Supreme Court's decision to overrule Albrecht v. The Herald
Co., 390 U.S. 145 (1968) and thereby to exclude from per se
treatment vertical maximum resale price fixing agreements. State
Oil Co. v. Khan, 118 S. Ct. 275 (1997).
Where a plaintiff proves conduct that falls within a per
se category, nothing more is needed for liability; the defendants'
power, illicit purpose and anticompetitive effect are all said to
be irrelevant. United States v. Socony-Vacuum Oil Co., 310 U.S.
150 (1940). But courts have been very careful to confine per se
treatment to conduct of the type that is almost always actually or
potentially anticompetitive and has no redeeming benefits (e.g.,
reduced costs, increased competition) worthy of being weighed
against the negative effects. Broadcast Music, Inc. v. Columbia
Broadcasting System, 441 U.S. 1 (1979). Per se offenses remain
very important--they include horizontal price fixing--but only for
conduct that fits squarely within the "ever narrowing per se
niche." U.S. Healthcare, 986 F.2d at 593.
Joint venture enterprises like OFC, unless they amount to
complete shams, cf. Palmer v. BRG of Georgia, 498 U.S. 46 (1990),
are rarely susceptible to per se treatment. Where the venture is
producing a new product--here, the OFC-1 software package--there is
patently a potential for a productive contribution to the economy,
and conduct that is strictly ancillary to this productive effort
(e.g., the joint venture's decision as to the price at which it
will purchase inputs) is evaluated under the rule of reason. This
is so even if we accept, pursuant to the stipulation, the arguendopremise that OSF and those connected with it represented a large
portion of the market for purchasing B-1 security software and
represented a large portion of some kind of output market for
integrated Unix software programs.
Addamax points to fragments of evidence that, assuming a
full context were established, might or might not suggest that OSF
was an aggressive response to the AT&T Sun venture and that
Hewlett-Packard had a secret agenda to favor SecureWare over
Addamax (for reasons that are never made quite clear), regardless
of whether Addamax offered a superior product. None of the
evidence pointed to by Addamax suggests that OSF-1 was other than
a legitimate, if ultimately unsuccessful, product; and there is
nothing to suggest that the ancillary decisions--what inputs to
purchase, at what price, and from whom--were not legitimately
related to this effort. In this context, flinging around terms
like "cartel" and "boycott" do not convert a rule of reason claim
into a per se one.
2. On the other hand, neither is a joint venture "per
se" legal. Any joint venture, especially one that involves
competitors, tends to be susceptible to attack under section 1's
rule of reason--on the theory that the operations of the joint
venture represent collaboration of the separate entities that own
or control it. How far this theory can be pressed in the case of
a truly integrated enterprise, whose "owners" were no more than
stockholders, is a matter we need not pursue; we will assume here
that the OSF joint venture, or some aspect of it, could be
condemned under section 1 if the balance of harms and benefits
tipped in favor of harms; questions of power and motive are
primarily clues to such effects.
At this point, Addamax's most straightforward claim would
be that OSF's concentration of purchasing power in the supposed
"market" for acquiring B-1 security software was so great that it
imposed a significant risk of forcing prices below competitive
levels, and that those risks outweighed any benefit from the
venture or, more plausibly, that the venture could achieve those
benefits in a less restrictive fashion, i.e., without creating a
substantial threat of monopsony pricing. Whether or not this
theory could be proved, we are here assuming liability arguendo.
The question remains whether Addamax established--either as a
matter of law or based on the evidence--some causal connection
between this assumed violation by defendants and Addamax's failure
in the B-1 security software business.
Addamax first argues that the stipulation required the
district court to find that the assumed violation was a material
cause of injury to Addamax. Addamax's reading is contrary both to
the explicit language of the stipulation and to its evident
purpose. The stipulation said that the first phase of the bench
trial was to determine "whether" defendants' conduct was "a
material cause of injury in fact" to Addamax (and, if so, the
amount of damages), assuming arguendo that a violation of law had
occurred. Pointedly, the stipulation went on to emphasize that
although the existence of a violation was to be assumed, the court
"will not assume, but will hear and take evidence on," the question
whether there was "injury in fact to the plaintiff as a result of"
the attributed conduct.
Certainly both Addamax's complaint and its experts
asserted that the alleged conduct has caused injury, but this was
not part of the facts to be assumed arguendo. The stipulation did
not assume the truth of the complaint, but only that defendants had
engaged in the "conduct" alleged in the complaint. Under the
stipulation, the reports of Addamax's experts were assumed to be
true only in describing the conduct and not its consequences; that
is made clear by the reference in the stipulation to assuming the
truth of the "non damages portions of the expert reports."
Addamax's more interesting argument is its claim that the
case law, and the economic theory that underlies it, require a
conclusion that the conduct assumed arguendo to comprise a
violation must have caused injury to Addamax. The broadest version
of this proposition is Addamax's claim that under the rule of
reason, conduct is condemned only because it has an anticompetitive
effect. Therefore, Addamax argues, there must have been some
injury to it, and the only question that remains is to calculate
the amount of damages.
A more specific version of the argument, also advanced,
is that in this very case the complaint's straightforward charge is
that the defendants engaged in an agreement that had the effect of
reducing price for B-1 security systems and since Addamax was a
provider of B-1 security programs, it necessarily was injured by a
reduction in price. An alternative version is Addamax's claim that
the joint venturers were engaged in suppressing demand for their
own output--Unix programs like OSF-1 embodying B-1 software--and
this in turn reduced the demand, and presumably therefore price,
volume or both, for suppliers of the input.
It is technically an overstatement to say that actual
anticompetitive impact is a requirement of liability in a rule-of-
reason case. True, as a practical matter, most courts would be
unlikely to condemn an otherwise legitimate joint venture absent
some showing of anticompetitive effect. But in principle, a
sufficiently high risk of an anticompetitive effect, coupled with
marginal benefits (or none at all that could not be achieved
through an easily available less restrictive alternative) might
justify condemnation under the rule of reason.
But all this is beside the point. Even if we assume that
the OSF purchasing consortium was capable of exercising monopsony
power directly or through coordination of its sponsor\members'
actions, it does not follow that Addamax was a victim or that the
alleged below-market price offered by the consortium materially
affected Addamax. The only formal purchase by OSF involving
Addamax was based on the November 1989 request for technology, in
which SecureWare was the successful supplier. If below market
price was paid, SecureWare, and not Addamax, was directly injured.
To be sure, Addamax claims that its sales opportunities
were indirectly curtailed. While the OSF sponsors and members were
free to purchase B-1 security programs from anyone they wanted on
an individual basis, Addamax claims that winning the OSF-1 sale
would have amounted to a valuable OSF endorsement, spurring other
sales. But it is hard to see this loss as a consequence of
monopsony pricing. In all events, Addamax's claim of secondary
injury reduces itself to an issue of fact--not an issue that can be
taken as resolved by stipulation or case law.
3. As a preface to its factual claims, Addamax asserts
that the district court's decision recognizes a "defense" of
mismanagement. A plaintiff's mismanagement is not a bar to
recovery where an antitrust violation is a material cause of
injury. Engine Specialties, Inc. v. Bombardier Ltd., 605 F.2d at
14 n.21. But mismanagement is certainly relevant to the factual
question whether the antitrust violation had such an effect or
whether the plaintiff is entirely the cause of its own failure.
Argus Inc. v. Eastman Kodak Co., 801 F.2d 38, 42-45 (2d Cir. 1986).
We turn then to the question whether the district court
erred in its factual determination that Addamax's inability to
succeed in its efforts to sell its B-1 security program for Unix
more widely was materially caused by the defendants' conduct. On
this issue, Addamax bore the burden of proof at trial. Irvin
Industries, Inc. v. Goodyear Aerospace Corp., 974 F.2d 241, 244 (2d
Cir. 1992). On appeal, the district court's findings and ultimate
conclusion of fact are entitled to stand unless found to be clearly
erroneous. La Esperanza De P.R., Inc. v. Perez Y Cia. De Puerto
Rico, Inc., 124 F.3d 10, 15 (1st Cir. 1997).
Addamax is mistaken in its suggestion that the district
court misunderstood the legal standard for causation: quite unlike
Haverhill Gazette Co. v. Union Leader Corp., 333 F.2d 798 (1st Cir.
1964), the district judge properly asked whether the defendants'
assumed conduct had been a substantial or material cause of the
losses claimed by Addamax. Although we there cautioned against an
unduly rigid view of causation in computing damages once injury had
been established, id. at 808 n.16, nothing in the opinion relieved
the plaintiff from making the statutory showing (fully satisfied in
Haverhill) that the violation had caused at least some injury to
the plaintiff's business or property. 15 U.S.C. 15.
Here, the district court was presented with two competing
versions of reality. Addamax's witnesses took the view that
Addamax developed a superior B-1 product and its failure to succeed
resulted from defendants' mechanizations which forced down the
price of the product to sub-competitive levels and suppressed
output for Unix software incorporating B-1 security programs. In
some places, Addamax describes itself as the target, and elsewhere
as the accidental victim, of a larger conspiracy directed against
AT&T.
The defense version, which the district court adopted,
was derived from defense depositions, cross-examination of the
plaintiffs' experts and numerous documents. In this view, Addamax
engaged in risky entry into a market dominated by AT&T, an
established supplier of B-1 security software for Unix; the Addamax
system was oversophisticated, expensive, arrived late, and never
received the important certification from NSA's National Computer
Security Center. And, in a market characterized by ever-changing
demands, AT&T's promised development of a follow-on B-2 system
made the market for B-1 security software for Unix inherently risky
and in some measure transitional. There is nothing inherently
implausible about either version; everything depends on the
evidence.
In fairness to Addamax, the factual analysis in its brief
represents exactly the kind of detailed critique of the evidence
that fairly presents the question whether the district judge's
findings are adequately supported by the record. Addamax
identifies specific findings with which it disagrees, cites to
record evidence it thinks was misconstrued, furnishes citations to
the record for propositions it advances, points to counterveiling
evidence that says the district judge ignored and misinterpreted.
In this respect, the brief deserves to be taken seriously, and we
have done so.
But while in a few instances the district court may have
misinterpreted an exhibit or ignored some fact that softens or
qualifies the inference it draws, Addamax's own critique is filled
with one-sided versions of events and refusals to confront evidence
in support of the district court's findings. What emerges from our
own review of the record is that the district court had evidence to
support each of its key findings: that the business was a risky
one; that Addamax entered late, with a high-priced, overbuilt and
uncertified product; that AT&T and SecureWare, in different ways,
posed major problems for Addamax; that many of Addamax's problems,
including losses of customers, had begun before the OSF selection
of SecureWare; and that changes in market conditions proved to be
adverse to Addamax.
Further, the evidence is largely derived from Addamax
itself, including the deposition and trial testimony of Dr.
Alsberg. The risky nature of Addamax's venture was stressed in its
disclosures to investors, and the delays and cost overruns
concerning its B-1 product emerge from its own records; it was
Addamax that expressed concern with competition from AT&T and
especially its ability to offer a smooth transition to its own
promised B-2 offering; and the defection of existing and
prospective customers to AT&T, apparently before OSF-1, can be
traced through Addamax records. The district court did not commit
"clear error" in finding the facts in favor of defendants.
Addamax attacks the district court's findings as
inadequate under Fed. R. Civ. P. 52(a). By this, it means not that
the explicit findings were wrong (it argues this elsewhere) but
that the findings are either not sufficiently detailed or did not
affirmatively find specific facts helpful to Addamax or did not
address evidence that favored Addamax where there was evidence both
ways. But the district court was not required to make findings on
every detail, was not required to discuss all of the evidence that
supports each of the findings made, and was not required to respond
individually to each evidentiary or factual contention made by the
losing side. Knapp Shoes, Inc. v. Sylvania Mills Co., 15 F.3d
1222, 1228 (1st Cir. 1994); Applewood Landscape & Nursery Co. v.
Hollingsworth, 884 F.2d 1502, 1503-04 (1st Cir. 1989).
There is no mechanical rule for determining the exact
level of findings required by Rule 52(a). In this instance, the
district court made explicit intermediate findings (e.g.,
competition from AT&T, the strengths and weaknesses of Addamax's
product) making clear the bases for its ultimate finding on
causation; and, with the help of counsel, it has been relatively
easy to discern the evidence underlying these key findings. The
district court could have written a 200-page decision on this case,
but the far more compact assessment it made was entirely adequate
under Rule 52(a).
Affirmed.