MEMORANDUM DECISION
ROBERT J. WARD, District Judge.
This case presents interesting and somewhat novel questions concerning the law of trade secrets in the context of computer software utility programs. Plaintiffs Integrated Cash Management Services, Inc., and Cash Management Corporation (collectively, “ICM”) allege that certain proprietary information of value to ICM was misappropriated by defendants Nicholas Mitsos (“Mitsos”), Alfred Newlin (“New-lin”), and Behrouz Vafa (“Vafa”), former employees of ICM. The individual defendants are presently employed by defendant Digital Transactions, Inc. (“DTI”). ICM has withdrawn a claim of copyright infringement which it originally asserted. Defendants have raised various counterclaims. On plaintiffs’ motion the Court severed the issues of liability on the plaintiffs’ claims from all other issues. Trial on
the severed issues commenced on October 16, 1989, and concluded October 25, 1989. The Court, having heard testimony and having considered the stipulations, exhibits and arguments of counsel, finds that DTI is liable for misappropriation and use of ICM trade secrets in creating the generic programs complained of in this lawsuit.
FINDINGS OF FACT
The Court’s findings of fact pursuant to Rule 52, Fed.R.Civ.P. are as follows:
ICM was founded in 1981 by Ben Massu-da and Eric Campbell. Its main business is the design and development of computer software which it then markets to its customers, which are banks. These banks in turn market the software to their customers, primarily treasury and financial departments of corporations. At present, the ICM products are commercially successful, and ICM estimates that it counts among its clients and their customers approximately 500 of the 1000 largest corporations in the United States.
During 1981 — 1983, ICM was one of many companies which rendered general computer consulting services to banks in order to aid them in designing and implementing dedicated computer programs specifically designed to address the needs of the banking industry. When personal, or micro-, computers (“PCs”) began to gain wide application, ICM recognized a need for software that would run on the PCs and yet would be economical to install and operate. According to the testimony of Eric Campbell (“Campbell”), a central theme of the ICM approach involved developing generic programs in the financial environment that could be customized by the bank’s clients to address client specifications without requiring the services of computer consultants for their programming and maintenance. Although defendants have asserted that the generic-type programs at issue in this case are not novel, and in any event are not secret information, there was testimony at trial to the effect that at the time this system was offered by ICM, several other companies claimed to have generic programming products which, when investigated, were not yet operable, (testimony of Mark Kane). Campbell testified that ICM invests millions of dollars in research and development in the design system architecture and details which ultimately are embodied in the programs licensed to its customers.
The specific ICM programs at issue in this case are (1) the ICM SEUNIMNT program, a generic universal database management system, (2) the ICM Telefon program, a generic communications program which allows for customization in response to specific communications protocols utilized by various banks, (3) the ICM Menu System/Driver, which allows the customer to perform various functions at the treasury workstation, and (4) the ICM Report Writer, which permits the user to customize its financial reports in accordance with its particular needs and desires. (A fifth program, the ICM Approval program, was originally included, but ICM withdrew its claim as to that program after learning that DTI’s approval program is not generic.) The gravamen of ICM’s complaint is that these generic components of its system work together in a particular “winning combination” which is not ascertainable by the public and which was known and utilized by defendants in creating their system. (Campbell testimony).
Defendants Mitsos, Newlin and Vafa all worked for ICM in various capacities prior to creating DTI. Both Newlin and Vafa signed nondisclosure agreements, admitted in evidence as plaintiffs’ exhibits five and six, in which they agreed not to disclose or use any confidential or proprietary information of ICM upon leaving that company’s employ, including software products and tools. Mitsos was employed as an independent consultant for ICM at various times between 1981 and 1986. He worked primarily in the marketing area, and has no advanced skills in computer science or programming. Mr. Mitsos is trained, among other things, as a lawyer, and he helped draft an earlier version of the nondisclosure agreements signed by Newlin and Vafa for ICM. He left ICM in September 1986. Newlin worked as a computer pro
grammer for ICM from September 1984 until March 1987. This was his first full-time position after receiving his graduate degree in computer science from the University of Pennsylvania. Vafa was employed by ICM as a computer programmer from June 1986 until March 1987. As with Newlin, this was Vafa’s first full-time position after completing his formal education. Vafa graduated from Purdue University with a degree in electrical engineering, and subsequently received two graduate degrees from the Massachusetts Institute of Technology (“MIT”), one in electrical engineering and one in computer science. Vafa excelled in his academic studies, and the Court is persuaded as to his high intelligence and skill as a computer programmer.
As an employee of ICM, Vafa was given the task of writing a version of the SEU-NIMNT utility in C language. ICM already possessed the SEUNIMNT program written in the BASIC computer language by Don Radlauer, another ICM employee. Vafa had access to the source code of SEU-NIMNT in BASIC, but he has testified that he did not transliterate the code into the C language, but instead “reverse engineered” the program by looking at the screens and the various functions that the program performed and then writing original source code to duplicate those functions and screens. The Court accepts this testimony of Vafa, and in general accepts his testimony as having been credible and forthright. Nonetheless, the SEUNIMNT in C was clearly written by Vafa in his capacity as an employee of ICM, and is owned in all respects by ICM. Defendants do not dispute this fact. In addition, Newlin assisted Vafa in writing the SEUNIMNT program in C. Vafa, along with Newlin, also wrote the program for the Report Writer module of the ICM system. The Communications and Menu modules were written by Newlin in his capacity as an ICM employee.
Campbell, who is the Director of Technical Development and a principal of ICM, testified as to the effort and time that went into designing, developing, debugging, and maintaining the above-mentioned system. In addition, he described to the Court the similarities which he found between the ICM and DTI systems. The Court found Mr. Campbell’s testimony credible, and relies upon it to a large extent because Mr. Campbell was intimately involved in the overall development of the system at issue, and is very familiar with its architecture, design, organization, structure, and logic flow. (Plaintiffs’ expert, Ms. Elizabeth Matlack, also described various similarities between the two systems. However, many of these points were credibly explained and refuted by Vafa and, to a lesser extent, by Dr. Howard Tompkins, one of defendants’ experts. Thus, the Court relies to a greater degree for its findings on the testimony and credibility of the principals.)
Campbell testified as to several specific similarities between the two systems. In addition, he described the similarity in the way that the various component parts fit together to form the entire generic utility system. He testified that ICM, through its hard work, put together a unique, winning combination that, while perhaps not on the “cutting edge” of technology, is not found anywhere else except in the DTI system. Specific similarities which the Court finds through the evidence presented at trial include:
1. ICM SEUNIMNT and DTI database manager: both maintain support files and define them similarly, both define interrelationships between support files as reflected in cross-referencing, both include user interface information in association with data files in the respective header files, both have encrypted fields, both reflect similar capacities and constraints, and both contain many of the same fields, values, and elements which are the value sets for the data files.
2. ICM Communications program (Tele-fon) and DTI Communications program: similar dialogue language with all its instructions including how those instructions work against theoretical communications buffers, hooks and ties that link communications to the applications they are supporting, the ability to pass status informations around, the ability to go in and out of protected protocols of different computers, the ability to tie to pri
vately encrypted files and to use confidential passwords, the ability to communicate with different types of Hayes modems from different manufacturers of Hayes modems, and the arbitrary breakdown of the communications files.
3. ICM Menu program and DTI Menu program: both are generic, and have similar methodology, structure, and functionality in accessing files from the menu program directly into the SEUNIMNT and similar interrelationship and communications between the two modules.
4. ICM and DTI report writer programs: both have similar joining of functions as generic and similar leaving of hard coding to other functions.
As to the architecture of the system, or the way in which the various components fit together as building blocks in order to form the unique whole, the Court is satisfied that this was secret information belonging to ICM, not publicly available, and of substantial business value to that company. In addition, the Court finds that defendants used that information in constructing their system. Specifically, Campbell testified that many of the specific functions of the ICM communications program, some of which have been discussed above in describing the similarities between the ICM and the DTI modules, were dictated and necessitated by the way in which that module interacted with SEUNIMNT. He stated that “there was a necessity to build a communications module that did all of these things in a certain way to work as a building block for an overall system.” (Tr. at 153). Similarly, with respect to the Menu module Campbell testified that SEU-NIMNT was defined to “handle a certain category of files, and the Menu system was designed to work with that category of files. So the three — Menu, Communications, and SE Unimnt (sic), really go hand in hand as a part of the overall building block.” (Tr. at 155). In addition, Campbell testified that at the time that Newlin and Vafa were employed by ICM, ICM was having difficulty in designing the Report Writer module so that it would work with the rest of the system. ICM expended substantial time and money in trial and error with different formulations of the Report Writer, and learned that it would not work with the system if it were too complex. Vafa and Newlin were privy to this ICM trial and error process, and when they left to form DTI defendants created a report writer module that, in Campbell’s opinion, made use of the knowledge gained through that “blind alley” process.
The Court is satisfied that the architecture of the ICM system, as embodied in the source code of the several modules and in the manner in which they relate to and interact with one another, was maintained as secret by ICM. Several witnesses testified that ICM maintained all of its source code as secret. Employees, including New-lin and Vafa, were required to sign nondisclosure agreements with ICM. The doors to the premises of ICM were kept locked. Further, plaintiffs have proven that the specific source code, structure, and the interrelationship between the parts of the ICM generic system were not in the public domain. Although defendants presented evidence that several component parts of the programs could be found described in textbooks or other publications, and that certain discreet utilities are available in commercial packages, they have not credibly refuted plaintiffs’ evidence demonstrating that the package as a whole, and the specifications used by ICM to make the parts of that package work together, are not in the public domain.
Newlin and Vafa left ICM on March 13, 1987 and began work at DTI on March 16, 1987. Within two weeks, DTI had created a prototype of its database manager program. When Newlin left ICM, he copied certain files in the machine he was using there onto diskettes, took those diskettes with him, and did not inform ICM that he had done so. Vafa, too, brought with him source code implementing a certain algorithm which he had written for ICM, but later destroyed that file because of questions in his mind about the propriety of using such source code at DTI. Although plaintiffs have not demonstrated that any of the source code taken from ICM by defendants was in fact used by them in
creating the programs at issue in this lawsuit, and although the Court believes Vafa’s testimony that he did not directly make use of any such code, the removal of the code, and the subsequent failure to produce it during discovery, is quite troubling. It may very well be that, in 1987, DTI had in its possession other ICM source code which it no longer has. For the most part, the Court credits the testimony of Newlin and Vafa that the removal of most of the code was inadvertent. However, this incident is an example of the rather cavalier way in which the defendants treated their secrecy and nondisclosure obligations toward ICM. Had they taken these obligations less lightly, they would perhaps have used more care in identifying the files on their personal diskettes before removing them from the ICM premises.
DTI, through Newlin and Vafa, continued to write additional generic programs which operate in substantially the same manner as comparable ICM generic programs. As described earlier, these DTI programs are similar in many significant respects to the ICM programs, and the structure of the entire system is also very similar. The Court does not believe that Vafa consciously “copied” the programs that he had written for ICM in the traditional sense. As defendants’ expert Dr. Tompkins rightly noted, Vafa did not have a “need to copy,” since he was certainly capable of writing the source code in his own right. In fact, he did write the source code for the ICM SEUNIMNT program in C, so it would be clearly erroneous to find that he could not do so. Nonetheless, the Court finds that Vafa, and Newlin as well, did make use of information learned while at ICM concerning which functions and relationships among the modules would and would not work in the generic program. This information was maintained as secret by ICM, and was valuable to its business. Vafa was, in the Court’s judgment, simply too bright to “make the same mistakes twice” in writing his programs. Not general experience, but specific experience with, and knowledge of, those particular types of generic programs, for banking and financial applications, were utilized, perhaps unavoidably, by Vafa and Newlin when they chose to take up the same task at a new company.
CONCLUSIONS OF LAW
The Court notes, first, that “[cjomputer software has received judicial recognition as a trade secret.”
Business Intelligence Services, Inc. v. Hudson,
580 F.Supp. 1068, 1072 (S.D.N.Y.1984) (citing
Matter of Belth v. Insurance Dep’t of New York,
95 Misc.2d 18, 406 N.Y.S.2d 649 (1977) (description of computer program and details of mathematical models, procedures, and statistical assumptions developed by an insurance company constitute trade secrets)).
See also Q-Co Industries v. Hoffman,
625 F.Supp. 608, 617 (S.D.N.Y.1985).
Specifically, a trade secret has been defined in this district to include “any formula, pattern, device or compilation of information which is used in one’s business, and which gives him an opportunity to obtain an advantage over competitors who do not know or use it.”
Rapco Foam, Inc. v. Scientific Applications, Inc.,
479 F.Supp. 1027, 1029 (S.D.N.Y.1979) (quoting Restatement of Torts § 757 comment b (1937)). Among the factors that the Court must consider in determining whether a trade secret exists in a particular case are (1) the degree of secrecy surrounding the process both inside and outside the plaintiff’s business, (2) the plaintiff’s efforts to develop the process and to preserve its secrecy, (3) its value to the plaintiff and to its competitors, and (4) the difficulty with which it could be duplicated by others. In order to prevail on a trade secret claim, plaintiff must prove that it possessed a trade secret, and that the defendant used that secret either in breach of an agreement, confidence, or duty, or as a result of discovery by improper means.
See id.
In this case, defendants do not contest the validity and enforceability of the nondisclosure agreements entered into between ICM and Newlin and Vafa respectively. Thus, if the Court finds that plaintiffs possessed a trade secret, and that it was used by defendants, the above test will
have been satisfied, since such use would clearly constitute a breach of those agreements.
It is well-settled that “a trade secret can exist in a combination of characteristics and components, each of which, by itself, is in the public domain, but the unified process and operation of which, in unique combination, affords a competitive advantage and is a protectable secret.”
Imperial Chemical Industries Ltd. v. National Distillers & Chem. Corp.,
342 F.2d 737, 742 (2d Cir.1965);
Q-Co Industries, Inc. v. Hoffman,
625 F.Supp. 608, 617 (S.D.N.Y.1985). Thus, the first question that must be decided is whether the programs of ICM, either separately or in combination, constitute a trade secret.
Clearly, plaintiffs did make reasonable efforts to maintain the secrecy of the source code of their programs. Their programmers were required to sign nondisclosure agreements, and the doors to the premises were kept locked. Plaintiffs have also demonstrated to the Court the value of the programs to their business, and the significant efforts that went into making the system work as a whole. Thus, the crucial threshold question which must be addressed is the extent to which the information claimed to be secret was in fact available to the public, and the ease or difficulty of duplication of those programs in the absence of knowledge of the allegedly proprietary information.
The Court does not pretend that this is an easy question for it to answer. Much, if not all, of the conflicting evidence on this issue was presented by experts employing jargon which at times sounded like a foreign language. On the other hand, the Court is mindful of the dilemma faced by the parties: in attempting to bring the explanation of the programs down to a level of simplicity understandable to one unskilled in the computer programming field, they risk abandoning much of the force of their detailed explanations. Defendants’ experts and defendants themselves (two of whom are computer scientists and thus “experts” in their own right), presented testimony and evidence to the effect that the various components of the ICM system are not in any way secret, but are available to the public through books, commercially-sold utilities, and scholarly publications. Plaintiffs, on the other hand, presented testimony showing that although one could purchase commercial products which performed the utilities at issue, the ICM generic utilities were constructed in a specific manner in order that they be especially suited to the task they were to perform in a specific context. In addition, Campbell testified that the several modules were constructed so as to fit and work together in a particular way. As another court concluded in a case almost factually identical to the one presented here,
at least to some degree, defendants’ position is correct in that some of the approaches utilized in the [ ] system, standing separately, are general concepts known to experts in the computer industry. The cases indicate, and plaintiff concedes that general concepts are not protectable [sic], per se, as trade secrets. However, the cases further indicate that while general concepts are not protecta-ble [sic], the specific implementation involving a particular combination of general concepts may well amount to a trade secret.
Cybertek Computer Products, Inc. v. Whitfield, et al.,
203 U.S.P.Q. 1020 (BNA) (Ca.Super.1977). The court in
Cybertek
relied on a case in which “the specifications of these basic mechanical elements and their relationship to each other ... were not publicly known, were arrived at only through painstaking research and extensive trial and error, and, therefore, constituted a trade secret entitled to protection. In other words, the court found that the specific embodiment of the general concepts and approach into a combination of parts was protectable [sic], even though all or some of them might well be known to the industry.”
Id.
Here, as in that case, the facts present a proper situation for application of that principle. It is clear to the Court that this particular combination provides a competitive advantage to ICM, and thus that it is entitled to protection from misappropriation. The Court finds
that Campbell’s testimony in this regard was as specific as reasonably possible in light of the necessity that he speak in a language understandable to a trier of fact uninitiated in computerese. In addition, the Court credits his testimony that a significant amount of time and money was spent in investigating alternatives that, in the end, were not fruitful. Such a trial and error process is also protectible as a trade secret of ICM. Thus, in resolving this admittedly difficult question, the Court finds that plaintiffs have met their burden of demonstrating that the combination of the several modules, as well as the specific source code of each individual module, constituted a trade secret of ICM.
Having found the existence of a trade secret, and noting that there is no dispute concerning the fact that defendants had access to that secret information and in fact created much of the source code embodying it, the Court must next decide whether defendants have misused that secret information in creating their own generic system. This presents a close question of fact and a difficult issue of public policy. The difficulty results from the need to preserve the proprietary information of ICM, while at the same time not unduly restricting Newlin and Yafa from utilizing their admittedly extensive skills, experience and general knowledge. This is especially the case where, as here, there exists a situation in which no direct copying occurred. The Court is convinced that Vafa, and to a lesser extent Newlin, sincerely attempted to refrain from using ICM source code and other information used by them while working at ICM. Yet it is equally convinced that the two programmers benefitted from their specific knowledge of the architecture and detail of the ICM programs in creating substantially similar programs for DTI.
Other courts have grappled with this issue, and have relied on certain factors in attempting to fashion a just result.
See generally,
Developments in the Law—Competitive Torts, 77 Harv.L.Rev. 888, 950-53 (1964) (recommending balancing approach in dealing with difficult issue of the technically-skilled employee). Certainly, “[a]n employee who achieves technical expertise or general knowledge while in the employ of another may thereafter use that knowledge in competition with his former employer, so long as he does not use or disclose protected trade secrets in the process.”
Wilmington Trust Co. v. Consistent Asset Management Co., Inc.,
No. 8867, 1987 WL 8459 (Del.Ct.Chanc.1987) (slip op.).
One important factor often considered is whether there existed an agreement between the employee and the former employer claiming misappropriation of trade secrets. This factor is important in two respects. The first, not relevant here, whether the process developed by the employee belonged to the employee, or instead belonged to the employer. That question is not at issue in this case, since defendants concede that the ICM programs written by Yafa and Newlin are the property of ICM. A second reason courts look to the existence of a contract is to help to balance the interests of former employers and skilled employees. The existence of a nondisclosure agreement puts the employee on notice that the programs are considered trade secrets.
See, e.g., Aries Information Systems, Inc. v. Pacific Management Systems Corp.,
366 N.W.2d 366, 369 (Minn.App.1985) (confidentiality agreements put employees on notice that information contained in computer programs was considered secret);
Jostens, Inc. v. National Computer Systems, Inc.,
318 N.W.2d 691, 702 (Minn.1982) (same). In this case, Vafa and Newlin were well aware of their duty not to disclose secrets of ICM. In addition, Mitsos himself helped to draft the agreements, and thus was also aware of this duty. Furthermore, it appears from the evidence adduced at trial that the three individual defendants explicitly discussed their duties toward ICM, yet resolved to create similar software for DTI while attempting not to use secrets of ICM.
Defendants assert that the DTI programs were created “from scratch,” and that Vafa and Newlin approached their tasks at DTI “with a fresh mind.” It is a well-recognized principle that, where a defendant in a trade secret case claims inde
pendent development, the burden shifts to the defendant to show that this was in fact the case.
See, e.g., Rapco Foam, Inc. v. Scientific Applications,
479 F.Supp. 1027, 1030-31 (S.D.N.Y.1979);
Cybertek Computer Products, Inc. v. Whitfield,
203 U.S.P.Q. 1020 (BNA) (Ca.Super.1077). Defendants have not sustained this burden.
The Court does not doubt the good intentions of Vafa. Yet good intentions may not be enough to protect ICM. In continuing a preliminary injunction prohibiting a former employee from working on a similar computer system for a competitor, another court has noted in this respect:
While it recognizes Boddie’s good intentions and professional integrity, [plaintiff] contends that it will be impossible for Boddie to develop a competing system for ATUSA without dwelling upon his experience in having done the same thing for [plaintiff] and that in the process, as part of properly performing his duties for his new employer, he will of necessity incorporate into any system for ATUSA many of the proven-workable features of the [plaintiffs’] systems, which features legally belong to [plaintiff], which are confidential and vital to the continued superiority of its business product, and which therefore, constitute trade secrets....
American Totalisator Systems, Inc. v. Automatic Totalisators (U.S.A.) Ltd.,
No. 5562, 1978 WL 4479 (Del.Ct.Chanc.1978) (slip op.). This argument is persuasive, and applicable to the instant case.
ICM does not attempt to prevent defendants from competing with it on a fair basis. It seeks only to prevent them from creating a substantially similar system of utility programs, the source code of which was in part written by them for ICM, within weeks after leaving its employ. Defendants’ expert, Dr. Tompkins, admitted that there are many similarities in the structure, functionality, organization, and logic flow between the ICM and the DTI programs. It has been represented to the Court by both parties that the four programs at issue are among approximately 1,000 programs or modules created by defendants. ICM has no complaint concerning the balance of these programs. In light of the equities presented, and the inadequacy of damages at law to compensate ICM for the misappropriation of its valuable trade secrets, injunctive relief is appropriate.
CONCLUSION
Accordingly, defendants are enjoined, for a period of six months from the date of this decision, from utilizing as part of DTI’s systems any versions of its database manager, menu, communications, and report writer programs created in whole or in part by either Vafa or Newlin. In addition, Vafa and Newlin are enjoined from contributing to the creation of any new programs embodying the above-mentioned utilities for the same period. DTI is otherwise free to develop new programs, provided that it does not in any way rely on the restricted programs in this development. The period of six months has been chosen with a view to the length of time ICM spent in creating its systems, the increased speed of the current hardware available to programmers, and the need to neutralize the “head start” gained by DTI from the improper use of ICM’s trade secrets. This injunction, due to its limited scope, does not prevent the individual defendants from employing their extensive skills in their new business or from earning their livelihoods.
Settle permanent injunction on notice.