Appeal from the United States District Courtfor the Northern District of CaliforniaCharles A. Legge, District Judge, PresidingArgued and SubmittedSeptember 14, 1999--San Francisco, CaliforniaFiled February 10, 2000Before: Herbert Y. C. Choy, William C. Canby, Jr. andBarry G. Silverman, Circuit Judges.Opinion by Judge Canby
_____________________________COUNSEL William S. Coats, III, Howrey & Simon, Menlo Park, Califor-nia, for the defendant-appellant.Ezra Hendon, Crosby, Heafey, Roach & May, Oakland, Cali-fornia; James G. Gilliland, Jr., Townsend and Townsend andCrew, San Francisco, California, for the plaintiffs-appellees.Annette L. Hurst, Howard, Rice, Nemerovski, Canady, Falk& Rabkin, San Francisco, California, for amicus Institute forElectrical and Electronics Engineers - USA. Peter M. C.Choy, American Committee for Interoperable Systems, PaloAlto, California, for amici American Committee forInteroperable Systems and Computer & CommunicationsIndustry Association. Mark Lemley, University of Texas atAustin, Austin, Texas, for amicus Law Professors. Steven J.Metalitz, Smith & Metalitz, Washington, D.C., for amici Nin-tendo of America, Inc., Sega of America, Inc., and 3dfx Inter-active.
_____________________________OPINION CANBY, Circuit Judge:In this case we are called upon once again to apply the prin-ciples of copyright law to computers and their software, todetermine what must be protected as expression and whatmust be made accessible to the public as function. Sony Com-puter Entertainment, Inc., which brought this copyrightinfringement action, produces and markets the Sony PlaySta-tion console, a small computer with hand controls that con-nects to a television console and plays games that are insertedinto the PlayStation on compact discs (CDs). Sony owns thecopyright on the basic input-output system or BIOS, which isthe software program that operates its PlayStation. Sony hasasserted no patent rights in this proceeding.The defendant is the Connectix Corporation, which makesand sells a software program called "Virtual Game Station."The purpose of the Virtual Game Station is to emulate on aregular computer the functioning of the Sony PlayStation con-sole, so that computer owners who buy the Virtual Game Sta-tion software can play Sony PlayStation games on theircomputers. The Virtual Game Station does not contain any ofSony's copyrighted material. In the process of producing theVirtual Game Station, however, Connectix repeatedly copiedSony's copyrighted BIOS during a process of "reverseengineering" that Connectix conducted in order to find outhow the Sony PlayStation worked. Sony claimed infringementand sought a preliminary injunction. The district court con-cluded that Sony was likely to succeed on its infringementclaim because Connectix's "intermediate copying " was not aprotected "fair use" under 17 U.S.C. S 107. The district courtenjoined Connectix from selling the Virtual Game Station orfrom copying or using the Sony BIOS code in the develop-ment of other Virtual Game Station products.Connectix now appeals. We reverse and remand withinstructions to dissolve the injunction. The intermediate cop-ies made and used by Connectix during the course of itsreverse engineering of the Sony BIOS were protected fair use,necessary to permit Connectix to make its non-infringing Vir-tual Game Station function with PlayStation games. Anyother intermediate copies made by Connectix do not supportinjunctive relief, even if those copies were infringing.The district court also found that Sony is likely to prevailon its claim that Connectix's sale of the Virtual Game Stationprogram tarnishes the Sony PlayStation mark under 15 U.S.C.S 1125. We reverse that ruling as well.I. BackgroundA. The productsSony is the developer, manufacturer and distributor of boththe Sony PlayStation and Sony PlayStation games. Sony alsolicenses other companies to make games that can play on thePlayStation. The PlayStation system consists of a console(essentially a mini-computer), controllers, and software thatproduce a three-dimensional game for play on a television set.The PlayStation games are CDs that load into the top of theconsole. The PlayStation console contains both (1) hardwarecomponents and (2) software known as firmware that is writ-ten onto a read-only memory (ROM) chip. The firmware isthe Sony BIOS. Sony has a copyright on the BIOS. It hasclaimed no patent relevant to this proceeding on any compo-nent of the PlayStation. PlayStation is a registered trademarkof Sony.Connectix's Virtual Game Station is software that"emulates" the functioning of the PlayStation console. That is,a consumer can load the Virtual Game Station software ontoa computer, load a PlayStation game into the computer's CD-ROM drive, and play the PlayStation game. The VirtualGame Station software thus emulates both the hardware andfirmware components of the Sony console. The Virtual GameStation does not play PlayStation games as well as Sony'sPlayStation does. At the time of the injunction, Connectix hadmarketed its Virtual Game Station for Macintosh computersystems but had not yet completed Virtual Game Station soft-ware for Windows.B. Reverse engineeringCopyrighted software ordinarily contains both copyrightedand unprotected or functional elements. Sega Enters. Ltd. v.Accolade, Inc., 977 F.2d 1510, 1520 (9th Cir. 1993) (amendedopinion); see 17 U.S.C. S 102(b) (Copyright protection doesnot extend to any "idea, procedure, process, system, methodof operation, concept, principle, or discovery" embodied inthe copyrighted work.). Software engineers designing a prod-uct that must be compatible with a copyrighted product fre-quently must "reverse engineer" the copyrighted product togain access to the functional elements of the copyrightedproduct. See Andrew Johnson-Laird, Software Reverse Engi-neering in the Real World, 19 U. Dayton L. Rev. 843, 845-46(1994).Reverse engineering encompasses several methods of gain-ing access to the functional elements of a software program.They include: (1) reading about the program; (2) observing"the program in operation by using it on a computer;"(3) performing a "static examination of the individual com-puter instructions contained within the program; " and(4) performing a "dynamic examination of the individualcomputer instructions as the program is being run on acomputer." Id. at 846. Method (1) is the least effective,because individual software manuals often misdescribe thereal product. See id. It would be particularly ineffective in thiscase because Sony does not make such information availableabout its PlayStation. Methods (2), (3), and (4) require thatthe person seeking access load the target program on to acomputer, an operation that necessarily involves copying thecopyrighted program into the computer's random accessmemory or RAM.1Method (2), observation of a program, can take severalforms. The functional elements of some software programs,for example word processing programs, spreadsheets, andvideo game displays may be discernible by observation of thecomputer screen. See Sega, 977 F.2d at 1520. Of course, thereverse engineer in such a situation is not observing the objectcode itself,2 only the external visual expression of this code'soperation on the computer. Here, the software program is cop-ied each time the engineer boots up the computer, and thecomputer copies the program into RAM.Other forms of observation are more intrusive. Operationssystems, system interface procedures, and other programs likethe Sony BIOS are not visible to the user when they are oper-ating. See id. One method of "observing" the operation ofthese programs is to run the program in an emulated environ-ment. In the case of the Sony BIOS, this meant operating theBIOS on a computer with software that simulated the opera-tion of the PlayStation hardware; operation of the program, inconjunction with another program known as a "debugger,"permitted the engineers to observe the signals sent betweenthe BIOS and other programs on the computer. This lattermethod required copying the Sony BIOS from a chip in thePlayStation onto the computer. The Sony BIOS was copiedagain each time the engineers booted up their computer andthe computer copied the program into RAM. All of this copy-ing was intermediate; that is, none of the Sony copyrightedmaterial was copied into, or appeared in, Connectix's finalproduct, the Virtual Game Station.Methods (3) and (4) constitute "disassembly" of objectcode into source code.3 In each case, engineers use a programknown as a "dissassembler" to translate the ones and zeros ofbinary machine-readable object code into the words and math-ematical symbols of source code. This translated source codeis similar to the source code used originally to create theobject code4 but lacks the annotations drafted by the authorsof the program that help explain the functioning of the sourcecode. In a static examination of the computer instructions,method (3), the engineer disassembles the object code of allor part of the program. The program must generally be copiedone or more times to perform disassembly. In a dynamicexamination of the computer instructions, method (4), theengineer uses the disassembler program to disassemble partsof the program, one instruction at a time, while the programis running. This method also requires copying the programand, depending on the number of times this operation is per-formed, may require additional copying of the program intoRAM every time the computer is booted up.C. Connectix's reverse engineering of the Sony BIOSConnectix began developing the Virtual Game Station forMacintosh on about July 1, 1998. In order to develop a Play-Station emulator, Connectix needed to emulate both the Play-Station hardware and the firmware (the Sony BIOS).Connectix first decided to emulate the PlayStation's hard-ware. In order to do so, Connectix engineers purchased aSony PlayStation console and extracted the Sony BIOS froma chip inside the console. Connectix engineers then copied theSony BIOS into the RAM of their computers and observed thefunctioning of the Sony BIOS in conjunction with the VirtualGame Station hardware emulation software as that hardwareemulation software was being developed by Connectix. Theengineers observed the operation of the Sony BIOS throughuse of a debugging program that permitted the engineers toobserve the signals sent between the BIOS and the hardwareemulation software. During this process, Connectix engineersmade additional copies of the Sony BIOS every time theybooted up their computer and the Sony BIOS was loaded intoRAM.Once they had developed the hardware emulation software,Connectix engineers also used the Sony BIOS to "debug" theemulation software. In doing so, they repeatedly copied anddisassembled discrete portions of the Sony BIOS.Connectix also used the Sony BIOS to begin developmentof the Virtual Game Station for Windows. Specifically, theymade daily copies to RAM of the Sony BIOS and used theSony BIOS to develop certain Windows-specific systems forthe Virtual Game Station for Windows. Although Connectixhad its own BIOS at the time, Connectix engineers used theSony BIOS because it contained CD-ROM code that the Con-nectix BIOS did not contain.Early in the development process, Connectix engineerAaron Giles disassembled a copy of the entire Sony BIOS thathe had downloaded from the Internet. He did so for the pur-pose of testing a "disassembler" program he had written. Theprint-out of the source code was not used to develop the Vir-tual Game Station emulator. Connectix engineers initiallyused this copy of the Sony BIOS to begin the reverse engi-neering process, but abandoned it after realizing that it was aJapanese-language version.During development of the Virtual Game Station, Connec-tix contacted Sony and requested "technical assistance" fromSony to complete the development of the Virtual Game Sta-tion. Connectix and Sony representatives met during Septem-ber 1998. Sony declined Connectix's request for assistance.Connectix completed Virtual Game Station for Macintoshcomputers in late December 1998 or early January 1999. Con-nectix announced its new product at the MacWorld Expo onJanuary 5, 1999. At MacWorld, Connectix marketed the Vir-tual Game Station as a "PlayStation emulator. " The materialsstated that the Virtual Game Station permits users to play"their favorite Playstation games" on a computer "even if youdon't yet have a Sony PlayStation console."D. Procedural historyOn January 27, 1999, Sony filed a complaint alleging copy-right infringement and other causes of action against Connec-tix. Sony subsequently moved for a preliminary injunction onthe grounds of copyright and trademark infringement. Thedistrict court granted the motion, enjoining Connectix:(1) from copying or using the Sony BIOS code in the devel-opment of the Virtual Game Station for Windows; and(2) from selling the Virtual Game Station for Macintosh orthe Virtual Game Station for Windows. Order on Mot. forPrelim. Inj. at 27. The district court also impounded all Con-nectix's copies of the Sony BIOS and all copies of worksbased upon or incorporating Sony BIOS. Id. at 27-28. Con-nectix now appeals from this order.II. Discussion[1] To prevail on its motion for injunctive relief, Sony wasrequired to demonstrate "either a likelihood of success on themerits and the possibility of irreparable injury or that seriousquestions going to the merits were raised and the balance ofthe hardships tip sharply in its favor." Cadence Design Sys.,Inc. v. Avant! Corp., 125 F.3d 824, 826 (9th Cir. 1997) (inter-nal quotation marks and bracket omitted), cert. denied, 523U.S. 1118 (1998). We reverse the grant of a preliminaryinjunction only when "the district court abused its discretionor based its decision on an erroneous legal standard or onclearly erroneous findings of fact." Roe v. Anderson, 134 F.3d1400, 1402 n.1 (9th Cir. 1998) (internal quotation marks omit-ted), aff'd on other grounds, sub nom. Saenz v. Roe, 526 U.S.489 (1999). We review the scope of injunctive relief for anabuse of discretion. SEC v. Interlink Data Network of L.A.,Inc., 77 F.3d 1201, 1204 (9th Cir. 1996).Connectix admits that it copied Sony's copyrighted BIOSsoftware in developing the Virtual Game Station but contendsthat doing so was protected as a fair use under 17 U.S.C.S 107. Connectix also challenges the district court's conclu-sion that Sony has established a likelihood that Connectix'sVirtual Game Station tarnishes the PlayStation trademark. Weconsider each of these claims below.A. Fair useThe fair use issue arises in the present context because ofcertain characteristics of computer software. The object codeof a program may be copyrighted as expression, 17 U.S.C.S 102(a), but it also contains ideas and performs functions thatare not entitled to copyright protection. See 17 U.S.C.S 102(b). Object code cannot, however, be read by humans.The unprotected ideas and functions of the code therefore arefrequently undiscoverable in the absence of investigation andtranslation that may require copying the copyrighted material.We conclude that, under the facts of this case and our prece-dent, Connectix's intermediate copying and use of Sony'scopyrighted BIOS was a fair use for the purpose of gainingaccess to the unprotected elements of Sony's software.[2] The general framework for analysis of fair use is estab-lished by statute, 17 U.S.C. S 107.5 We have applied this stat-ute and the fair use doctrine to the disassembly of computersoftware in the case of Sega Enterprises Ltd. v. Accolade,Inc., 977 F.2d 1510 (9th Cir. 1993) (amended opinion). Cen-tral to our decision today is the rule set forth in Sega: [W]here disassembly is the only way to gain access to the ideas and functional elements embodied in a copyrighted computer program and where there is a legitimate reason for seeking such access, disassem- bly is a fair use of the copyrighted work, as a matter of law.Id. at 1527-28 (emphasis added). In Sega , we recognized thatintermediate copying could constitute copyright infringementeven when the end product did not itself contain copyrightedmaterial. Id. at 1518-19. But this copying nonetheless couldbe protected as a fair use if it was "necessary " to gain accessto the functional elements of the software itself. Id. at 1524-26. We drew this distinction because the Copyright Act pro-tects expression only, not ideas or the functional aspects of asoftware program. See id. at 1524 (citing 17 U.S.C. S 102(b)).We also recognized that, in the case of computer programs,this idea/expression distinction poses "unique problems"because computer programs are "in essence, utilitarian articles-- articles that accomplish tasks. As such, they contain manylogical, structural, and visual display elements that are dic-tated by the function to be performed, by considerations ofefficiency, or by external factors such as compatibilityrequirements and industry demands." Id. Thus, the fair usedoctrine preserves public access to the ideas and functionalelements embedded in copyrighted computer software pro-grams. This approach is consistent with the " `ultimate aim[of the Copyright Act], to stimulate artistic creativity for thegeneral public good.' " Sony Corp. of Am. v. Universal CityStudios, Inc.,
464 U.S. 417, 432
(1984) (quoting TwentiethCentury Music Corp. v. Aiken,
422 U.S. 151, 156
(1975)).We turn then to the statutory fair use factors, as informedby our precedent in Sega. 1. Nature of the copyrighted work[3] Under our analysis of the second statutory factor, natureof the copyrighted work, we recognize that "some works arecloser to the core of intended copyright protection thanothers." Campbell v. Acuff-Rose Music, Inc. ,
510 U.S. 569
,586 (1994). Sony's BIOS lies at a distance from the corebecause it contains unprotected aspects that cannot be exam-ined without copying. See Sega, 977 F.2d at 1526. We conse-quently accord it a "lower degree of protection than moretraditional literary works." Id. As we have applied this stan-dard, Connectix's copying of the Sony BIOS must have been"necessary" to have been fair use. See id. at 1524-26. We con-clude that it was.[4] There is no question that the Sony BIOS containsunprotected functional elements. Nor is it disputed that Con-nectix could not gain access to these unprotected functionalelements without copying the Sony BIOS. Sony admits thatlittle technical information about the functionality of the SonyBIOS is publicly available. The Sony BIOS is an internaloperating system that does not produce a screen display toreflect its functioning. Consequently, if Connectix was to gainaccess to the functional elements of the Sony BIOS it had tobe through a form of reverse engineering that required copy-ing the Sony BIOS onto a computer.6 Sony does not disputethis proposition.[5] The question then becomes whether the methods bywhich Connectix reverse-engineered the Sony BIOS werenecessary to gain access to the unprotected functional ele-ments within the program. We conclude that they were. Con-nectix employed several methods of reverse engineering(observation and observation with partial disassembly) eachof which required Connectix to make intermediate copies ofcopyrighted material. Neither of these methods renders fairuse protection inapplicable. Sega expressly sanctioned disas-sembly. See id. at 1527-28. We see no reason to distinguishobservation of copyrighted software in an emulated computerenvironment. Both methods require the reverse engineer tocopy protected as well as unprotected elements of the com-puter program. Because this intermediate copying is the gra-vamen of the intermediate infringement claim, see 17 U.S.C.S 106(1); Sega, 977 F.2d at 1518-19, and both methods ofreverse engineering require it, we find no reason inherent inthese methods to prefer one to another as a matter of copy-right law. Connectix presented evidence that it observed theSony BIOS in an emulated environment to observe the func-tional aspects of the Sony BIOS. When this method of reverseengineering was unsuccessful, Connectix engineers disassem-bled discrete portions of the Sony BIOS to view directly theideas contained therein. We conclude that intermediate copy-ing in this manner was "necessary" within the meaning ofSega.[6] We decline to follow the approach taken by the districtcourt. The district court did not focus on whether Connectix'scopying of the Sony BIOS was necessary for access to func-tional elements. Instead, it found that Connectix's copyingand use of the Sony BIOS to develop its own softwareexceeded the scope of Sega. See Order at 17 ("[T]hey disas-sembled Sony's code not just to study the concepts. Theyactually used that code in the development of [their]product."). This rationale is unpersuasive. It is true that Segareferred to "studying or examining the unprotected aspects ofa copyrighted computer program." 977 F.2d at 1520 (empha-sis added). But in Sega, Accolade's copying, observation anddisassembly of Sega's game cartridges was held to be fair use,even though Accolade "loaded the disassembled code backinto a computer, and experimented to discover the interfacespecifications for the Genesis console by modifying the pro-grams and studying the results." Id. at 1515. Thus, the distinc-tion between "studying" and "use" is unsupported in Sega.Moreover, reverse engineering is a technically complex, fre-quently iterative process. Johnson-Laird, 19 U. Dayton L.Rev. at 843-44. Within the limited context of a claim of inter-mediate infringement, we find the semantic distinctionbetween "studying" and "use" to be artificial, and decline toadopt it for purposes of determining fair use.7[7] We also reject the argument, urged by Sony, that Con-nectix infringed the Sony copyright by repeatedly observingthe Sony BIOS in an emulated environment, thereby makingrepeated copies of the Sony BIOS. These intermediate copiescould not have been "necessary" under Sega , contends Sony,because Connectix engineers could have disassembled theentire Sony BIOS first, then written their own ConnectixBIOS, and used the Connectix BIOS to develop the VirtualGame Station hardware emulation software. We acceptSony's factual predicate for the limited purpose of this appeal.8Our doing so, however, does not aid Sony.[8] Sony contends that Connectix's reverse engineering ofthe Sony BIOS should be considered unnecessary on therationale that Connectix's decision to observe the Sony BIOSin an emulated environment required Connectix to make moreintermediate copies of the Sony BIOS than if Connectix hadperformed a complete disassembly of the program. Under thislogic, at least some of the intermediate copies were not neces-sary within the meaning of Sega. This construction stretchesSega too far. The "necessity" we addressed in Sega was thenecessity of the method, i.e., disassembly, not the necessity ofthe number of times that method was applied. See 977 F.2dat 1524-26. In any event, the interpretation advanced by Sonywould be a poor criterion for fair use. Most of the intermedi-ate copies of the Sony BIOS were made by Connectix engi-neers when they booted up their computers and the SonyBIOS was copied into RAM. But if Connectix engineers hadleft their computers turned on throughout the period duringwhich they were observing the Sony BIOS in an emulatedenvironment, they would have made far fewer intermediatecopies of the Sony BIOS (perhaps as few as one per com-puter). Even if we were inclined to supervise the engineeringsolutions of software companies in minute detail, and we arenot, our application of the copyright law would not turn onsuch a distinction.9 Such a rule could be easily manipulated.More important, the rule urged by Sony would require that asoftware engineer, faced with two engineering solutions thateach require intermediate copying of protected and unpro-tected material, often follow the least efficient solution. (Incases in which the solution that required the fewest number ofintermediate copies was also the most efficient, an engineerwould pursue it, presumably, without our urging.) This is pre-cisely the kind of "wasted effort that the proscription againstthe copyright of ideas and facts . . . [is] designed to prevent."Feist Publications, Inc. v. Rural Tel. Serv. Co.,
499 U.S. 340
,354 (1991) (internal quotation marks omitted). Such anapproach would erect an artificial hurdle in the way of thepublic's access to the ideas contained within copyrighted soft-ware programs. These are "aspects that were expressly deniedcopyright protection by Congress." Sega, 977 F.2d at 1526(citing 17 U.S.C. S 102(b)). We decline to erect such a barrierin this case. If Sony wishes to obtain a lawful monopoly onthe functional concepts in its software, it must satisfy themore stringent standards of the patent laws. See Bonito Boats,Inc. v. Thunder Craft Boats, Inc.,
489 U.S. 141
, 160-61(1989); Sega, 977 F.2d at 1526. This Sony has not done. Thesecond statutory factor strongly favors Connectix. 2. Amount and substantiality of the portion used[9] With respect to the third statutory factor, amount andsubstantiality of the portion used in relation to the copyrightedwork as a whole, Connectix disassembled parts of the SonyBIOS and copied the entire Sony BIOS multiple times. Thisfactor therefore weighs against Connectix. But as we con-cluded in Sega, in a case of intermediate infringement whenthe final product does not itself contain infringing material,this factor is of "very little weight." Sega, 977 F.2d at 1526-27; see also Sony Corp. of Am. v. Universal City Studios, Inc.,
464 U.S. 417, 449
-50 (1984) (copying of entire work does notpreclude fair use). 3. Purpose and character of the use[10] Under the first factor, purpose and character of the use,we inquire into whether Connectix's Virtual Game Station merely supersedes the objects of the original cre- ation, or instead adds something new, with a further purpose or different character, altering the first with new expression, meaning, or message; it asks, in other words, whether and to what extent the new work is "transformative."Campbell v. Acuff-Rose Music, Inc.,
510 U.S. 569, 579
(1994)(internal quotation marks and citations omitted). As an initialmatter, we conclude that the district court applied an errone-ous legal standard; the district court held that Connectix'scommercial purpose in copying the Sony BIOS gave rise toa "presumption of unfairness that . . . can be rebutted by thecharacteristics of a particular commercial use." Order at 14-15(citing Sega, 977 F.2d at 1522). Since Sega, however, theSupreme Court has rejected this presumption as applied to thefirst and fourth factor of the fair use analysis. Acuff-Rose, 510U.S. at 584, 594 (clarifying Sony,
464 U.S. at 451
). Instead,the fact that Connectix's copying of the Sony BIOS was fora commercial purpose is only a "separate factor that tends toweigh against a finding of fair use." Id. at 585 (internal quota-tion marks omitted).10[11] We find that Connectix's Virtual Game Station ismodestly transformative. The product creates a new platform,the personal computer, on which consumers can play gamesdesigned for the Sony PlayStation. This innovation affordsopportunities for game play in new environments, specificallyanywhere a Sony PlayStation console and television are notavailable, but a computer with a CD-ROM drive is. Moreimportant, the Virtual Game Station itself is a wholly newproduct, notwithstanding the similarity of uses and functionsbetween the Sony PlayStation and the Virtual Game Station.The expressive element of software lies as much in the orga-nization and structure of the object code that runs the com-puter as it does in the visual expression of that code thatappears on a computer screen. See 17 U.S.C.S 102(a)(extending copyright protection to original works of author-ship that "can be perceived, reproduced, or otherwise commu-nicated, either directly or with the aid of a machine ordevice"). Sony does not claim that the Virtual Game Stationitself contains object code that infringes Sony's copyright. Weare therefore at a loss to see how Connectix's drafting ofentirely new object code for its VGS program could not betransformative, despite the similarities in function and screenoutput.[12] Finally, we must weigh the extent of any transforma-tion in Connectix's Virtual Game Station against the signifi-cance of other factors, including commercialism, that militateagainst fair use. See Acuff-Rose,
510 U.S. at 579
. Connectix'scommercial use of the copyrighted material was an intermedi-ate one, and thus was only "indirect or derivative." Sega, 977F.2d at 1522. Moreover, Connectix reverse-engineered theSony BIOS to produce a product that would be compatiblewith games designed for the Sony PlayStation. We have rec-ognized this purpose as a legitimate one under the first factorof the fair use analysis. See id. Upon weighing these factors,we find that the first factor favors Connectix.The district court ruled, however, that the Virtual GameStation was not transformative on the rationale that a com-puter screen and a television screen are interchangeable, andthe Connectix product therefore merely "supplants " the SonyPlayStation console. Order at 15. The district court clearlyerred. For the reasons stated above, the Virtual Game Stationis transformative and does not merely supplant the PlaySta-tion console. In reaching its decision, the district court appar-ently failed to consider the expressive nature of the VirtualGame Station software itself. Sony's reliance on InfinityBroadcast Corp. v. Kirkwood, 150 F.3d 104 (2d Cir. 1998),suffers from the same defect. The Infinity court reasoned thata "change of format, though useful, is not technically atransformation." Id. at 108 n.2. But the infringing party in thatcase was merely taking copyrighted radio transmissions andretransmitting them over telephone lines; there was no newexpression. Id. at 108. Infinity does not change our conclu-sion; the purpose and character of Connectix's copying pointstoward fair use. 4. Effect of the use upon the potential market[13] We also find that the fourth factor, effect of the useupon the potential market, favors Connectix. Under this fac-tor, we consider not only the extent of market harm caused by the particular actions of the alleged infringer, but also "whether unrestricted and widespread conduct of the sort engaged in by the defendant . . . would result in a substantially adverse impact on the potential market" for the original.Acuff-Rose,
510 U.S. at 590
(quoting 3 M. Nimmer & D.Nimmer, Nimmer on Copyright, S 13.05[A][4], at 13-102.61(1993)). Whereas a work that merely supplants or supersedesanother is likely to cause a substantially adverse impact on thepotential market of the original, a transformative work is lesslikely to do so. See id. at 591; Harper & Row, Publishers, Inc.v. Nation Enters, Inc.,
471 U.S. 539, 567
-69 (1985).[14] The district court found that"[t]o the extent that sucha substitution [of Connectix's Virtual Game Station for SonyPlayStation console] occurs, Sony will lose console sales andprofits." Order at 19. We recognize that this may be so. Butbecause the Virtual Game Station is transformative, and doesnot merely supplant the PlayStation console, the VirtualGame Station is a legitimate competitor in the market for plat-forms on which Sony and Sony-licensed games can be played.See Sega, 977 F.2d at 1522-23. For this reason, some eco-nomic loss by Sony as a result of this competition does notcompel a finding of no fair use. Sony understandably seekscontrol over the market for devices that play games Sony pro-duces or licenses. The copyright law, however, does not con-fer such a monopoly. See id. at 1523-24 ("[A]n attempt tomonopolize the market by making it impossible for others tocompete runs counter to the statutory purpose of promotingcreative expression and cannot constitute a strong equitablebasis for resisting the invocation of the fair use doctrine.").This factor favors Connectix.[15] The four statutory fair use factors must be "weighedtogether, in light of the purposes of copyright. " Acuff-Rose,
510 U.S. at 578
. Here, three of the factors favor Connectix;one favors Sony, and it is of little weight. Of course, the statu-tory factors are not exclusive, Harper & Row, 471 U.S. at560, but we are unaware of other factors not already consid-ered that would affect our analysis. Accordingly, we concludethat Connectix's intermediate copying of the Sony BIOS dur-ing the course of its reverse engineering of that product wasa fair use under 17 U.S.C. S 107, as a matter of law. Withrespect to its claim of copyright infringement, Sony has notestablished either a likelihood of success on the merits or thatthe balance of hardships tips in its favor. See Cadence DesignSys., Inc. v. Avant! Corp., 125 F.3d 824, 826 (9th Cir. 1997),cert. denied,
523 U.S. 1118
(1998). Accordingly, we need notaddress defenses asserted by Connectix under 17 U.S.C.S 117(a)(1) and our doctrine of copyright misuse. We reversethe district court's grant of a preliminary injunction on theground of copyright infringement.11B. Tarnishment[16] The district court found that Connectix's sale of theVirtual Game Station tarnished Sony's "PlayStation" markunder 15 U.S.C. S 1125(c)(1). The district court based its pre-liminary injunction, however, exclusively on Sony's copy-right claim, and did not cite its tarnishment finding as aground for the injunction. Although we can "affirm the dis-trict court on any ground supported by the record, " Charley'sTaxi Radio Dispatch Corp. v. SIDA of Haw., Inc., 810 F.2d869, 874 (9th Cir. 1987), we decline to affirm on this alterna-tive ground. Sony has not shown a likelihood of success oneach element of the tarnishment claim.[17] To prevail on its tarnishment claim, Sony must showthat (1) the PlayStation "mark is famous;" (2) Connectix is"making a commercial use of the mark;" (3) Connectix's"use began after the mark became famous;" and(4) Connectix's "use of the mark dilutes the quality of themark by diminishing the capacity of the mark to identify anddistinguish goods and services." Films of Distinction, Inc. v.Allegro Film Prods., Inc., 12 F. Supp. 2d 1068, 1078 (C.D.Cal. 1998); 15 U.S.C. SS 1125(c)(1), 1127 (definition of"dilution"). Connectix does not dispute the first and third ofthese elements. We address only the fourth element.Because Sony proceeds under a tarnishment theory of dilu-tion, it must show under this fourth element that its PlaySta-tion mark will "suffer negative associations" through Connec-tix's use. Hormel Foods Corp. v. Jim Henson Prods., Inc., 73F.3d 497, 507 (2d Cir. 1996); see also 4 J. Thomas McCarthy,McCarthy on Trademarks and Unfair CompetitionS 24.95(4th ed. 1996 & Supp. 1999). The district court found the Vir-tual Game Station does not play PlayStation games as well asthe PlayStation console, and that although the Virtual GameStation's packaging contains a disclaimer to this effect, "gameplayers do not comprehend this distinction." Order at 24-25.The Sony PlayStation mark therefore suffers negative associa-tions because of this confusion on the part of consumers whoplay Sony games on the Virtual Game Station software. Id. at25.[18] The evidence on the record does not support such afinding of misattribution. The district court relied primarily ona series of semi-anonymous reviews posted on the Internetand submitted by Connectix. As the district court acknowl-edged, these reviews were neither authenticated nor identi-fied. More important, the print-out of the comments does notreveal the context in which the comments were made; thisomission makes the extent of any confusion by game playersdifficult to assess reliably. The district court also referred totwo focus group studies conducted by market research firmsat Sony's bequest. These studies address the difference ofquality between the Virtual Game Station and PlayStation, butshed no light on the question of misattribution. Thus, wereject as clearly erroneous the district court's finding that theVirtual Game Station tarnishes the Sony PlayStation mark ona misattribution theory of tarnishment.[19] Nor are we persuaded by Sony's argument that the dif-ference in quality between the two platforms is itself suffi-cient to find tarnishment. See Deere & Co. v. MTD Prods,Inc., 41 F.3d 39, 43 (2d Cir. 1994) (" `Tarnishment' generallyarises when the plaintiff's trademark is linked to products ofshoddy quality," diminishing the value of the mark "becausethe public will associate the lack of quality . . . with the plain-tiff's unrelated goods."). Even if we assume, without decid-ing, that the concept of tarnishment is applicable to thepresent factual scenario, there is insufficient evidence to sup-port a finding of tarnishment. "The sine qua non of tarnish-ment is a finding that plaintiff's mark will suffer negativeassociations through defendant's use." Hormel Foods, 73 F.3dat 507. The evidence here fails to show or suggest that Sony'smark or product was regarded or was likely to be regardednegatively because of its performance on Connectix's VirtualGame Station. The evidence is not even substantial on thequality of that performance. The Sony studies, each ofincluded eight participants, presented a range of conclusions.One study concluded that "[o]n balance, the results of thisfocus group study show that the testers preferred the PlaySta-tion gaming experience over the Virtual Game Station gamingexperience." The other concluded that consumers found theVirtual Game Station was "generally acceptable " for onegame, but "nearly unplayable" on another. The internetreviews submitted by Connectix also presented a range ofopinion; while some anonymous reviewers loved the VirtualGame Station, some were ambivalent, and a relative fewhated the Virtual Game Station emulation. In the only reviewfor attribution, Newsweek said the software played"surprisingly well," and that some games on the Virtual GameStation "rocked." Steven Levy, "Play it Your Way,"Newsweek, Mar. 15, 1999, at 84. This evidence is insufficientto support a conclusion that the shoddiness of the VirtualGame Station alone tarnishes the Sony mark. Sony's tarnish-ment claim cannot support the injunction.CONCLUSIONConnectix's reverse engineering of the Sony BIOSextracted from a Sony PlayStation console purchased by Con-nectix engineers is protected as a fair use. Other intermediatecopies of the Sony BIOS made by Connectix, if they infringedSony's copyright, do not justify injunctive relief. For thesereasons, the district court's injunction is dissolved and thecase is remanded to the district court. We also reverse the dis-trict court's finding that Connectix's Virtual Game Station hastarnished the Sony PlayStation mark.REVERSED AND REMANDED.
___________________________FOOTNOTES 1 Any purchaser of a copyrighted software program must copy the pro-gram into the memory of a computer in order to make any use at all of theprogram. For that reason, 17 U.S.C. S 117(a)(1) provides that it shall notbe an infringement for one who owns a software copy to make anothercopy "created as an essential step in the utilization of the computer pro-gram in conjunction with a machine and that it is used in no othermanner." Connectix contends that its copying is within the protection ofsection 117, but our disposition of the fair use issue makes it unnecessaryfor us to address that contention. See Sega, 977 F.2d at 1517-18 (rejectingcontention that disassembly is protected by section 117).2 Object code is binary code, consisting of a series of the numerals zeroand one, readable only by computers.3 Source code is readable by software engineers, but not by computers.4 Software is generally written by programmers in source code (and inother more conceptual formats) and then assembled into object code.5 The factors for determining fair use include: (1) the purpose and character of the use, including whether such use is of a commercial nature or is for nonprofit educational purposes; (2) the nature of the copyrighted work; (3) the amount and substantiality of the portion used in relation to the copyrighted work as a whole; and (4) the effect of the use upon the potential market for or value of the copyrighted work.17 U.S.C. S 107.6 Connectix offers evidence that it attempted to gain access to the func-tionality of the Sony BIOS by attaching a "logic analyzer" to the input andoutput leads of the chip on which the Sony BIOS was located within thePlayStation console. This form of observation does not appear to requirethe making of an intermediate copy, but was of limited value because itpermitted the observation of inter-chip, but not intra-chip signals. Sonydoes not suggest that this form of observation alone would have permittedConnectix engineers to gain access to the functional elements of the SonyBIOS.7 We are unable to locate evidence in the record to support the districtcourt's finding that Connectix "gradually convert[ed] Sony's code to theirown code," Order at 11, if by this statement the court meant that Connec-tix engineers failed to create an original work. True, Connectix engineersadmitted to combining the Sony BIOS with the Virtual Game Stationhardware emulation software to test and develop the hardware emulationsoftware. But in drafting the Connectix BIOS, Connectix engineers neverclaimed to do anything other than write their own code, even though theyused, observed, copied and sometimes disassembled the Sony BIOS asthey did so. Sony presents no evidence to the contrary, nor does Sony con-tend that Connectix's final product contains infringing material.8 The depositions of Connectix engineers Aaron Giles and Eric Trautsuggest that Connectix engineers recognized that other engineering solu-tions were sometimes available.With respect to the observation of the Sony BIOS in an emulated envi-ronment, Traut admitted that it was easier to use the Sony BIOS todevelop the hardware emulation software than to develop Connectix ownBIOS first, and then use the Connectix BIOS to develop the hardwareemulation software.With respect to the observation of the Sony BIOS with selective disas-sembly of the code, Traut stated with respect to one bug that there wouldhave been no way to fix the bug without disassembling a portion of theSony BIOS. He also stated that at other times he disassembled portions ofthe Sony BIOS when doing so was "the most efficient way of finding thatbug." In a subsequent question, he clarified that disassembly was not theonly way to fix the bug, just the fastest way to do so.With respect to Connectix's observation of the Sony BIOS in the devel-opment of the Virtual Game Station for Windows, other solutions, pre-sumably disassembly, may have been possible. Connectix engineer Gilesresponded "I don't know" when asked by Sony counsel if "it would havebeen possible to write the CD-ROM code before building the emulator."9 Sony relies on these RAM copies for its contention, which we reject,that there is no significant difference between the facts of this case and ourdecisions in Triad Systems Corp. v. Southeastern Express Co., 64 F.3d1330 (9th Cir. 1995) and MAI Systems Corp. v. Peak Computer, Inc., 991F.2d 511 (9th Cir. 1993). Those cases are inapposite to our fair use analy-sis. Neither involved reverse engineering of software to gain access tounprotected functional elements.10 Sony points to Micro Star v. Formgen, Inc., 154 F.3d 1107 (9th Cir.1998), for the proposition that commercial use creates a presumption ofunfairness. See id. at 1113 (quoting Sony Corp. of Am. v. Universal CityStudios, Inc.,
464 U.S. 417, 451
(1984)). We do not read Micro Star thatway; moreover, such a reading would be contrary to Acuff-Rose. Acuff-Rose expressly rejected such a "hard evidentiary presumption" and statedthat the Court of Appeals "erred" by giving such dispositive weight to thecommercial nature of the use.
510 U.S. at 584
. Also, Micro Star itselfinvolved a use that was non-transformative, which is not the case here. SeeMicro Star, 154 F.3d at 1113 & n.6. Cf. American Geophysical Union v.Texaco, Inc., 60 F.3d 913, 921-22 (2d Cir. 1995) (amended opinion)(rejecting, on grounds of Acuff-Rose and collected cases, presumption ofunfairness for commercial use as applied to Texaco's intermediate copy-ing of copyrighted articles).11 We do not accept Sony's argument that the downloading of Sony'sBIOS from the Internet was itself an infringement justifying the injunc-tion. The evidence of record suggests that the downloaded BIOS playeda minimal role, if any, in development of the Virtual Game Station. Weconclude that, on this record, the downloading infringement, if such itwas, would not justify our upholding the injunction on the developmentand sale of the Virtual Game Station. The Virtual Game Station itselfinfringes no copyright. Bearing in mind the goals of the copyright law, "tostimulate artistic creativity for the general public good," Sony, 464 U.S.at 432 (internal quotation marks omitted), we conclude that there is a legit-imate public interest in the publication of Connectix's software, and thatthis interest is not overborne by the record evidence related to the down-loaded BIOS. The imposition of an injunction is discretionary. See 17U.S.C. S 502(a). On this record, we conclude that it would be inappropri-ate to uphold the injunction because of Connectix's copying and use of thedownloaded Sony BIOS; damages would adequately protect Sony's inter-est with respect to that alleged infringement. See Acuff-Rose, 510 U.S. at578 n.10 (discussing factors to be evaluated in deciding whether to enjoinproduct found to have exceeded bounds of fair use).