MEMORANDUM AND ORDER
This matter is before the Court on plaintiffs motion for a preliminary injunction. Plaintiff seeks an order of the Court enjoining defendant from publishing, selling, marketing, or otherwise disposing of any copies of defendant’s LTR-compatible radio program in any form, and impounding during the pendency of this action and destruction upon conclusion thereof any materials, programs, or other articles of information by means of which plaintiff’s copyrighted computer software has been or may be produced by defendant. The Court heard testimony and arguments on plaintiff’s motion September 9-10, 1985. Plaintiff’s motion will be granted. This order incorporates the findings of fact and conclusions of law required by Federal Rule of Civil Procedure 52.
I. FACTS
A. Parties
Plaintiff E.F. Johnson Co. (EFJ) is a Minnesota corporation engaged in the business of manufacturing and selling two-way land-based communications systems. EFJ’s principal place of business is in Waseca, Minnesota. Defendant Uniden Corporation of America (Uniden) is an Indiana corporation, a subsidiary of Uniden Corporation of Japan. Uniden imports and distributes electronic equipment including land-based communications systems.
B. Background
In the spring of 1980 EFJ introduced into the market its newly developed “Clearchannel LTR” logic trunked radio system (LTR). A logic trunked radio system is one consisting of mobile radi.o units, typically installed in motor vehicles such as taxis, police cars, delivery trucks, etc., and “repeaters,” base stations which receive and transmit signals to and from the mobile radio units. The heart of the EFJ LTR system is computer software contained in the mobile radios and repeaters. The computer software, independently developed by EFJ engineers Mervin Grindahl, Keith Barnes, and Phillip Keefer, allows the LTR system to pool radio frequency channels, thereby making all assigned radio channels accessible to all system users, at a significant gain in operational efficiency. The software contained in the EFJ 8700 series mobile radio units are subject to copyright.
In April, 1985, defendant Uniden introduced into the market its model FTS 250T two-way 800 MHZ FM trunked mobile radio compatible with LTR-system radios and repeaters. The Uniden mobile radio also contains computer software, which allows it to receive and transmit messages from and to EFJ’s LTR-system radios.
Shortly after the Uniden FTS 250T radios came onto the market EFJ engineers subjected one of the radios to scrutiny at their Waseca laboratories. Concluding that the software contained in the FTS 250T radio is identical to software contained in EFJ radios, EFJ commenced the instant litigation, claiming copyright infringement, 17 U.S.C. § 501 et seq., and seeking preliminary injunctive relief.
C. Definitions
Before launching into the technological complexities of this matter, a few definitions of terminology as used in this opinion may be helpful.
ALGORITHM—A step-by-step procedure, or defined set of instructions, designed to solve a particular problem or produce a particular result.
ASSEMBLY LANGUAGE—A computer program written in humanly-recognizable *1488 commands or mnemonics. An assembly language program is converted into machine-readable form by an assembler.
BINARY—A system of numeration consisting of only two digits, zero and one.
BIT—The smallest unit of information used in a computer, represented as zero or one.
BYTE—A grouping of eight bits.
EPROM—Anagram for Erasable Programmable Read-Only Memory. A microchip on which data can be stored, and which may be erased and reprogrammed.
HEXADECIMAL—A base 16 numbering system used as a shorthand representation of a string of binary instructions.
MICROPROCESSOR—A miniature computer placed on a single microchip which is capable of performing arithmetic, logic, and control functions.
PROGRAM—“a set of statements [i.e., data] or instructions to be used directly or indirectly in a computer in order to bring about a certain result,” Pub.L. No. 96-517, § 10(a), 94 Stat. 3015, 3028 (1980) (amending 17 U.S.C. § 101 (1976)).
PROM—Anagram for Programmable Read-Only Memory. A microchip-mounted program from which a computer may read instructions. Nonerasable.
OBJECT CODE—Computer program expressed as sequence of ones and zeroes, readable only by machines.
SOURCE CODE—Disassembled object code, expressed in humanly-readable form with accompanying comments and labels.
D. EFJ’s Development of the LTR-System Radio
EFJ engineers first conceived the idea of a trunked mobile radio system in late summer, 1977. At that time no trunked radio systems were on the market. EFJ engineer Mervin Grindahl was assigned the task of developing the LTR code. In pursuit of this objective, Grindahl first developed signal methods for sending data transmissions over radio waves to direct mobile radios. Grindahl then developed a system architecture which permitted use of the signalling method, and signalling protocols to implement the system architecture. Finally, Grindahl created the algorithms necessary to development of the LTR computer software.
At this point the baton was passed to Keith Barnes, another EFJ engineer, who wrote the LTR software program from specifications detailed in the Grindahl algorithms. Barnes completed the detailed program sometime in 1978. Following an extensive period of “debugging” and program modification, the LTR system was ready for public unveiling in April, 1980.
The LTR mobile radio introduced into the market in 1980 was the model 8800 radio. The model 8800 radio contains a computer software program authored by Grindahl and Barnes. The Grindahl-Barnes EFJ LTR software is the subject of a valid copyright, Registration No. TX 957-037, registered June 30, 1982.
During the period 1982-85 EFJ engineers made certain modifications to the LTR mobile radio. Specifically, EFJ introduced into the market mobile radio models 8700, 8710, 8805, 8810, 8855, 8865, and 8870. Each of these models contains computer software programs, designated by EFJ as versions 1.01, 2.0, 3.0, 4.0, and 5.0. Version 3.0 was authored by EFJ engineer Phillip A. Keefer. Version 3.0 includes certain design improvements and enhancements of the Grindahl-Barnes program made by Keefer, who joined the LTR project in October, 1982. Version 3.0 of the EFJ LTR software is the subject of a valid copyright, Registration No. TX1-568-701, registered June 4, 1985. EFJ mobile radios containing version 3.0 or derivations thereof are models 8805, 8855, 8865, 8870, and 8710.
Sales of EFJ’s LTR-system radios since their introduction have been brisk. EFJ’s annual sales volume of trunked radios is approximately $25 million factory net, $40 *1489 million retail. Approximately 30 percent of EFJ’s total sales are of mobile trunked radios. The national retail market for mobile trunked radios is on the order of $200-$250 .million per year. Of 500 authorized EFJ retail dealers, 150 carry trunked mobile radios.
E. Logic Trunked Radio Systems
The concept of “trunking” had its genesis in the telephone industry. 1 As the Court understands it, the “trunking” of frequency channels permits the system to afford all system users automatic access to all channels for maximum efficiency. Rather than assigning each user a discrete channel, the trunked system, through the use of sophisticated computer software, patches together unutilized airwave “spaces” to create an uninterrupted channel of communication. The net result is that the system can accommodate more users than it has frequency channels available.
EFJ’s clearchannel LTR system is composed of two elements: (1) repeaters, which control access to the system and receive and retransmit signals from mobile radio units; and (2) mobile radio and control units. The system works in the following manner. Each mobile radio is assigned to a repeater. Repeaters are analogous to radio station transmitters. The mobile radios send signals—high speed digital data bursts at sub-audible frequencies—to the repeater, which identifies the sending unit and which retransmits and amplifies the signal so that it can be received by the appropriate mobile unit. The “trunked logic” aspect of the system comes into play in the system’s selection of open frequency channels. The FCC has assigned certain radio frequencies for use by trunked radio systems. Dealers who sell repeaters are licensed by the FCC to operate a trunked system on an assigned frequency. 2 In addition, any purchaser of a mobile radio or radios must purchase the right to use the repeater from the licensed dealer or system operator and must also obtain an FCC license for the radios. Thus, the universe of potential utilizers of trunked radios is finite, as constrained by the number of frequency channels assigned by the FCC for trunked radio use. A “fully loaded” trunked radio channel will support between 100 and 500 mobile radios. Hence, the market for mobile radios is much larger than the market for repeaters.
EFJ manufactures both repeaters and mobile radios. The heart of the LTR system, as mentioned above, is the copyrighted computer software found in both repeaters and mobile radios. EFJ’s software is mounted on a Read-Only Memory (ROM) microchip. EFJ’s mobile radios use an Intel 8049 microprocessor.
Currently, there are four major manufacturers of mobile trunked radio systems in the United States: General Electric, Motorola, Tac-Tel, and EFJ. Three companies manufacture mobile radios compatible with LTR-system repeaters and radios: Standard Communications, Regency, and Uni-den.
F. Uniden Enters the Fray
Uniden commenced development of a mobile radio compatible with EFJ’s LTR radi *1490 os and repeaters sometime in 1984. In the course of developing an LTR-compatible radio, Uniden engineer S. Katsukura disassembled the software found in EFJ’s model 8855 and 8800 radios. Disassembly of a computer program is done by translating the machine or object code into humanly-readable assembly language. Having disassembled the EFJ software program, Uniden engineers Katsukura and S. Uwabe prepared and studied flow charts of the EFJ program. Uniden engineers also studied EFJ hardware and service manuals in developing the FTS 250T software.
When through industry sources EFJ learned of Uniden’s plans to market an LTR-compatible radio, EFJ in letters dated January 24, 1984 and March 5, 1984, advised Uniden that the software in the LTR mobile radios was copyrighted, and that any attempt by competitors to infringe the EFJ copyrights would be vigorously contested. Uniden unveiled its model FTS 250T 800 MHZ trunked mobile radio at the Expo West industry trade show in Las Vegas, Nevada on April-23-26, 1985. EFJ engineer Phillip A. Keefer attended Expo West, visually examined the Uniden radio and obtained Uniden promotional literature.
Thereafter EFJ engineers obtained a Uniden model FTS 250T radio and subjected it to scrutiny at EFJ’s Waseca laboratories. EFJ engineers Grindahl and Keefer first cheeked to see if the Uniden radio was in fact compatible with LTR repeaters. Finding that it was, Keefer removed from the Uniden radio the Eraseable Programmable Read-Only Memory microchip (EPROM) on which the Uniden software was stored. Keefer and Grindahl “dumped” the Uniden program or “code” from the EPROM, using for this purpose a PROM programmer, before “uploading” it into an EFJ computer for purposes of making a comparison of the Uniden and EFJ programs.
EFJ engineers Grindahl and Keefer first compared the “data tables” contained in the Uniden and EFJ software programs. They came to the conclusion that the data tables were identical.
Grindahl then wrote a disassembly program—one which converts machine language code into humanly-readable assembly language. Grindahl made a brief comparison of the Uniden and EFJ codes in assembly language, and when he found evidence of copying, instructed Keefer to conduct a more thorough examination. Keefer did a line-by-line comparison of the codes. As a consequence of their comparisons, Grindahl and Keefer came to the conclusion that the EFJ software had been copied by Uniden.
The Uniden model FTS 250T radio utilizes a Hitachi HD63B05X2P microprocessor. Uniden manufactures its FTS 250T radios in Taiwan and distributes them throughout the United States. Many of the same dealers who carry EFJ mobile trunked radios also carry Uniden FTS 250T radios. Retail prices for Uniden’s radios are some 31-39 percent less than the retail price of EFJ radios.
II. DISCUSSION
Plaintiff has moved the Court for an order enjoining the defendant from publishing, selling, marketing, or otherwise disposing of any copies of defendant’s LTR-compatible radio program, and is further seeking impoundment of any and all materials, programs, and other articles and information by means of which plaintiff’s copyright has been or is being produced by defendant. The test for whether preliminary injunctive relief should issue is set forth in
Dataphase Systems, Inc. v. CL Systems, Inc.,
(1) the threat of irreparable harm to the movant;
(2) the state of balance between this harm and the injury that granting the injunction will inflict on other parties litigant;
(3) the probability that movant will succeed on the merits; and
*1491 (4) the public interest.
Dataphase,
It is axiomatic that no one of the four
Dataphase
factors is determinative.
Dataphase,
Plaintiffs complaint charges defendant with two statutory wrongs. First, plaintiff contends that defendant has infringed its valid copyright on the software for its LTR-system mobile radios. Second, plaintiff contends that Uniden’s alleged pirating of the LTR-system software constitutes a violation of the Minnesota Uniform Trade Secrets Act, Minn.Stat. §§ 325C.01 to .07. For purposes of its motion for a preliminary injunction, plaintiff has concentrated on the copyright infringement count solely.
A. Likelihood of Success on the Merits
1. Copyright Ownership
In order to establish a claim of copyright infringement, plaintiff must prove its ownership of a valid copyright, and copying, or infringement, of the copy
*1492
righted work by the defendant.
Animal Fair,
at 185,
citing Atari, Inc. v. North American Philips Consumer Electronics Corp.,
The Court finds that adequate notice of the EFJ software copyright was included in every EFJ Model 8800 and 8855 radio. Phillip Keefer included a copyright notice in the version 3.0 software program in ASCII language. ASCII language is an internationally understandable combination of control characters and alpha numeric characters established in the American Standard Code for Information Interchange. The disassembled object code analyzed by Uniden engineers Katsukura and Uwabe bears EFJ’s copyright notice. 3
Because direct evidence of copying often is unavailable, copying may be inferred where two elements are proven: (1) that defendant had access to the copyrighted work, and (2) that the accused work is substantially similar to the copyrighted work.
Atari, Inc. v. North American Philips Consumer Electronics Corp.,
a. Access
The Court finds that defendant had access to EFJ’s copyrighted software program. By their own admission, Uniden engineers removed the ROM on which the EFJ program is mounted from EFJ model 8800 and 8855 radios, dumped the EFJ programs, disassembled the EFJ programs, and analyzed and flow charted the EFJ programs in the course of developing the software for Uniden’s radio.
b. Substantial Similarity
The test of substantial similarity
4
in the Eighth Circuit is “whether the work is recognizable by an ordinary observer as having been taken from the copyrighted source.”
Wihtol v. Crow,
Because a copyrighted computer program is stored on a computer chip or disc well-hidden from public view, application of the ordinary observer test in a computer software context has proven problematical. The absence of an easily perceived general aura or feeling emanating from a silicon chip has led some commentators to suggest an “iterative” approach to substantial similarity. The iterative approach requires proof (1) that the defendant “used” the copyrighted work in preparing the alleged copy, which may be established by proof of access and similarity sufficient to reasonably infer use of the copyrighted work; and (2) that the defendant’s work is an iterative reproduction, that is, one produced by iterative or exact duplication of substantial portions of the copyrighted work. Note, Copyright Infringement of Computer Programs: A Modification of the Substantial Similarity Test, 64 MINN.L.REV. 1264, 1294-1300 (1984). 6 Under the iterative approach, adopted in form if not name by several courts, 7 the factfinder’s focus shifts from the hypothetical ordinary observer’s impressions of the “total concept and feel” of the copyrighted and allegedly infringing works to an analysis of the “quantitative and qualitative evidence of similarities” as gauged by the Court’s evaluation of expert testimony. The fiction of the lay observer is thus abandoned in favor of an analysis of similarities and differences in the copyrighted and allegedly offending computer programs. Note, Copyrighted Computer Program, 68 MINN.L.REV. at 1296-99.
Under either the traditional “ordinary observer” or the contemporary “iterative” test of substantial similarity, it is dear that the software program found in Uniden’s FTS 250T mobile trunked radio is substantially similar to EFJ’s copyrighted version 3.0 software program.
The indicia of substantial similarity may perhaps most usefully be set forth in tabular form:
(1) Barker Code
Due to the fact that Uniden designed its radio to be compatible with EFJ’s LTR system, both parties acknowledge that some similarities in software design were inevitable. The Court finds that in order to make its radios compatible with LTR repeaters Uniden was required to copy the *1494 “Barker code” found in the copyrighted EFJ program. A “Barker code” is a pattern of ones and zeroes alternated in a prepatterned sequence. Both the sending and receiving units must identify the Barker code in order for communication to be established. The EFJ Barker code is numerically depicted as 1011000. In order to make its radios compatible, Uniden was required to and did copy this aspect of the EFJ program.
The matter does not end there, however. In the LTR scheme of transmission identification, synchronization is achieved by a comparison of incoming streams of data with the seven-bit Barker code. By counting the number of matches and nonmatches, and by comparing the number of non-matches or “errors” to the “sample error” data table contained in the EFJ program, the Intel microprocessor is able to determine if an incoming stream of data is in fact the 1011000 Barker word, in which event the number of errors is nil, or whether it is something else, in which event the number of errors is very high. Only when a certain threshold of matches is achieved has synchronization word detection occurred. This is known in the parlance as the shifting correlator scheme for detection.
In order to make this comparison of incoming data with the Barker code the LTR system takes eight “samples” of seven bits each and compares the 56 bits with an internal table, called the “sample error” or “lookup” table. The significance of the 56 bit sampling technique is threefold.
First, EFJ engineers chose to sample 56 bits at a time because the Intel microprocessor is incapable of sampling at a greater rate. By sampling at a 56 bit rate the EFJ program utilizes 96 percent of the Intel microprocessor’s top speed. In contrast, the Hitachi microprocessor used by Uniden in its radios is capable of sampling bits at a much higher rate of speed. Sampling at a rate in excess of 56 bits is preferable because, as admitted by Uniden engineers, the error detection capability increases exponentially with increases in the number of bits sampled. Nevertheless, Uniden chose to sample incoming data streams in exactly the same manner as EFJ—eight samples of seven bits each. This coincidence of sampling rates indicates inferentially that Uni-den engineers copied the EFJ code verbatim without considering the utility which the greater speed of the Hitachi microprocessor afforded them.
Second, the sample error table found in the EFJ program has its exact duplicate in the Uniden program. The duplication is not necessary to compatability, in that other methods of counting errors exist, as, for example, by counting the number of errors in the cross-correlation word by shifting the word one bit at a time and counting the number of bits that have ones. Further, Uniden’s choice of the sample-error method of detecting errors is contrary to logic, in light of design differences in the Intel and Hitachi microprocessors. The sample error table works particularly well with an Intel microprocessor because of the availability in the Intel microprocessor of a “swap command,” one that allows the program to swap bits four at a time. In the Hitachi microprocessor, in contrast, bits must be swapped one at a time, at a corresponding loss of efficiency. 8 Hence, Uniden’s use of a sample error table technique identical to EFJ’s when more efficient methods were available raises an inference of copying.
Third, as originally conceived, the EFJ “error threshold” was set at eight. Synchronization was deemed to have occurred when eight or fewer errors were detected. *1495 Later experimentation by Phillip Keefer revealed that better synchronization could be achieved by setting the threshold at six errors, which is the threshold incorporated in later versions of the EFJ code. Defendant’s code retains the eight error threshold copied from version 3.0, at a corresponding loss of efficiency.
(2) H-Matrix
The Court finds that in order to make its radios compatible with the LTR system Uniden was required to include an H-Matrix table in its software program. An H-Matrix is a series of ones and zeroes arranged in rows and columns in a matrix format. An H-Matrix is used in the LTR system to detect errors in transmission, once communication has been established by matching of Barker codes. To make its radios compatible with the LTR system, Uniden was required to and did employ some form of H-matrix in its software program, as EFJ’s experts admitted in their testimony.
In fact, Uniden’s H-matrix is identical to the H-matrix found in the EFJ program. This exact duplication was not necessary, in that the H-matrix can be configured any of 32 different ways. The fact that Uniden configured its H-matrix in precisely the same manner as EFJ is evidence of copying.
In addition, Uniden’s program includes an inverse H-matrix identical to the inverse H-matrix found in the EFJ program. The EFJ inverse H-matrix is, as the name implies, an H-matrix in which the rows of the matrix have been interchanged. The inverse H-matrix is used in conjunction with the H-matrix to detect transmission errors and is found in both repeaters and radios. The inverse H-matrix in the Uniden program serves the same purpose. The significance of an inverse H-matrix for present purposes lies in the fact that it is almost completely superfluous to the LTR system’s operation. The inverse H-matrix was included in the EFJ program solely due to a miscommunication between EFJ engineers Grindahl, who designed the LTR algorithms, and Barnes, who reduced the algorithms to computer program form. The end result, transmission error detection, could have been achieved through use of the H-matrix solely. Nevertheless, Uniden’s program includes an inverse H-matrix identical to the inverse H-matrix mistakenly included in the EFJ program.
Finally, the EFJ H-matrix was loaded into the software code in reverse order. The reason for this apparent anamoly is that the Intel microprocessor more efficiently decrements (subtracts) than increments (adds) resulting in system efficiencies with a reverse-loaded matrix. In contrast, the Hitachi microprocessor is equally efficient at incrementing and decrementing. Nevertheless, the Uniden H-matrix is loaded in reverse order.
(3) Duplex Function Instructions
The EFJ software program contains three lines of code which are completely unnecessary to operation of the LTR system. The unnecessary instructions are found in the “transmit audio” subroutine. The instructions were originally written into the code by EFJ engineer Barnes to accommodate a “duplex feature”—one which would allow a system user speaking over the LTR radio to simultaneously hear transmissions directed to the user from other radios. When it was determined that duplex transmission was not feasible, that part of the code was removed, with the exception of the offending three instructions, which Barnes left in the code by mistake.
Precisely the same three superfluous instructions are found in the Uniden program, in precisely the same location, the “transmit audio” subroutine. Uniden engineer S. Uwabe acknowledged the presence of the unnecessary instructions. Uwabe contended that the surplusage was written into the code purposely to accommodate a duplex feature in the event that Uniden determined to add such a feature to their *1496 radio at some future date. The Court has difficulty comprehending how Uniden engineers could have independently determined to include precisely the same unnecessary instruction at precisely the same location in the program without having in fact copied the program. The more likely explanation, and the one the Court finds to be credible, is that Uniden unwittingly copied the unnecessary instructions when translating the EFJ code virtually verbatim from Intel to Hitachi code.
The existence of the identical unnecessary instructions in both codes is strong proof of substantial similarity. In
SAS Industries, Inc. v. S & H Computer Systems, Inc.,
(4) Select Call Prohibit
The select call prohibit feature of the EFJ radio prevents LTR system users from attempting to communicate with the “dispatcher” or system operator when the dispatcher is transmitting to another user. Essentially, the select call prohibit causes a “busy signal” to be communicated. In version 3.0 of the LTR software program an error in the select call prohibit feature was written into the program. In brief, the error causes transmissions to be blocked to all dispatchers aligned with a particular repeater when in fact only one dispatcher is transmitting. In later versions of the EFJ code the error has been corrected.
The Uniden select call prohibit feature incorporates the same error found in version 3.0 of the EFJ software. For this comity of errors the Court can conceive of only two plausible explanations: one, that EFJ and Uniden engineers independently committed the same inadvertences; or two, that Uniden engineers unknowingly wrote the error into their code when copying the EFJ code. The Court finds the latter explanation to be the likelier one.
The presence of identical errors in copyrighted and infringing computer programs was held to be evidence of copying in
Williams Electronics, Inc. v. Arctic International, Inc.,
There is overwhelming evidence in the present case that the [plaintiffs] computer program has been copied in some form. The following facts, among others, manifest the similarities between the [plaintiffs] program and that stored in the [defendant’s] memory devices:
(1) The game created by the [defendant’s] circuit boards contained an error which was present in early versions of. the [defendant’s] computer program—it displays the wrong score value for destroying a particular alien symbol[.]
Williams,
(5) Other Evidence of Substantial Similarity
Other, miscellaneous evidence of copying abounds. One of the unnecessary lines of code is a “counter” set arbitrarily by EFJ engineers at 20 code cycles. The Uniden counter is also set at 20 code cycles. The EFJ “receive executive” routine, as modified by Phillip Keefer, is “neat,” in the sense of concise, while the “transmit executive” routine is rambling. In the Uniden code the receive executive routine is neat and the transmit executive routine is rambling. Further, EFJ’s experts testified that 38 of 44 subroutines found in its version 3.0 code are duplicated in the Uniden
*1497
code.
9
Finally, a manual prepared by defendant for use with its FTS 250T radios includes large segments lifted verbatim from an EFJ manual.
See
Plaintiffs exh. 16. Verbatim copying of a computer manual is inferential evidence of pirating of the underlying software.
Synercom Technology v. University Computing Co.,
In sum, the Court finds that the similarities between the EFJ program and the Uniden program are substantial. 10 Both programs achieve Barker word synchronization in precisely the same manner, despite the fact that alternative techniques are available. Both programs “sample” incoming bits at precisely the same speed, despite the fact that Uniden’s Hitachi microprocessor is capable of sampling at a higher speed. Both programs utilize identical sample error tables, H-matrices and inverse H-matrices. Perhaps most tellingly, both programs contain precisely the same superfluous instructions and precisely the same select call prohibit error. The duplicative tables and errors, the identity of 38 out of 44 subroutines and the marked similarity in overall design lead the Court to conclude that the programs are substantially similar.
Defendant has introduced into evidence an exhibit consisting of a line-by-line, side-by-side comparison of the EFJ and Uniden codes, both translated into Hitachi language. While it is obvious at a glance that the programs thus depicted are not line-byline duplications, the Court finds the comparison on the whole unconvincing. Because of differences in addressing features of the Intel and Hitachi microprocessors a given step may take a differing number of commands depending on whether the instructions are set forth in Intel or Hitachi language. Thus, literal translation of plaintiff’s Intel instructions into Hitachi language necessarily involves a skewing of the program such that line-by-line comparison becomes meaningless. In
Whelan Associates, Inc. v. Jaslow Dental Laboratory, Inc.,
[I]t would be very difficult if not impossible to literally translate a program written in EDL to a program written in BASIC. The evidence makes clear that transferring or converting from one computer language to another is not comparable to translating a book written in English to French. At least, it would be a very inefficient method of copying a program to attempt to work solely from the source code and literally translate it from EDL to BASIC.
Whelan,
While acknowledging the presence of certain similarities between its and plaintiff’s program, defendant contends that plaintiff’s code is not copyrightable because a computer program is a “useful” as opposed to a literary work.
See, e.g., Kepner-Tregoe, Inc. v. Carabio,
The Court cannot accept defendant’s characterization of plaintiff’s programs as “a useful work.” Congress has clearly defined computer programs as “literary works.”
Apple Computer, Inc. v. Franklin Computer Corp.,
Against the Court’s finding of substantial similarity, the defendant raises two somewhat related defenses: (1) preexisting material, and (2) the idea/expression dichotomy.
(i) Preexisting Material
Originality is widely recognized as the
sine qua non
of copyrightability. See
L. Batlin & Son, Inc. v. Snyder,
Defendant argues that plaintiff’s version 3.0 software program is not copyrightable because derived in large part from preexisting material in the public domain. Specifically, defendant contends that the H-matrices, Barker word, error look-up tables and Barker correlation tables have long been in general scientific use and are not original with EFJ.
The Court finds that while some aspects of the EFJ program are taken from the public domain, the degree of originality contributed by plaintiff to version 3.0 is far from “trivial.” For example, Mervin Grindahl testified that the EFJ Barker word is a unique permutation of the textbook Barker. EFJ engineers turned the “classic” Barker end for end and inverted all the bits, changing ones to zeroes and zeroes to ones, for the purpose of eliminating inter-symbol interference produced by the textbook Barker word. 12 Further, EFJ engineers “rolled” or inverted the ones on each corner of the Barker digits to accommodate the Intel hardware and changed the textbook Barker correlation table from seven to 56 bits. In addition, the Barker correlation table found in the EFJ code was empirically derived by Grindahl using data streams and an emulator. 13 Thus, while the concept of a Barker word and Barker correlation table is not original with EFJ, the particular word and particular configuration of that table found in version 3.0 is.
Further, the Court finds that the H-matrix found in the EFJ code is not a “classic” matrix taken from textbooks, but rather one created by Grindahl through the use of a computer program which generated a number of pseudorandom six bit sequences which were in turn compiled to form the H-matrix table. 14
In sum, the Court finds that EFJ has contributed something “recognizeably its own” to the compilation of standardized programming techniques and innovative EFJ-generated instructions. The test of originality is a “modest” one,
Durham Industries,
An apt analogy to the case at bar can be found in the Synercom decision of the United States District Court for the Northern District of Texas. In Synercom the court found that 30 percent of a copyrighted computer manual was composed of excerpts from works in the public domain. Nevertheless, the court held the manual to be copyrightable, stating that:
under copyright law the 30% nonoriginal content does not void the copyright as to the 70% or even the 30%, at least as an integrated part of a product whose whole is original. This is not to infer that [plaintiffs] copyright removed the nonoriginal material from the public domain. It is to say that if an assembly of parts— old and new—results in original expression, the whole is protected.
Synercom,
(ii) Idea/Expression Dichotomy
A threshold question in any copyright case is what characteristics of the plaintiff's work are entitled to copyright protection.
Animal Fair,
at 186,
citing Eden Toys, Inc. v. Marshall Field & Co.,
To the extent that defendant premises its idea/expression contentions on the bare fact that compatability with the LTR system was its objective, its arguments must be rejected.
17
This argument was addressed by the United States Court of Ap
*1502
peals for the Ninth Circuit in
Apple Computer, Inc. v. Formula Intern., Inc.,
In the computer context this means that when specific instructions, even though previously copyrighted, are the only and essential means of accomplishing a given task, their later use by another will not amount to an infringement.
National Commission on New Technological Uses of Copyrighted Works, Final Report 20 (1979) (hereafter CONTU Report).
The record amply demonstrates that an LTR-compatible software program could have been written without verbatim duplication of EFJ’s version 3.0 program. The H-matrix contained in both plaintiff’s and defendant’s code, for example, can be configured in any of 32 different ways, as discussed above. Thus, exact duplication of the EFJ H-matrix by the defendant was not the “only and essential” means of achieving compatability. The Barker word was of necessity identical in both codes, but identity of Barker word correlation techniques and sampling rates was not. The *1503 Court finds that while both plaintiff and defendant employed the “shifting correlator” scheme of Barker word detection, other ways of achieving the same task are recognized in the industry—such as the “phase-locked loop method” and the more conventional technique of synchronizing Barker with hardware rather than software. In addition, rather than duplicate EFJ’s inverse H-matrix, the defendant could have accomplished the same task by inverting the check sum obtained from the H-matrix, as defendant’s engineers admit. Phillip Keefer testified and defendant did not deny that more than one possible configuration of the sample error table could have been created and that defendant’s duplication of the EFJ sample error table was not a requisite to compatability. In fact, EFJ has created differing versions of the table in the past.
Thus, the mere fact that defendant set out with the objective of creating an LTR-compatible radio does not, without more, excuse its copying of plaintiff’s code. The Court finds that copying plaintiff’s code was not the only and essential means of creating an LTR-compatible software program. Defendant was required to copy plaintiff’s Barker word, as discussed above. Virtually all other aspects of defendant’s program could have been independently created, however, without violence to defendant’s compatibility objective. Defendant has reproduced the expression, not merely the idea of plaintiff’s copyrighted program. 19
B. Irreparable Harm
The general rule in copyright infringement cases is that irreparable injury is presumed for the purposes of a preliminary injunction motion once the moving party has established a case of copyright infringement.
Dallas Cowboy Cheerleaders, Inc. v. Scoreboard Posters, Inc.,
The Court notes that even if there were no such presumption of irreparable harm, the evidence in the present case would support such a finding. Ongoing infringement of plaintiff’s copyrighted program by defendant jeopardizes plaintiff’s substantial investment in LTR-system radios as well as its competitive position in the marketplace, thus satisfying the requirement of irreparable harm.
See Apple Computer v. Franklin,
C. Balance of Harms
The balance of harms does not militate against the issuance of a preliminary injunction. The only harm which the defendant will suffer is the lost profits.on its
*1504
infringing radios. This is not a weighty-equitable consideration.
Atari, Inc.,
D. Public Interest
The copyright laws are designed to “... encourage individual effort and creativity by granting valuable enforceable rights.”
Atari, Inc.,
Defendant contends that by granting plaintiff’s motion the Court will in effect confer on plaintiff a market monopoly, with deleterious effect to the public. First, the Court notes that plaintiff asserted at trial and defendant did not dispute that two manufacturers other than Uniden, Standard Communications and Regency, market LTR-compatible radios. Second, defendant is perfectly free to develop and market a radio that is LTR-compatible. But is may not do so by pirating plaintiff’s mobile radio programs, if an alternative is available. As stated by the district court in
Apple Computer, Inc. v. Formula Intern., Inc.,
Defendant will still be free to produce programs which result in the machine performing the same calculations, setups, memory loading, etc., as Plaintiff’s software does and thus compete with Plaintiff in the software market. Defendant will not, however, be free to do so by copying the exact number and sequence of bytes or items by which Plaintiff’s program causes the machine to operate. As the CONTU Report aptly put it, “One is always free to make the machine do the same thing as it would if it had copyrighted work placed in it, but only by one’s own creative effort rather than by piracy.”
III. CONCLUSION
The plaintiff has demonstrated a substantial likelihood of success on the merits of its copyright claim. Plaintiff has also established the remaining equitable prerequisites for injunctive relief set forth in
Dataphase Systems, Inc. v. CL Systems, Inc.,
*1505 Based on the foregoing, IT IS ORDERED that:
1. defendant, its officers, employees, agents, servants, and all those in active concert or participation with defendant in the manufacture, promotion, distribution, or sale of its LTR-compatible radio program are hereby preliminarily enjoined and restrained from infringing, in any manner, plaintiffs copyrighted LTR-system computer programs and from publishing, selling, marketing, or otherwise disposing of any copies of defendant’s LTR-compatible radio program in any form, pending the final judgment after trial on the merits of this action;
2. defendant is to take all reasonable steps to make its officers, agents, servants, employees, attorneys and those in active concert and participation with it aware of the existence and terms of this preliminary injunction order, as well as their duty to abide by said order;
3. the parties shall confer and advise the Court if agreement can be reached as to the amount of bond to be posted by the plaintiff, and if no agreement can be reached, each side shall submit to the Court within 21 days of the date of this order briefs detailing a proposed bond amount.
Notes
. Because the telephone industry has more users than it has circuits, it "trunks” or pools circuits so that an open circuit can be automatically selected whenever a call is made. Not every owner of a telephone has a private "line.” Rather, the system takes advantage of pauses in speech to create a single uninterrupted communications montage.
. The FCC originally allocated 150 frequencies from 806-810 MHZ for conventional (nontrunked) radio use and 200 frequencies from 816-821 MHZ for trunking use. The channels in the 810-816 MHZ band were reserved for future use. In 1982 the FCC allocated 50 of these reserve channels for use by specialized mobile radio operators (SMRO). EFJ sells their repeaters to SMROs, who then sell mobile radios to "fill” the assigned channels. SMRO channels are assigned by the FCC and no more than five are assigned at any given time. One group must be constructed and loaded before another group will be assigned. The SMROs operate and maintain the trunked channel which has been assigned to them. EFJ, Motorola, GE and Tac-Tel market repeaters and radios to the SMROs.
. Disassembled object codes discovered in defendant’s files contained the EFJ copyright notice. See Plaintiff’s ex. 38.
. The question of substantiality of similarity is one of fact.
SAS Industries, Inc. v. S & H Computer Systems, Inc.,
. The substantial similarity test had its genesis in the opinion of Judge Learned Hand in
Nichols v. Universal Pictures Corp.,
The test was later modified to include the requirement of "access,” defined generally as opportunity to view the copyrighted work.
See Smith v. Little, Brown & Co.,
. See also Note, Toward a Unified Theory of Copyright Infringement For an Advanced Technological Era, 96 HARV.L.REV. 450 (1982).
. In
Williams Electronics v. Artie International, Inc.,
. Uniden engineer Uwabe contended in an unsworn declaration filed in the case that the swap command was not necessary to four bit sampling, and that the same result could be achieved with a sequence of four shift right operations. Considerable doubt was cast upon this assertion by the testimony of EFJ engineer Keefer, however. The Court finds that Uniden’s duplication of EFJ's sampling technique is evidence of copying, and that use of the sample error table in a microprocessor without a swap command is contrary to logic.
. In the case of In the Matter of Certain Personal Computers and Components Thereof, Investigation No. 337-TA-140 (Final Decision, March 1984), the International Trade Commission found a computer software program to be substantially similar to a copyrighted program where substantial identity of subroutines was present. There it was stated:
As to substantial similarity, this does not require complete identity. Substantial similarity has been found in numerous instances where the similar material is quantitatively important. Sufficient similar material is involved here, and that material appears to be
qualitatively significant, since it includes 23 of the 32 most useful subroutines of the approximately 70 subroutines in the Autostart ROM program.
U.S.I.T.C. Publication 1504 at 22-24.
. A finding of substantial similarity would be made under the iterative test, since the Court has found proof of use, defined as access plus similarity, and iterative or verbatim reproduction of substantial sections of the EFJ code, i.e., the data tables and 38 of 44 subroutines.
. The question of whether computer programs are the valid subject of copyright was authoritatively resolved by a 1980 amendment to the Copyright Act of 1976, Act of Dec. 12, 1980, Pub.L. No. 96-517, § 10, 94 Stat. 3028, which explicitly refers to computer programs.
See
1
Nimmer on Copyright
§ 2.04[c] (1981). Copyright protection is available whether the program is written in humanly readable source code, machine readable object code and whether the code is stored on ROM silicon chip or a disc or tape.
Midway Mfg. Co.
v.
Strohon,
Of course, it is true that because "[t]he computer software industry progresses by a stepping-stone improvement process, with each innovation building on past innovations to produce an improved product,” Note, Copyrighted Computer Programs, 68 MINN.L.REV. at 1291, too loose an application of the substantial similarity test in this field could inhibit salutary innovation by in effect requiring software engineers to “reinvent the wheel” in order to achieve technological progress. Id.
. Keith Barnes admitted and the Court finds that Barker codes are routinely taught in college level computer programming courses. Because the particular Barker word created by plaintiffs engineers is unique, however, the Court further finds that the EFJ Barker, copied by defendant, was not "preexisting material."
. The Barker comparison table was created by Grindahl who stated in a deposition that “[t]he receiver was run with an emulator replacing the microprocessor, a data word was transmitted from some test equipment and received by the receiver, decoded by the emulator a number of times, and the average of these times was used to generate a proper comparison table.” Grindahl deposition at 37.
. Defendant’s exhibit L, Table 6.2 is a textbook H-matrix similar to the EFJ H-matrix. As defendant’s witness admitted, however, the textbook H-matrix is not identical to the EFJ H-matrix, but is rather the product of a shifting of columns.
. In
Batlin,
the seminal case in this area of copyright law, the Second Circuit stated that, while in order to gain a copyright a derivative work must contain "some substantial, not merely trivial originality,"
. Defendant characterizes the idea/expression dichotomy as an element of the substantial similarity test. In
Amstein v. Porter,
More importantly, the Amstein/Krofft approach does not change the nature of the inquiry as dictated by statute, 17 U.S.C. § 102(b), prohibiting copyright protection of “any idea ... [or] principle____” The "bottom line” in a computer software context is whether the copyrighted instructions are the sole means of accomplishing a given task, as the Ninth Circuit recognized in the Apple Computer cases, discussed infra. Thus, whether the idea/expression dichotomy is characterized as an "element” of the substantial similarity test or as a "defense” or "exception” to a claim of infringement is essentially irrelevant—under any formulation, the primary inquiry is whether the stated objective can be accomplished in only one or a few ways.
. The mere fact that defendant’s engineers dumped, flow charted, and analyzed plaintiffs code does not, in and of itself, establish pirating. As both parties’ witnesses admitted, dumping and analyzing competitors’ codes is a standard practice in the industry. Had Uniden contented itself with surveying the general outline of the EFJ program, thereafter converting the scheme into detailed code through its own imagination, creativity, and independent thought, a claim of infringement would not have arisen. As stated in
Synercom,
”[t]he preparation of a computer program in any language from a general description of the problem to be solved (as, for example, is contained in the forms and manuals, which prescribe a problem involving a set of ordered inputs in a particular arrangement which must be accepted by the computer and transmitted to the FRAN program) is very dissimilar to the translation of a literary work, or to the translation of a program from one language to another ... [and] can in no way be said to be merely a copy____”
Synercom,
*1502 Beginning with a broad and general statement of the overall purpose of the program, the author must decide how to break the assigned task into smaller tasks, each of which must in turn be broken down into successively smaller and more detailed tasks. At the lowest levels the detailed tasks are then programmed in source code. At every level, the process is characterized by choice, often made arbitrarily, and only occasionally dictated by necessity-
. This approach to the idea/expression dichotomy in a computer software context has also been adopted in other cases. In the
SAS Institute
case,
. In
Apple Computer v. Franklin
the court noted that: "Franklin claims that ... there are a limited ‘number of ways to arrange operating systems to enable a computer to run the vast body of Apple-compatible software.’ ”
Similarly, in
Apple Computer v. Formula,
the Ninth Circuit found that "Apple introduced evidence that numerous methods exist for writing the programs involved here, and Formula does not contend to the contrary.” This finding led the court to conclude that the Apple program embodied expression and not an idea, and was thus copyrightable.
See also SAS
Industries,
.
See, e.g., Stem Electronics, Inc. v. Kaufman,
writing a new computer program that would interact with the hardware components of a video game to produce on the screen the same images seen in "Scramble," accompanied by the same sounds. Such replication is possible because many different computer programs can produce the same "results," whether those results are an analysis of financial records or a sequence of images and sounds____ To take an elementary example, the result of displaying a "4” can be achieved by an instruction to add 2 and 2, subtract 3 from 7, or in a variety of other ways. Obviously, writing a new program to replicate the play of "Scramble” requires a sophisticated effort, but it is a manageable task.
Stem,
