Microsoft Corp. v. United States

United States Court of Appeals For the First Circuit No. 98-2133 IN RE: MICHAEL A. CUSUMANO AND DAVID B. YOFFIE [UNITED STATES OF AMERICA v. MICROSOFT CORPORATION]. MICROSOFT CORPORATION, Petitioner, Appellant. APPEAL FROM THE UNITED STATES DISTRICT COURT FOR THE DISTRICT OF MASSACHUSETTS [Hon. Richard G. Stearns, U.S. District Judge] Before Selya, Circuit Judge, Coffin and Bownes, Senior Circuit Judges. D. Stuart Meiklejohn, with whom John L. Warden, Richard J. Urowsky, Steven J. Holley, Michael E. Swartz, Hilary M. Williams, Sullivan & Cromwell, Thomas J. Sartory, Lynne Alix Morrison, and Goulston & Storrs, P.C. were on brief, for petitioner. Jeffrey Swope, with whom Palmer & Dodge LLP was on brief, for respondents Michael A. Cusumano and Massachusetts Institute of Technology. Jonathan M. Albano, with whom Shaun B. Spencer, Bingham Dana LLP, and Kimberly S. Budd, Office of the General Counsel, Harvard University, were on brief, for respondents David B. Yoffie and Harvard University. December 15, 1998 SELYA, Circuit Judge. In this appeal, petitioner- appellant Microsoft Corporation (Microsoft) invites us to reverse the district court's denial of its motion to compel production of research materials compiled by two academic investigators. Microsoft wants to use the subpoenaed materials in defending a civil antitrust case, United States v. Microsoft Corp., presently being tried in the United States District Court for the District of Columbia. Mindful that important First Amendment values are at stake, we decline Microsoft's invitation. I. THE ANTITRUST CASE We draw our description of the antitrust litigation in large part from the court in which that litigation pends. SeeUnited States v. Microsoft Corp., 1998 WL 614485 (D.D.C. 1998). Microsoft is one of the most profitable companies in the computer industry. It first attained a significant foothold in the production of operating systems for the personal computer (PC) market when a leading computer manufacturer, International Business Machines Corporation, chose Microsoft's "MS-DOS" operating system for its PCs in the early 1980s. An operating system is the "command center" of a PC. Microsoft, 1998 WL 614485, at *2. It facilitates the integrated use of hardware and software by controlling the interaction between a PC's processor, its memory, and devices like keyboards and disk drives. In relatively short order, Microsoft's operating systems achieved a preeminent market position. Microsoft continued to introduce new operating systems, including its phenomenally successful "Windows" systems, which allow a user to control a PC's operations by manipulating images on the computer screen with a mouse, rather than by typing commands. The dominance of Microsoft's operating systems has been maintained, in part, because of the symbiotic relationship that exists between software and operating systems. Software programs utilize certain general functions of operating systems and are written to work with particular systems. Since more PCs depend on Windows than on any rival, software creators tend to write products for use on that system. In turn, most PC users want this operating system for their PCs, so that they can access the widest possible range of software programs. To keep this lucrative circle spinning, Microsoft licenses its operating system to PC manufacturers for pre-installation on new computers. Microsoft's achievements in the operating systems market have encouraged it to spread its corporate wings. It now produces an internet browser product known as "Internet Explorer." Browsers are software programs that allow computer users to access, manipulate, and display portions of the world-wide web (the Web). The Web is a set of sites that employ graphics, text, and other media to provide information to viewers. Among other things, browsers can be used to translate these sites from the language of their creation into a format intelligible to a user's particular PC. A browser can be purchased individually, acquired as an accessory to a newly purchased PC, or downloaded from internet access providers or other Web sites. Microsoft's success has not gone unremarked. On May 18, 1998, the United States Department of Justice (DOJ) and several state attorneys general brought suit in the United States District Court for the District of Columbia, charging Microsoft with various antitrust violations. The complaint's main allegations center around Microsoft's accretion of market share for its Internet Explorer product. DOJ asserts that Microsoft, mindful that browsers potentially can be used as platforms on which to run software and thus replace, or at least compete with, operating systems, set out to increase its share of the browser market in a no-holds-barred campaign to safeguard its hegemony in the operating systems market. In January 1997, Navigator, a competing browser produced by Netscape Communications Corporation (Netscape), boasted an 80% share of the browser market. Explorer enjoyed less than 20%. DOJ charges that Microsoft first essayed to increase its market share by colluding with Netscape. When Netscape rebuffed Microsoft's overtures, DOJ alleges, Microsoft illegally "tied" Explorer to its Windows operating system refusing to grant computer manufacturers licenses to pre-install Windows for their customers unless the manufacturers agreed to pre-install Explorer and no other browser and thereby increased its share of the browser market to approximately 50% by May of 1998. Microsoft denies the government's accusations. It avers that it never tried to split the browser market between itself and its competitors or to use monopoly power in the operating systems market to capture a lion's share of the browser market in an illegal fashion. Instead, it attributes its increased share of the browser market to Explorer's superiority. The district court placed the antitrust litigation on a fast track. Among other things, the court established a tightly compressed schedule for pretrial discovery; shortened the usual time within which parties and non-parties alike might respond to discovery requests; directed that witness lists (to include no more than twelve trial witnesses per side, absent special permission) be submitted no later than August 24, 1998 (just over three months after the government sued); and closed discovery as of October 9, 1998 (save for discovery already underway). After that date, new discovery could be initiated only with leave of court. A bench trial commenced on October 19, 1998. That trial is ongoing. II. THE RULE 45 PROCEEDINGS In the course of pretrial discovery in the antitrust case, Microsoft learned about a forthcoming book entitled Competing on Internet Time: Lessons from Netscape and the Battle with Microsoft (Lessons) and obtained a copy of the manuscript. As its title implies, Lessons deals extensively with the "browser war" waged between Microsoft and Netscape. Its authors (respondents- appellees here) are distinguished academicians: Michael A. Cusumano, a tenured full professor at Massachusetts Institute of Technology's Sloan School of Management, and David B. Yoffie, a tenured full professor at Harvard Business School. As part of their research for Lessons, the respondents interviewed over 40 current and former Netscape employees. Their interview protocol dealt with confidentiality on two levels. First, the respondents signed a nondisclosure agreement with Netscape, in which they agreed not to disclose proprietary information conveyed to them in the course of their investigation except upon court order, and then only after giving Netscape notice and an opportunity to oppose disclosure. Second, the respondents requested and received permission from interview subjects to record their discussions, and, in return, promised that each interviewee would be shown any quotes attributed to him upon completion of the manuscript, so that he would have a chance to correct any errors or to object to quotations selected by the authors for publication. On September 18, 1998, believing that certain statements from Netscape employees reported in Lessons offered succor for its defense, Microsoft subpoenaed the professors' notes, tape recordings and transcripts of interviews, and correspondence with interview subjects. See Fed. R. Civ. P. 45. The respondents produced some correspondence, but declined to surrender the notes, tapes, or transcripts. Microsoft moved to compel the production of these items on October 1, 1998. Because the documents, if produced at all, would be produced in Cambridge, the subpoenas issued from the United States District Court for the District of Massachusetts and the motion to compel was docketed there as an independent proceeding. See id. (requiring a subpoena commanding only document production to "issue from the court for the district in which the production . . . is to be made" and allowing enforcement "pursuant to an order of th[at] court"). On October 7, the respondents filed their oppositions. Following a hearing held the next day, the district court denied the motion to compel ore tenus. The court performed a case- specific balancing analysis, taking into account a myriad of factors. On one hand, it found that Microsoft's need for the information sought by the subpoenas, though real, was not great. Microsoft could have obtained that information directly from the sources revealed by the manuscript, and, in all events, its main thrust likely would be for purposes akin to impeachment. On the other hand, the court found that the respondents had a substantial interest in keeping the subpoenaed information confidential and that significant First Amendment values favored its protection. Balancing these and other elements, the court declined to compel production of the notes, tapes, and transcripts. Withal, the court retained jurisdiction in order to review individual items in camerafor materiality on Microsoft's later motion and proclaimed its readiness to order specific material produced upon a showing of particularized need. Microsoft now appeals. III. APPELLATE JURISDICTION Federal courts, as courts of limited jurisdiction, may not presume the existence of subject matter jurisdiction, but, rather, must appraise their own authority to hear and determine particular cases. See Viqueira v. First Bank, 140 F.3d 12, 16 (1st Cir. 1998). We do so here. The procedural posture of this case is unusual. Microsoft, a litigant in a court in another district, moved in the District of Massachusetts to compel the respondents (who are not parties to that litigation) to honor subpoenas duces tecum. When the respondents objected, the district court obliged them, but retained jurisdiction to hear particularized claims of need. Meanwhile, the primary action is ongoing and any appeal therein will go, in the first instance, to the District of Columbia Circuit. Our customary appellate jurisdiction extends to "final decisions of the district courts." 28 U.S.C. 1291. Still, a decision or order can be final in the sense needed to confer appellate jurisdiction even if the proceeding in which it is rendered is ancillary to some other proceeding. Thus, in Horizons Titanium Corp. v. Norton Co., 290 F.2d 421 (1st Cir. 1961), a disappointed patent-seeker subpoenaed documents from a competitor for use in an action challenging the dismissal of its patent application. See id. at 421-22. The subpoenaed firm did not comply, and the patent-seeker moved to compel production in the United States District Court for the District of Massachusetts (although the principal dispute was pending elsewhere). See id. at 421. When the district court refused to order production of the documents, the patent-seeker appealed. In rejecting the subpoenaed firm's assertion that the court of appeals lacked jurisdiction, we explained that "the order of the district court made a final disposition of the only proceedings in its district growing out of a particular controversy, and the only proceeding pending between these particular parties anywhere," and, thus, was final and appealable. Id. at 424. So it is here. Nor does the district court's decision to retain jurisdiction detract from the immediate appealability of its order. It is settled law that a court's retention of jurisdiction in order to facilitate the consideration of possible future relief does not undermine the finality of an otherwise appealable order. See FTCv. Standard Fin. Mgmt. Corp., 830 F.2d 404, 407 (1st Cir. 1987); FTC v. Texaco, Inc., 555 F.2d 862, 873 n.21 (D.C. Cir. 1977); seegenerally 15B Charles Alan Wright, et al., Federal Practice and Procedure 3915.3 (2d ed. 1992) (explaining that an order's vulnerability to possible change through subsequent proceedings does not automatically deprive it of finality). Consequently, we have jurisdiction to hear and determine this appeal. IV. THE MERITS Microsoft argues that the district court underestimated its need for the subpoenaed information because that information would be useful not only for impeachment purposes, but also as independent evidence of Netscape's business miscalculations. SeeFed. R. Evid. 807; see also United States v. American Tel. & Tel. Co., 516 F. Supp. 1237, 1239-42 (D.D.C. 1981) (admitting into evidence in an antitrust prosecution documents authored by agents of defendants' non-party competitors under the residual hearsay exception). Moreover, it had no other feasible way to obtain the information since the accelerated schedule in the antitrust case effectively thwarted direct discovery. Turning to the other side of the balance, Microsoft claims that the district court erred in affording substantial protection to the subpoenaed materials because those materials were not confidential and emanated from disclosed sources. In this regard, Microsoft tells us that the only protectable data obtained by the respondents is proprietary information covered by the nondisclosure agreement an agreement signed with Netscape, not with the individual interviewees. Microsoft further asserts that the interviews were not confidential because the authors presented the manuscript, including quotes from interviews, to Netscape executives and outside reviewers prior to showing it to the persons quoted. The respondents dispute virtually all of Microsoft's assertions. They maintain that Microsoft does not really need the information at all and that it can secure the same data through other, less intrusive avenues. Furthermore, the respondents asseverate that they are sufficiently like journalists for the protection afforded to journalists' materials to be applied here; that the information which they procured is confidential because of their interview protocol; and that forcing them to disclose the contents of the notes, tapes, and transcripts would endanger the values of academic freedom safeguarded by the First Amendment and jeopardize the future information-gathering activities of academic researchers. A. Standard of Review. To resolve these issues, we focus first on the applicable standard of review. Discovery orders ordinarily are reviewed for abuse of discretion. See Dykes v. DePuy, Inc., 140 F.3d 31, 36-37 (1st Cir. 1998); Mack v. Great Atl. & Pac. Tea Co., 871 F.2d 179, 186-87 (1st Cir. 1989). Microsoft seeks to bend this rule and obtain plenary review. It proffers two reasons. Microsoft's first ground is barren. It points out that the district court did not take testimony, but, rather, decided the case solely on a paper record, assisted by arguments of counsel. Thus, Microsoft contends, there is no occasion to defer since this court is equally well-equipped to evaluate the corpus of evidence upon which the district court based its decision. This line of argument is neither original nor persuasive. It has been tried before, and regularly rejected. See, e.g., United States Liab. Ins. Co. v. Selman, 70 F.3d 684, 688 (1st Cir. 1995); In re Tully, 818 F.2d 106, 108-10 (1st Cir. 1987). We, too, dismiss it. Microsoft's second ground holds out more promise as a theoretical matter. It asserts that the lower court applied an incorrect legal standard and ample authority supports the proposition that, whatever the procedural context, pure questions of law warrant de novo review. See, e.g., McCarthy v. Azure, 22 F.3d 351, 354 (1st Cir. 1994); In re Howard, 996 F.2d 1320, 1327 (1st Cir. 1993). Here, however, Microsoft's argument fails because, as we explain below, the district court's test, which balanced the need for disclosure against the desirability of confidentiality, is correct as a matter of law. Refined to bare essence, Microsoft's quarrel with the district court's determination concerns the manner in which the court struck the required balance, rather than its selection of a legal rule. Since abuse of discretion is the precise rubric under which an appellate court should review a district court's application of a legal rule to the facts it has found, our review proceeds accordingly. B. The Analytic Approach. The discovery rules apply to subpoenas issued under Fed. R. Civ. P. 45. See 9A Wright et al., supra, 2452. Thus, we start with Fed. R. Civ. P. 26(b)(2), which admonishes generally that: the . . . extent of the use of the discovery methods otherwise permitted under these rules . . . shall be limited by the court if it determines that: (i) the discovery sought is unreasonably cumulative or duplicative, or is obtainable from some other source that is more convenient, less burdensome, or less expensive; (ii) the party seeking discovery has had ample opportunity by discovery in the action to obtain the information sought; or (iii) the burden or expense of the proposed discovery outweighs its likely benefit, taking into account the needs of the case, the amount in controversy, the parties' resources, the importance of the issues at stake in the litigation, and the importance of the proposed discovery in resolving the issues. This admonition forms the backdrop against which a court must consider whether to enforce a subpoena duces tecum issued as an instrument of discovery in a civil case. From that point forward, our study of the merits of this appeal must proceed in steps. Initially, we must determine whether the respondents' academic research is protected in a manner similar to the work product of journalists. This inquiry entails two aspects: (1) whether the respondents' positions warrant conferral of any special consideration, and (2) whether their research comprises confidential information. If these hurdles are cleared, we next must determine the type and kind of protection that the law affords. Finally, we must assess the district court's application of the law to the facts, and the appropriateness of its order. C. The Availability of Protection. 1. Who Is Protected? Microsoft acknowledges that the law supplies a measure of protection for materials compiled by journalists. See Bruno & Stillman, Inc. v. Globe Newspaper Co., 633 F.2d 583, 595-98 (1st Cir. 1980). The respondents, however, are academic researchers and commentators, not professional newsmen. We do not think that this makes a dispositive difference in whether special protection vests. Academicians engaged in pre- publication research should be accorded protection commensurate to that which the law provides for journalists. Courts afford journalists a measure of protection from discovery initiatives in order not to undermine their ability to gather and disseminate information. See United States v. LaRouche Campaign, 841 F.2d 1176, 1181 (1st Cir. 1988). Journalists are the personification of a free press, and to withhold such protection would invite a "chilling effect on speech," id., and thus destabilize the First Amendment. The same concerns suggest that courts ought to offer similar protection to academicians engaged in scholarly research. After all, scholars too are information gatherers and disseminators. If their research materials were freely subject to subpoena, their sources likely would refuse to confide in them. As with reporters, a drying-up of sources would sharply curtail the information available to academic researchers and thus would restrict their output. Just as a journalist, stripped of sources, would write fewer, less incisive articles, an academician, stripped of sources, would be able to provide fewer, less cogent analyses. Such similarities of concern and function militate in favor of a similar level of protection for journalists and academic researchers. Given this mise-en-scne, it is unsurprising that several of our sister circuits have held that the medium an individual uses to provide his investigative reporting to the public does not make a dispositive difference in the degree of protection accorded to his work. See In re Madden, 151 F.3d 125, 128-31 (3d Cir. 1998); Shoen v. Shoen, 5 F.3d 1289, 1293-94 (9th Cir. 1993); von Bulow v. von Bulow, 811 F.2d 136, 142-44 (2d Cir. 1987). Whether the creator of the materials is a member of the media or of the academy, the courts will make a measure of protection available to him as long as he intended "at the inception of the newsgathering process" to use the fruits of his research "to disseminate information to the public." von Bulow, 811 F.2d at 144. This case fits neatly into the architecture of these precedents. The sole purpose of the respondents' interviews of Netscape personnel was to gather data so that they could compile, analyze, and report their findings anent management practices in the internet technology industry. Thus, the respondents are within a group whose pre-publication research merits a modicum of protection. 2. What Is Protected? This aspect of the question raises vexing theoretical issues. Leaving confidential sources to one side, the prototypical situation in which a court provides protection from disclosure for a journalist's or researcher's materials involves confidential information. See, e.g, United States v. Cuthbertson, 630 F.2d 139, 146-48 (3d Cir. 1980). When the information cannot fairly be characterized as confidential, the courts are divided as to whether any protection is warranted. Compare Shoen, 5 F.3d at 1295-96 (holding that materials should be afforded protection even if the information contained therein is not confidential) with Gonzales v. National Broad. Co., 155 F.3d 618, 626 (2d Cir. 1998) (disavowing protection for non-confidential information). Although we have not ruled definitively on this issue, we have noted, in a situation involving only nonconfidential information, "a lurking and subtle threat to journalists and their employers if disclosure of outtakes, notes, and other unused information, even if nonconfidential, becomes routine and casually, if not cavalierly, compelled." LaRouche, 841 F.2d at 1182. We perceive no need to resolve this unanswered question today. The district court found that the information sought by Microsoft was confidential in character, and the record sufficiently supports this finding. To be sure, confidentiality comes in varying shapes and sizes. The confidentiality agreed upon by the authors and the interviewees in this instance does not ensure the most perfect privacy. The respondents did not offer the interviewees the sort of detailed nondisclosure agreement that they signed with Netscape. Moreover, they conducted the interviews in the presence of a Netscape official, gave Netscape's management a preview of planned quotations prior to showing those statements to the persons who made them, and circulated their manuscript (or portions of it) for pre-publication peer review. Still, determinations of where particular disclosures fall along the continuum of confidentiality, and related determinations anent the degree of protection that attaches to them, must take into account the totality of the circumstances. When the respondents began their inquiry into Netscape's management practices early in 1997, neither they nor their interview subjects knew (or had any reason to believe) that Microsoft later would be sued for antitrust violations based upon its peregrinations in the browser market. In that environment, handing individual interview subjects complex legal agreements might have made them squeamish (and thus less candid). Instead, the respondents gave each interviewee a personal, albeit verbal, assurance that he would be accorded the opportunity to correct, comment upon, and/or disclaim attributed quotations prior to publication. At the very least, this assurance is a species of confidentiality. Microsoft's demand for the full tapes (including outtakes) and transcripts of interviews with all Netscape employees materials which have not to date been shown to anyone obviates this assurance. While the level of confidentiality that characterizes a journalist's or researcher's confidential information may, in the end, affect the degree of protection conferred upon that information in a discovery dispute, we agree with the district court that the interviews here fall along the continuum of confidentiality at a point sufficient to justify significant protection. D. The Degree of Protection. We turn now to the degree of protection that attaches to the respondents' materials. We begin with Bruno & Stillman, an opinion that had its genesis in a libel suit brought by a commercial boatbuilder against the Boston Globe after one Coughlin, a Globe reporter, wrote an unflattering article. See Bruno & Stillman, 633 F.2d at 584-85. During discovery, the defendants refused to turn over portions of Coughlin's notes, which named confidential sources and outlined information that the sources had supplied. See id. at 585. The district court compelled disclosure of the information, based on a finding that the information was critical to Bruno & Stillman's non-frivolous claim and was unavailable from other sources. See id. at 586. On appeal, without deciding the "semantic[]" question of whether the protection afforded to a journalist's sources and research is a type of privilege, id. at 595, Judge Coffin explained the necessity for attention to First Amendment concerns in such situations, seeid. at 595-96. The opinion instructed district courts that when faced with enforcing requests for the discovery of materials used in the preparation of journalistic reports [they] should be aware of the possibility that the unlimited or unthinking allowance of such requests will impinge upon First Amendment rights. In determining what, if any, limits should accordingly be placed upon the granting of such requests, courts must balance the potential harm to the free flow of information that might result against the asserted need for the requested information. Id. Eight years later, the court reiterated this instruction. SeeLaRouche, 841 F.2d at 1181. We, too, decline to spend our energies on semantics. It suffices to say that, when a subpoena seeks divulgement of confidential information compiled by a journalist or academic researcher in anticipation of publication, courts must apply a balancing test. This test contemplates consideration of a myriad of factors, often uniquely drawn out of the factual circumstances of the particular case. Each party comes to this test holding a burden. See Bruno & Stillman, 633 F.2d at 597. Initially, the movant must make a prima facie showing that his claim of need and relevance is not frivolous. See id. Upon such a showing, the burden shifts to the objector to demonstrate the basis for withholding the information. See id. The court then must place those factors that relate to the movant's need for the information on one pan of the scales and those that reflect the objector's interest in confidentiality and the potential injury to the free flow of information that disclosure portends on the opposite pan. See id. at 597-98. E. Calibrating the Scales. We now examine the district court's handiwork in light of this legal framework. Microsoft's need admittedly is substantial in the sense that relevant information likely exists indeed, the district court specifically found that Microsoft had not embarked on a fishing expedition and Microsoft has a legitimate use for it. The company, after all, is in the throes of defending a complex case of extraordinary importance to its future, and its primary defense is that Netscape suffered a series of self- inflicted wounds that dissipated its dominant position in the browser market. Lessons includes several quotations that suggest missteps by Netscape management during the browser war, and it is reasonable to assume that the notes, tapes, and transcripts include more evidence of this genre. Hence, Microsoft has made a prima facie showing of need and relevance. The district court discounted this showing somewhat because it found that the same information was otherwise available to Microsoft by direct discovery. We view this finding of fact deferentially, see Ross-Simons of Warwick, Inc. v. Baccarat, Inc., 102 F.3d 12, 16, 19-20 (1st Cir. 1996), and decline to disturb it. Despite the accelerated trial schedule in the antitrust case, Microsoft had the Lessons manuscript while discovery was still open, and the quotations in the book are attributed to named individuals. Microsoft which deployed no fewer than eight lawyers in the preparation of its brief in this appeal had enough time, enough knowledge, and enough resources to depose those individuals (or some subset of them), or otherwise obtain discovery from them. This factor must figure in the balance. See Haworth, Inc. v. Herman Miller, Inc., 998 F.2d 975, 978 (Fed. Cir. 1993). The opposite pan of the scale is brim-full. Scholars studying management practices depend upon the voluntary revelations of industry insiders to develop the factual infrastructure upon which theoretical conclusions and practical predictions may rest. These insiders often lack enthusiasm for divulging their management styles and business strategies to academics, who may in turn reveal that information to the public. Yet, pathbreaking work in management science requires gathering data from those companies and individuals operating in the most highly competitive fields of industry, and it is in these cutting-edge areas that the respondents concentrate their efforts. Their time-tested interview protocol, including the execution of a nondisclosure agreement with the corporate entity being studied and the furnishing of personal assurances of confidentiality to the persons being interviewed, gives chary corporate executives a sense of security that greatly facilitates the achievement of agreements to cooperate. Thus, in the Bruno & Stillman taxonomy, the interviews are "carefully bargained-for" communications which deserve significant protection. Bruno & Stillman, 633 F.2d at 597. Considering these facts, it seems reasonable to conclude as the respondents' affidavits assert that allowing Microsoft to obtain the notes, tapes, and transcripts it covets would hamstring not only the respondents' future research efforts but also those of other similarly situated scholars. This loss of theoretical insight into the business world is of concern in and of itself. Even more important, compelling the disclosure of such research materials would infrigidate the free flow of information to the public, thus denigrating a fundamental First Amendment value. It is also noteworthy that the respondents are strangers to the antitrust litigation; insofar as the record reflects, they have no dog in that fight. Although discovery is by definition invasive, parties to a law suit must accept its travails as a natural concomitant of modern civil litigation. Non-parties have a different set of expectations. Accordingly, concern for the unwanted burden thrust upon non-parties is a factor entitled to special weight in evaluating the balance of competing needs. SeeHaworth, 998 F.2d at 978; Dart Indus. Co. v. Westwood Chem. Co., 649 F.2d 646, 649 (9th Cir. 1980); Addamax Corp. v. Open Software Found., Inc., 148 F.R.D. 462, 468 (D. Mass. 1993). We need go no further. The district court used the proper test, balanced the right array of factors, and acted well within its discretion in determining that the scales tipped in favor of preserving confidentiality and against the wholesale disclosure of investigative materials gleaned in the course of pre- publication academic research. Our confidence in the appropriateness of this ruling is fortified by the fact that the district court took pains to protect Microsoft's legitimate interests. After denying the motion to compel, the court announced that it would retain jurisdiction so that, should a material conflict develop between quotations from the book and other evidence, it could review the notes, tapes, and transcripts incamera for purposes of verification and, if necessary, order production. This even-handed ruling treats all parties fairly. We discern no error. Affirmed.