Computer Associates International, Inc. v. Bryan

784 F.Supp. 982 (1992)

COMPUTER ASSOCIATES INTERNATIONAL, INC., Howard A. Rubin and Howard Rubin Associates, Inc., Plaintiffs,
v.
David W. BRYAN and Application Development Technologies, Inc., Defendants.

No. CV 91-0195(ADS).

United States District Court, E.D. New York.

January 29, 1992.

Gold, Farrell & Marks, New York City (Jane G. Stevens, Mark N. Diller, of counsel), for plaintiffs.

Cullen and Dykman, Garden City, N.Y. (Peter Mastaglio, John P. McEntee, William F. Logan on the brief, of counsel), for defendants.

DECISION AND ORDER

SPATT, District Judge.

Claiming misappropriation of trade secrets and confidential and proprietary information, the plaintiffs move, pursuant to Fed.R.Civ.P. 65(a), for a preliminary injunction to prevent the defendants from misappropriating, using, releasing or reproducing the computer software system known as "CA-ESTIMACS" or "ESTIMACS". In this regard, the plaintiffs also claim that the defendant David W. Bryan breached contractual and fiduciary duties owed to the plaintiffs by the creation and marketing of the defendants' product known as AD/TEC-SYS ("AD/TEC").

FACTUAL BACKGROUND

The defendant David W. Bryan ("defendant" or "Bryan"), was employed by the plaintiff Computer Associates International, Inc. ("Computer Associates" or "CA"), for approximately seven years starting in 1982. CA is a Delaware corporation with its principal place of business in Garden City, Long Island, New York. Upon commencing his employment with CA as a sales representative in the Chicago office, Bryan signed an employment agreement, in which he agreed not to use or misappropriate any trade secret or confidential information for his own benefit or for the benefit of any other person or entity. Upon his departure in July, 1989, Bryan also signed a "departing employee agreement", which again reiterated that he would not disclose or use any trade secrets or confidential information gathered during the course of his employment with CA.

In 1981, the plaintiff Howard A. Rubin ("Rubin"), through his wholly-owned corporation Rubin Systems, Inc., began marketing a computer software product that permits users to estimate the time, staff, risks, money required and type of computer hardware it would take to develop new software or modify existing software applications. This system was made up of several products, based loosely on an earlier IBM model, and was called "Quest".

In 1982, by agreement, Rubin granted a license to Management and Computer Services, Inc. ("MACS"), giving MACS exclusive rights to use and sell "Quest" in exchange for royalty payments. By agreement in 1985, Computer Associates acquired all of the MACS rights under the agreement with Rubin. Since that time, CA has used the tradenames "CA-ESTIMACS" and "CA-PLANMACS" for that product. CA alleges that it has invested millions of dollars in improving, designing, enhancing and guarding these programs, and that they contain valuable trade secrets and confidential information.

From mid- to late-1987, Bryan was a member of the CA-ESTIMACS sales force and later became an ESTIMACS product manager responsible for marketing the product. During 1988 and 1989, while in the employ of CA, Bryan worked on software similar to the ESTIMACS system which would also permit a user to estimate the effort, cost and risks in developing software programs. He called this program "HIQ". Bryan demonstrated the HIQ product to various co-employees at CA.

Bryan left the employ of CA in July, 1989 and ultimately formed a corporation known as Application Development Technologies, Inc. ("ADT"), a California corporation located in Dallas, Texas, which is also a defendant in this case. Bryan is the President and the sole shareholder of ADT. While at ADT, Bryan has been testing and marketing the AD/TEC program, which, according to CA, utilizes the trade secrets embodied in CA's programs. CA also contends that the AD/TEC program is actually based on the HIQ product developed by Bryan while in the employ of CA.

The plaintiffs commenced this action against Bryan and ADT, alleging causes of action based on misappropriation of trade secrets, breach of employment agreement, breach of fiduciary duty and unfair competition. The plaintiffs seek compensatory damages, punitive damages and injunctive relief. At this juncture, the plaintiffs move for a preliminary injunction pursuant to Fed.R.Civ.P. 65(a), to prevent Bryan and ADT from misappropriating and using CA's trade secrets, specifically the trade secrets combined and contained in the CA-ESTIMACS program, allegedly used to create and develop the defendants' AD/TEC program.

In opposition to this motion for a preliminary injunction, the defendants allege the following: (1) the plaintiffs' CA-ESTIMACS system does not in fact contain "trade secrets"; (2) the AD/TEC program is substantially different from the ESTIMACS program; (3) Rubin developed the ESTIMACS system by using printed material readily available to anyone in the trade and he did not conceive any "trade secrets" in the ESTIMACS system; (4) the employment contract is unconscionable and therefore invalid; (5) Bryan did not breach the employment agreement, since ADT did not do business until more than one year after the termination of Bryan's employment; and (6) the plaintiffs' delay in seeking to obtain an injunction undermines any claim of "irreparable harm." Furthermore, the defendants contend that the plaintiffs have not demonstrated irreparable harm.

WHAT IS ESTIMACS?

ESTIMACS is a software product which estimates the resources required to develop new software or modify existing software. As good an explanation as any provided to the Court of the purpose of ESTIMACS is set forth in the testimony of Marion Harlow, as follows:

"Q Perhaps you can describe for the Court an example you would give at the seminars that you gave.

A We always used the same example. It was a financial system. In a financial system you have accounts receivable, accounts payable, you have a billing system, you have a general ledger.

Q Please slow down.

A I am sorry. Those are all the different what would be considered subsystems of a financial system. And what the developers would actually do is to code — to actually write programs in COBOL or PL-1 or whatever the computer language is, to actually allow the computer to compute the billing for the company. That's the best I can do in layman's terms.

Q And Estimacs would be able to tell you what with respect to developing this system?

A How many woman hours or staff hours it would take to build the project. It would also tell you how many people were going to be required, insofar as how many project leaders, how many system analysts, how many programmers, how much it was going to cost and how long it was going to take" (Tr. at pp. 1128-29).

PROCEDURAL SETTING

On January 18, 1991, this Court signed an Order to Show Cause, scheduling a hearing on the motion for a preliminary injunction for January 25, 1991. At that time, the Court also temporarily sealed the affidavits and exhibits in support of the motion due to the alleged trade secrets and confidential information contained in the papers.

At the request of counsel, the commencement of the hearing was adjourned several times to permit the defendants to file papers in opposition. The Court thereafter began the hearing on the motion for preliminary injunction on February 22, 1991, and intermittently continued the hearing over the course of several months until its conclusion on July 19, 1991. Upon the conclusion of the hearing, in order to assist the Court in its determination, the parties submitted proposed findings of fact and conclusions of law (see Rule 23 of the Civil Rules of the United States District Court for the Eastern District of New York). In the final stage of the hearing, summations were heard by the Court on October 7, 1991.

After the hearing was terminated, the parties continued to submit post-hearing letters, affidavits and memoranda of law. By letter dated November 22, 1991, the Court advised counsel that it would not consider the post-hearing submissions or any issues raised therein, in the absence of a formal motion on notice (cf. United States v. Wallach, 935 F.2d 445 743 n. 1 [2d Cir.1991] [Altimari, J., concurring] [post-argument submissions on appeal are "looked upon with extreme disfavor"], citing United States v. Bortnovsky, 820 F.2d 572, 575 [2d Cir.1987] [per curiam]).

On December 5, 1991, a motion was made by the defendants for leave to supplement the record on the ground that the defendants "released a new version of the AD/TEC-SYS program referred to as AD/TEC-SYS 3.0" which program includes certain changes from the prior AT/TEC products. On January 17, 1992, the Court denied the motion without prejudice in an oral decision, essentially finding that the motion was premature and there was no reason to supplement the record with new facts which occurred subsequent to the conclusion of the hearing. The Court determined that after this lengthy and complex trial both sides were entitled to a decision on the merits of the charges and defenses.

In the interim, without waiving any rights, counsel for the plaintiffs consented to the defendants selling and marketing the subject products.

Pursuant to Fed.R.Civ.P. 52(a), set forth below are the Court's required findings of fact and conclusions of law (see Tekkno Laboratories, Inc. v. Perales, 933 F.2d 1093, 1097 [2d Cir.1991]; Society for Good Will to Retarded Children, Inc. v. Cuomo, 902 F.2d 1085, 1088 [2d Cir.1990] see also Weitzman v. Stein, 897 F.2d 653, 658 [2d Cir.1990] [district court "must" set forth reasons for grant or denial of preliminary injunction]).

APPLICABLE LAW

A. Jurisdiction

Subject matter jurisdiction over this action exists pursuant to 28 U.S.C. § 1332, since the plaintiffs and defendants are of diverse citizenship, and the amount in controversy exceeds $50,000.

B. Standard on Preliminary Injunction

The grant of preliminary injunctive relief is considered an "extraordinary" (Patton v. Dole, 806 F.2d 24, 28 [2d Cir. 1986]), and "drastic" (Borey v. National Union Fire Ins., 934 F.2d 30, 33 [2d Cir. 1991]) remedy. Accordingly, in order to obtain a preliminary injunction in the Second Circuit, it is now well established that the movant must clearly establish the following required elements:

"(1) irreparable harm or injury, and (2) either (a) a likelihood of success on the merits or (b) sufficiently serious questions going to the merits to make them a fair ground for litigation and a balance of hardships tipping decidedly in favor of the movant" (Jackson Dairy, Inc. v. H.P. Hood & Sons, Inc., 596 F.2d 70, 72 [2d Cir.1979] [per curiam]).

(See also Western Publishing Co. v. Rose Art Indus., Inc., 910 F.2d 57, 59 [2d Cir. 1990]; LeSportsac, Inc. v. K Mart Corp., 754 F.2d 71, 74 [2d Cir.1985]; Vera, Inc. v. Tug "Dakota", 769 F.Supp. 451, 454 [E.D.N.Y.1991]; Ecolab Inc. v. Paolo, 753 F.Supp. 1100, 1109 [E.D.N.Y.1991]).

A showing of probable irreparable harm is usually considered the single most important requirement (see Reuters Ltd. v. United Press International, Inc., 903 F.2d 904, 907 [2d Cir.1990]), and an applicant must establish more than a mere "possibility" of irreparable harm, namely, "that it is likely to suffer irreparable harm if equitable relief is denied" (JSG Trading Corp. v. Tray-Wrap, Inc., 917 F.2d 75, 79 [2d Cir.1990] [emphasis in original]). Where money damages are adequate compensation, a preliminary injunction will not issue (see Jackson Dairy, Inc. v. H.P. Hood & Sons, Inc., supra, 596 F.2d at p. 72).

Of particular relevance to this action is the recognition by the Second Circuit that the loss of "trade secrets" is not measurable in terms of money damages (see FMC Corp. v. Taiwan Tainan Giant Indus. Co., 730 F.2d 61, 63 [2d Cir.1984] [per curiam] ["A trade secret once lost is, of course, lost forever"]), and is thus considered "irreparable harm."

Moreover, the potential loss of an industry leader's present market and loss of the advantage of being the pioneer in the field and the market leader, may constitute irreparable harm. (See, for example, Natural Organics, Inc. v. Proteins Plus, Inc., 724 F.Supp. 50 [E.D.N.Y.1989]).

As to "likelihood of success," the movant "need not show that success is an absolute certainty. It need only be shown that the probability of the movant prevailing is better than fifty percent" (Abdul Wali v. Coughlin, 754 F.2d 1015, 1025 [2d Cir. 1985]).

C. The Defense of Laches

The defendants contend that the plaintiffs' delay in seeking this preliminary injunctive relief constituted laches. Even though delay may not rise to the level of "laches", it may nonetheless indicate the absence of irreparable harm (see Majorica, S.A. v. R.H. Macy & Co., 762 F.2d 7, 8 [2d Cir.1985]). As the Second Circuit noted in another context namely, a trademark infringement action:

"Preliminary injunctions are generally granted under the theory that there is an urgent need for speedy action to protect the plaintiffs' rights. Delay in seeking enforcement of those rights, however, tends to indicate at least a reduced need for such drastic, speedy action. Gillette Co. v. Ed Pinaud, Inc., 178 F.Supp. 618, 622 (S.D.N.Y.1959). Accord Le Sportsac, Inc. v. Dockside Research, Inc., 478 F.Supp. 602, 609 (S.D.N.Y.1979)."

(Citibank, N.A. v. Citytrust, 756 F.2d 273, 276 [2d Cir.1985]).

The Court finds that the plaintiffs did not delay to such an extent that laches is a bar to the preliminary injunctive relief sought. The record demonstrates that in the time period prior to commencing this action, the plaintiffs conducted an extensive investigation in order to gather the necessary facts required to support this complex action. (See, for example, Borey v. National Union Fire Ins. Co., 934 F.2d 30 [2d Cir.1991] ["undue delay" of three years]; Gilder v. PGA Tour, Inc., 936 F.2d 417 [9th Cir.1991] [the Court found the plaintiffs had pursued their claim with reasonable diligence in taking five to seven months to file for a preliminary injunction]).

D. Unfair Competition Based on Misappropriation of Trade Secrets

Circuit Judge Altimari recently enunciated the law of misappropriation of trade secrets, also in the context of an action involving computer software, as follows:

"A plaintiff claiming misappropriation of a trade secret must prove: `(1) it possessed a trade secret, and (2) defendant is using that trade secret in breach of an agreement, confidence, or duty, or as a result of discovery by improper means'"

(Integrated Cash Mgt. Servs., Inc. v. Digital Transactions, Inc., 920 F.2d 171, 173 [2d Cir.1990] [citations omitted]).

The Second Circuit in Integrated Cash, supra, approved the definition of a "trade secret" from the Restatement of Torts § 757, comment b, which has been adopted as the law in New York, and provides:

"`A trade secret may consist of 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'" (920 F.2d at p. 173 [citations omitted]).

Under New York law, the following factors are considered relevant in determining whether a "trade secret" in fact exists:

"(1) the extent to which the information is known outside of his business; (2) the extent to which it is known by employees and others involved in his business; (3) the extent of measures taken by him to guard the secrecy of the information; (4) the value of the information to him and to his competitors; (5) the amount of effort or money expended by him in developing the information; (6) the ease or difficulty with which the information could be properly acquired or duplicated by others"

(Eagle Comtronics, Inc. v. Pico, Inc., 89 A.D.2d 803, 803-04, 453 N.Y.S.2d 470, 472 [4th Dep't 1982] quoting Restatement of Torts § 757, comment b).

"The most important consideration remains whether the information was secret.... (and) Secrecy is a question of fact" (Lehman v. Dow Jones & Co., 783 F.2d 285, 298 [2d Cir.1986]).

Chief Judge Platt also recently discussed misappropriation of trade secrets under New York law, as follows:

"An aggrieved plaintiff need not show that the information it seeks to protect is vital to its business, but only that the information would provide the unauthorized user of it with an unfair competitive advantage which it would not otherwise have enjoyed. Greenwich Mills Co., Inc. v. Barrie House Coffee Co., Inc., 91 A.D.2d 398, 405, 459 N.Y.S.2d 454, 459 (2d Dep't 1983). In determining whether matter is protectable as a trade secret, courts frequently examine the method by which defendant acquired it, e.g., such as by stealing or copying. Ferranti Elec., Inc. v. Harwood, 43 Misc.2d 533, 251 N.Y.S.2d 612, 619-620 (Sup.Ct.Nassau County 1964).

* * * * * *

Moreover, even where the information would not otherwise qualify as a trade secret, the unauthorized physical taking and exploitation of internal company documents, including detailed customer information by an employee for use in his future business or employment is to be enjoined as unfair competition ..."

(Ecolab Inc. v. Paolo, supra, 753 F.Supp. at p. 1111 [citations omitted]).

Also, "a former employee can be prohibited from taking trade secrets from his former employ[er] whether or not there exists an employment contract which prohibits competition" (Le Juda, Inc. v. Diversified Recreations, Inc., N.Y.L.J., Nov. 25, 1991, p. 25, col 6 [Sup.Ct.N.Y.County 1991], citing Hecht Foods, Inc. v. Sherman, 43 A.D.2d 850, 351 N.Y.S.2d 711 [2d Dep't 1974]).

Matters of public or general knowledge in an industry are incapable of being designated as a "trade secret" (see Minnesota Mining and Mfg. Co. Inc. v. Technical Tape Corp., 23 Misc.2d 671, 192 N.Y.S.2d 102 [Sup.Ct.Westchester County 1959], aff'd, 15 A.D.2d 960, 226 N.Y.S.2d 1021 [2d Dep't 1962]). However, trade secrets may nevertheless exist where the information consists of an accumulation or combination of generic computer programs that are in the public domain, but are linked together, share information or form a unique combination in a way not generally known outside the business or industry (see Integrated Cash Mgt., supra, 920 F.2d at p. 174). That is, even though independently each of the "building blocks" may consist of general or public knowledge, collectively the various components may form a "unique whole" (id.).

Further, with regard to an employment agreement restricting the use of confidential information, such as the employment agreements at issue, under New York law, an agreement is enforceable if the restriction is reasonable in scope and duration and necessary to protect against the disclosure of trade secrets or other confidential information (see Business Intelligence Services, Inc. v. Hudson, 580 F.Supp. 1068 [S.D.N.H.1984]; Continental Group, Inc. v. Kinsley, 422 F.Supp. 838, 843 [D.Conn.1976]; Reed Roberts Assoc., Inc. v. Strauman, 40 N.Y.2d 303, 308, 386 N.Y.S.2d 677, 680, 353 N.E.2d 590 [1976]; Purchasing Assocs., Inc. v. Weitz, 13 N.Y.2d 267, 272, 246 N.Y.S.2d 600, 604, 196 N.E.2d 245 [1963]).

Mindful of these principles of law, the Court turns to the evidence adduced at the hearing.

THE HEARING

The following summarizes the testimony and evidence adduced at the hearing, which constitutes the Court's findings and conclusions (see Fed.R.Civ.P. 52[a]).

Plaintiff HOWARD A. RUBIN, holds a Ph.D. in computer science and is a Professor and Department Chair of Computer Science at Hunter College. Rubin testified that he is the creator of CA-ESTIMACS and CA-PLANMACS. He is retained by Computer Associates as a consultant in product development.

The idea of an estimating system originated when the question was posed to him by a consulting client in the late 1970's. At that time, he looked at the feasibility of being able to estimate the total system development effort with regard to high level business questions. He developed these systems to answer certain key questions, namely "How long it will take and what will the cost be, what are the ... risks ... associated with the software development and also (the) maintenance process" (Tr. at p. 30). He explained that "developing" a software system means to build such a system when none existed before. "Maintenance" means changes to an existing system, namely, to have it perform new or advanced functions or correct defects. The software system at issue allows the user to estimate key dimensions of creating software in effort, staff, duration, risk and financial issues "and a whole host of other technical factors."

Another dimension to this software system is not only to allow the user to estimate what it takes to do this work, but to allow the user to look at different ways of construction and at the relevant impact, which is called a "life cycle." As explained by Rubin:

"Q At each of these life cycles that you identified, they appear in the Estimacs product and is capable of arriving as an estimate of each of the life cycles?

A Yes. An estimate of the effort in each of the life cycles, the costs associated with each of the life cycles, the kinds of staff that are needed in each of the life cycle, the number of staff people that are needed in each life cycle phase. So Estimacs in fact gives each user a complete picture of all estimate dimensions, plus the staffing and allows it to look with the life cycle to see what the proposed projects look like." (Tr. at p. 37).

The clearest definition of a life cycle was given by the defendant Bryan as follows:

"Q Could you describe for the Court what a life cycle is. Slowly, please.

A A life cycle within the context of application development would be the steps and procedures taken from the beginning of a project's conception to the time it is finally installed and approved and all of the steps that it takes to do that" (Tr. at p. 1463).

A key element of ESTIMACS is the use of a dialogue where the user answers a series of questions, which Rubin explained as follows:

"It focuses on effort projection, staff cost and duration projection, risk, financial and a host of other factors.

The way that a user applies Estimacs, is that the Estimacs product interacts with its user through a dialogue in which they are asked a series of questions. Estimacs was designed to ask questions that are available very early in the process of developing systems. This was specifically done to contrast with the way Estimacs tools have worked prior to the existence of Estimacs" (Tr. at p. 38) (emphasis supplied).

The effort model of ESTIMACS allows its users to describe the software system they wish to construct and the life cycle they intend to use. Upon answering questions and running through formulas, "ESTIMACS can then predict the expected effort required both at the total project level and at each life cycle phase."

In addition, ESTIMACS has a staffing and cost model, which predicts the duration of the project, what size team should be deployed to carry out that effort, over what period of time, at what cost and what skilled classification of persons are needed, including managers, analysts and programmers.

Significantly, there are "equations" used in the staffing and cost models and in the maintenance model. These equations have never been published and only appear in the "source code" for the product.

Another term in the computer lexicon is the "architecture," which refers to structure of the product — how its individual components are organized; how they relate together; and how someone interacts with the products.

Asked to describe the architecture of ESTIMACS, Rubin responded:

"Estimacs has an architecture that has models for effort. It has a unique model for predicting the maintenance efforts which is the unique feature of the architecture. Has a model for cost, a model for risk, another model for what we call portfolio impact, another model dealing with financial constraints and considerations, another piece that deals with a technical aspect of computing which is given the name function points spelled as it sounds, function points.

It is a method that scientists at IBM developed to understand the size of the computer system.

It has another model that deals with assessing the value of a computer system to a business organization.

These are what I call the architectural building blocks of Estimacs.

The other aspects of the architecture is how those components feed each other or interfaces would be more of a jargon word, but feed each other, and how a person interfaces, works with them would be a better word perhaps.

The other basis of architecture for Estimacs I would sort of classify is look and feel how it works with somebody, where you answer a set of questions and you see the results of answering those questions in terms of graphic feedback. And then you are given some analysis of how your answers influenced the result you got ..." (Tr. at pp. 42-4).

Rubin testified that until the development of defendants' product (AD/TEC), no other estimation product was available in the market that utilized architecture similar to that of ESTIMACS.

"Estimacs is the only product that allows the user — until the defendant's products — that allows the user separate entries to let me look at effort, let me look at maintenance separately and then piece the puzzle together. So it is the architecture which is very, very unique" (Tr. at pp. 44-5).

The software was developed by Rubin beginning in 1978. He developed the hypothesis that there is a way to estimate projects from business based information early in the software development and maintenance life cycle. He then embarked on a "multi-point" research program involving models in universities and in commerce. After this research he concluded that, at that time, there was no research or products specifically addressing the issue of estimating entire life cycles from a business base.

Initially, Rubin entered into a contract with a company named Management In Computer Service, known as MACS; (hence, the origin of the "macs" in ESTIMACS). When MACS assigned his contract to Computer Associates, a one page document was signed stating that the parties would abide by all the trade secrets and confidentiality in the original contract between Rubin and MACS.

Getting to a crucial part of the case, Rubin explained his use of an IBM book which is designated as defendants' exhibit "A". This book was copyrighted by International Business Machines Corp. in 1980. It is part of the "SRA Management Library" and is entitled "Estimating Application Development Projects," a Personal Reference Guide ("The IBM Manual").

A major thrust of the defense in this case is that Rubin obtained important information from the IBM manual and this material, available to anyone who purchased the manual, rather than originally conceived material, was used to create ESTIMACS.

Rubin himself described his use of the IBM manual, as follows:

"Now, in acquiring materials that I got the first step and research is typically to recreate the results of other researchers. And in going through all the material I culled, and this goes from the period of 1978 to 1981, the most promising piece of work I found in the area of effort by work done by IBM that was published in a book by IBM called Estimating the Application Develop Process. In that book IBM had a question set which was used to estimate how long it would take to design a system, not to design and dole the system, just to design the system.

That looked like the most promising piece of work to springboard off of. The limitations of that work had to do with, again, design focus only and only in the context of a traditional life cycle.

The other issue with regard to it is IBM only provided the question set of that model to a small number. And what I did then is went through the steps of recreating other researcher's work by as best as possible tracing down the mathematics that they did and applying it to the day to day use to see if I was getting the same results.

The next step, and this was specifically with regard to the IBM model—and let me describe the IBM model. The IBM model for effort consisted of a list of 25 questions in a specific order, and where each question had what is called a waiting factor associated with it. So you give the answers to the questions, and multiply that answer to the questions by a number and feed that into an equation. After feeding the numbers into the equations the results would be an estimate of design effort.

IBM had a very, very small sample size. Nor did they have any kind of accurate method in places to go out and estimate full development effort from that.

What I then decided to do was to attempt to use that question set as a basis for estimating full life cycle development. And that this was not an easy task.

. . . . .

Now, the steps that I went through, IBM had 25 questions. Each question had a weight associated with it and a series of questions with respect to design effort. These equations, of course, worked over the IBM sample data. But IBM did not attempt to replicate reliably total system development effort.

I then obtained project data from a number of sources to try to build the model over a broader base.

. . . . .

[t]hrough other activities I had done through the consulting field through Bell Labs and other sources, I was able to consult data and bring up a total base of several hundred projects in which I then attempted to redo the IBM question set and make it estimate total project development effort.

What this involved was extensive work. And the reason the work is extensive because imagine for a moment you have 25 questions which were part of that set. And the issue in developing an estimation model was to ordering the questions so a user can answer them in a sensible fashion; secondly, determining the appropriate weights for the question; thirdly, developing coefficients to successfully translate the design estimates into a total system estimate within an acceptable tolerance that was reliable" (Tr. at pp. 49-51) (emphasis supplied)

In sum, Rubin testified that he developed ESTIMACS by scanning the existing research, acquiring data from his own source and other sources in the field and selecting the IBM manual question set as a basis for building a total life cycle productive model. Thus, he evolved new life cycles through the addition of new capabilities. This occurred between 1978 and 1981.

In 1981 Rubin marketed his new estimation product under the name of Quest (short for quick estimator). At that point the product consisted primarily of the effort model with traditional life cycle information. He had good marketing success with the product. By 1982 he expanded the product to include a staffing and cost estimation model.

In 1982, gaining feedback from hundreds of users, Rubin continued to improve his product. At that time, he entered into the contractual relationship with MACS. He then added a software maintenance model, which is an estimating system to compute what it takes to change existing software systems. To achieve this result Professor Rubin had to transform the initial question set into a parallel maintenance question set. He developed a new set of weights and created a maintenance model, which he described as follows:

"I would say that this is probably one of the bigger breakthroughs in some sense than the development model because there is no published reference anywhere to using the IBM question set in any literature, IBM or otherwise, to estimate software maintenance" (Tr. at p. 58).

In 1985, the company Management & Computer Service was acquired by the plaintiff Computer Associates and the products were transferred to CA. At that point other changes were made in the product. A "function point" feature was added "for sizing of the computer system." Also introduced was the concept of alternative life cycles. According to Rubin, the product continued to evolve from 1981 with only one module or component to more than nine modules or components at the present time.

The new mathematics developed in connection with the maintenance model have not been published, but the question set was published in an article in 1985. According to Rubin, the new mathematics do not appear anywhere in print.

"Q Have the new weights and coefficients that you have developed in connection with the development of the maintenance model of Estimacs ever been published, Professor Rubin?

A No. The new weights and coefficients has not been. The only thing that has been is through a professional society publication we described what the question set looks like, describing what the questions we asked.

THE COURT: The new weights and what?

MS. STEVENS: Coefficients.

A The details of the model do not appear anywhere in print.

Q If those are also only embodied in the source code?

A Only embodied in the source codes.

Q Is the source code provided to the licensee of the product?

A Definitely not. It is a matter of Computer Associates' policy" (Tr. at p. 62).

As stated previously, a "life cycle" is a way of constructing a system. Life cycles have steps called phases. Rubin developed the coefficient for ESTIMACS by his research and observations. These coefficients did not appear in the IBM literature and they have never been published without prohibition on disclosure or copyright notices.

The target markets of ESTIMACS are large commercial organizations such as banks, insurance companies and manufacturers who develop or maintain internal computer systems. The product is typically acquired by management level executives who are called project managers.

On cross-examination, Rubin testified that ESTIMACS is the only product of its kind that works from a high level question set and does not require lines of code or function point information as its primary input.

Capers Jones is another expert in this field who developed estimation products called SPQR/20 and checkpoint, which are of different architecture and have other features different from ESTIMACS. In the market, ESTIMACS has ten times more users than SPQR, which is its major competitor. Rubin described some of the unique features of ESTIMACS, as follows:

"Q Thank you.

Would you say one of the unique qualities of Estimacs is the series of nontechnical and general questions?

A I would say the unique aspects of Estimacs is the ability to turn the answers to those non-technical questions into total life cycles. The existence of a set of questions doesn't make it unique. The ability to turn it into an outcome of value to somebody so it would pay for the product is of worth" (Tr. at pp. 84, 85)

Rubin again conceded that the original source of the questions is the IBM manual and that he reproduced the IBM question set in ESTIMACS with changes in the order of the questions and in the weights of several of the questions. He changed the IBM weights because ESTIMACS was designed to estimate "total project estimates or effort" (Tr. at p. 90). He further defined a weight as "a multiplier for another number that adjusts it into a linear equation or non-linear equation" (Tr. at p. 89).

Significantly, Rubin testified that these weights or multipliers were never published. Further he stated that he recast the IBM material "to come up with new weights and new coefficients" (Tr. at p. 92).

Again, Rubin testified about his changing the IBM material, as follows:

"Q What is a core set when you say that?

A We look at groups of equations. So my mission in this was to use the IBM ingredients as a base and to be able to generate something new from it.

So, we have the IBM question set, the 25 questions, reordered and reweighted in a significant fashion. And the results from that went into a set of equations, the same base set that IBM used, and then modified ... to be able to use total estimation, and then mapped into the life cycle distribution which we developed" (Tr. at p. 93).

This testimony as to changes made by Rubin to the IBM manual material was never refuted.

Rubin defined another highly technical and important part of the ESTIMACS product called the "Source code," as follows:

"Q Give us a layman's definition of what a source code is?

A A course [sic] code for a computer product is the actual programing language listings of the instructions to the computer.

THE COURT: Just one minute now. The actual —

THE WITNESS: Programing language list of the instructions that have been written by the programmer to enact the product to make it work. It is readable and it has to be translated into an internal machine language typically to be of use, and it is also another process which is more abstract which is called interpretation.

THE COURT: You say if it is readable why must it be translated?

THE WITNESS: It is readable to people. Computers basically work in numeric code. So you write something which is called a source language, a semi-English like language, and another computer program called a compiler translates it into the numeric codes to get the computer to do what you want. So the computer essentially speaks in one language, has numeric language, people used to write in the numeric language and it was very difficult. And now people write in source languages, which is English like languages and lets other computer programs do the work of translating into the machine language.

. . . . .

THE COURT: Getting back to the source code, where is this kept, the source codes? What does it look like? Is it in a book?

THE WITNESS: It is usually kept in two forms. It is usually kept on magnetic media, which is on a diskette, and also it is typically kept in a book, listed what is on the magnetic media. It might be on a tape or on a disc. It is kept in a secured safe area.

Q Do you know where the codes for Estimacs was kept at Computer Associates headquarters?

A No. That is, again, making the distinction of what they do versus what I did, they are responsible for product management and evolution —

Q I didn't ask that. Do you know where it was kept?

A No.

Q You don't?

A No.

Q Are these formulas and weights contained within the source code?

A Yes.

Q So, you have the formulas, the weights, plus this language within the source codes?

A Yes" (Tr. at pp. 96-99).

In the essence of his testimony, Rubin stated time and again that although he used the IBM manual as a base, he himself changed the order of the questions and the weights so as to develop a new concept.

The Court finds that Rubin reorganized the IBM question set, assigned different weights from those in the IBM manual and derived a new estimation tool able to produce an estimate of total project development as well as maintenance. In this regard, this new estimation tool did contain "trade secrets." It is clear that, using the IBM material as a basis, an infinite number of weights and formulas could be conceived.

On cross examination, Rubin was also asked what trade secrets the plaintiff is asserting:

"Q Dr. Rubin, could you please give us a brief statement as to exactly what trade secrets the plaintiff asserted, is asserting? Let's get them into separate categories, what separate secrets the plaintiff is asserting?

A The trade secrets that we are asserting covered the formula that embodied the mathematical models.

Q Would the formulas be one component of the trade secrets?

A I would say the mathematical models to be one component.

The architectural components of the product and relationship to each other.

. . . . .

Q So the architecture would be one of the components also?

A Architecture is the component of the product and their relationship to each other and how they interact.

. . . . .

Now, these are essentially the architectural what I call building blocks of the product. Each one of these building blocks has some value independently, just as things that might be used in parts of a house where a sink might have value or a refrigerator might have value. But the other aspects of the architecture is how these pieces are combined and relate together through common paths through the products.

. . . . .

Q Thank you.

So you said one side of the trade secrets would be the formulas and the mathematics and the other part would be the architecture?

A Yes.

. . . . .

Q Let me backtrack, the idea of the question set that we spoke about last week, there would be no assertion of a trade secret of those 25 question sets that we referred to from the IBM article; is that true?

A I would say correct. It is not the questions themselves that are the trade secrets. That's what I referred to the last time. It is how the questions were applied and used ... to generate a total development estimate for a variety of different life cycles, and those life cycles we enumerated before as traditional life cycles, as a prototyping life cycle, and a family of life cycles that deal with what we call package acquisition.

. . . . .

Q Would the source code be within the formulas of mathematics or would it ba a separate category by itself?

A This source code contains the mathematics. And not being expert on what constitutes a trade secret, but if it has the trade secrets by my definition then access to it is looking at the formulas which I claim are the trade secrets. The mathematics I should say. The mathematics contain a few components other than just formulas.

So, I would say the sources code is a trade secret, and we have treated it as such since I first introduced my products in 1981, through both written agreements by anyone who got a product demonstration in the initial days or licensed the products or acknowledged what they were seeing was the trade secret. If they got a copy of the product they have had to acknowledge it is a trade secret.

THE COURT: What do you mean by that? You say when you give the source codes to anyone?

THE WITNESS: No, when they get a copy of the products. The source code went to places like management in Computer Services and Computer Associates when they acquired Management to Computer Services.

THE COURT: Just to the people who own the system?

THE WITNESS: The people who own the system get the source code. And that's the product vendor. In all my contacts with them they have acknowledged that these are trade secrets" (Tr. at pp. 116-126) (emphasis supplied).

Again, with regard to how much he learned and used from the IBM manual, Rubin stated that he used "modification to the IBM idea and formula as a starting point because the IBM question set was designed to estimate design effort only." Significantly, Rubin stated that ESTIMACS was the first and only estimating system to use the total effort ideas, except for the defendants' product. (Tr. p. 130). He also testified that ESTIMACS was the first system to estimate total system development effort and maintenance effort for high level business oriented questions.

Each ESTIMACS licensee signs an agreement that it will not try to re-engineer or reverse engineer the model or attempt "to figure out what the weights are and how they are used within the product."

On the subject of confidentiality and safeguarding the alleged trade secrets in ESTIMACS, Rubin testified that Computer Associates provides a user manual to its licensees. However, the manuals are copyrighted material. Also, the licensees produce their own internal manuals for their employees. For example, Allstate Insurance wrote its own manual (Defendants' Exh. C) which contains a list of the 25 questions as well as the weights.

Rubin testified that ESTIMACS has many original ideas:

"Q Dr. Rubin, do you have any original ideas in Estimacs, or are the ideas in the Estimacs products derived from industry wide standards and literature and publications?

A Original ideas existed in Estimacs in many levels.

Number one, the original idea was to create an environment in which effort, staff cost, risk and financial considerations along with business values all can be considered in one computer programing environment.

Number two, original idea was to separate the issue of large project development estimations from small project estimations from maintenance estimation entirely.

Original level number three, as I am labeling them as I discuss them, had to do with in fact bounding estimates with floors and ceilings in terms of presentation.

Original idea number four was to provide the user with interactive feedback as to what were the estimate drivers. And the next original idea as I have lost count for the moment, had to do with the factor of giving these drivers both at a high level and at a question level, and one of the most original ideas of its time was allowing someone to estimate a project and look at the dynamics of it in multiple different life cycles.

Until Estimacs existed no product existed that allowed all those things to be explored in those manners.

The mathematics of Estimacs is unique and original, in that there is no published source that exists that has our sequence of questions, the weights, the coefficients, the linking of equations, the floor —

Q Dr. Rubin—

THE WITNESS: This is my—the floor and ceiling equations and a whole flow of connecting all these things together, for small projects, large projects and maintenance projects" (Tr. at pp. 334-336) (emphasis supplied).

Rubin persistently protected his ideas:

"Q So it is your testimony you had a master trade secret agreement prior to the sale of the products to MACS?

A To anyone who saw that product, I made them sign what my attorneys called a nondisclosure agreement, and which the signee, I don't know what the word is, would agree that what they are looking at is something that contains trade secrets and proprietary information therein, and they would not try the to reverse engineer or do any of the other things.

. . . . .

Q If it was broken would all the source code and secrets be out of the bag?

A It would be like having a chemical essayist going through Coca Cola and deriving the formula for it" (Tr. at pp. 345, 349).

In a series of telling questions, it was forcibly brought home that although the IBM question set and the data in the IBM manual were readily available to everyone in the field, apparently no one but Rubin and the defendant Bryan were able to develop an equivalent estimation system:

"Q Professor Rubin, in the more than 12 years since you began your research on estimations, and in the ten years since QUEST was first released, has anyone other than you been able to use the IBM materials as a source to create a viable estimation tool, and I exclude defendant's product from that estimation?

A No ... More than 1,500 customers have adopted the product, 12 IBM sites, major cites to my knowledge, and other researchers have all had access to the IBM information and none have developed an equivalent product.

Q Again, excluding defendant's product from this question, Professor Rubin, in the more than ten years since you began marking (sic) your estimation program questions, now known as Estimacs, has any other product or program used the coefficients you developed for analyzing estimates according to alternative life cycles?

A No.

Q And during that same periods of time, and again excluding defendant's product, has anyone to your knowledge come up with a maintenance model using the IBM materials as a source?

A Definitely not.

Q Has the complete formula utilized in the Estimacs maintenance mode ... ever been published?

A No" (Tr at pp. 381-382).

BERNARD YAGED, is the holder of Bachelor and Master degrees from M.I.T. and a doctoral degree in Mathematical Planning for computer communications network. He has been employed by Computer Associates since February, 1990, as Product Development Manager for ESTIMACS and other products. He is responsible for bringing out the new version of ESTIMACS known as the 7.0. The 6.0 is the ESTIMACS version at issue.

In October, 1990, Yaged learned of the defendants' software program, called AD/ TEC, when one of the licensees of Computer Associates allowed the defendant Bryan to test his new product. While living in Dallas, he contacted Bryan in November, 1990, and by way of a subterfuge masking his identity and his true name, he obtained an evaluation copy of the defendants' AD/ TEC program in the form of four program disks. Thereafter he obtained a copy of the second version of AD/TEC. He loaded the program onto the hard disk of his personal computer at his home. He was asked to compare ESTIMACS with AD/TEC:

"Q With respect to the AD/TEC product, how does the AD/TEC product operate?

A The AD/TEC product operates in essentially the same way as the Estimacs product does. Upon entering the products the user is asked to make certain decisions and choices with regard to the project and how it can be represented by general (sic) certain program settings. By that I mean such things as which life cycle we would use, what is the annual salary cost for programmer or analyst or manager, in such general parameters.

. . . . .

Q And how do the questions used in AD/TEC compare with the questions used in Estimacs?

A The information requested of the user is identical to the information requested of the Estimacs user. The only difference is in the wording of the questions and the fact that one of the questions in Estimacs, which is a compound question, is broken down into two simple questions.

Q Is the phrasing of the AD/TEC questions set the same as in Estimacs?

A No.

There is — it is not exactly the same as the wording in Estimacs, but it does have the same content. And moreover, the wording of the question is precisely the same as the wording proposed to Computer Associates in a confidential communication with the Minneapolis user group so we can use it in the next version of Estimacs, 7.0" (Tr. at pp. 245-247).

Yaged was familiar with the HIQ computer program developed by Bryan in March, 1989, while he was employed by Computer Associates. He examined and tested the HIQ program, and compared it with the defendants' later AD/TEC program and testified that "the two question sets are identical" (Tr. at p. 263). He further compared HIQ with ESTIMACS and AD/TEC with ESTIMACS, in detail. Many charts and other exhibits were introduced in evidence to demonstrate the similarities and differences between the three products. In one phase, for example, in comparing HIQ with ESTIMACS, in each of twenty cases, the effort hour estimates in both systems were identical.

After testing the two systems, Yaged concluded that ESTIMACS and AD/TEC "were virtually indistinguishable" (Tr. at p. 423). He further concluded that 1) the design of AD/TEC is extremely similar to the HIQ program; 2) the architecture of HIQ is exactly replicative of AD/TEC; and 3) the origin of the AD/TEC product was in fact the HIQ product.

As to a comparison of ESTIMACS and AD/TEC, Yaged's conclusions were contained in the following testimony:

"Q Have you formed an opinion to a reasonable degree of scientific certainty as to whether the nearly exact correspondence of the coefficients that you have testified to in AD/TEC and Estimacs could have occurred due to independent research and development?

A Yes, I have.

Q And what is that opinion?

A It is my opinion that it would be virtually impossible for two independent researchers to come up with such similarities, even if they were to have the same raw data upon which to base these estimates. Even with the same general procedure, independent researchers would use different detail procedures, they would apply judgment, and the likelihood of ever coming up with the same final results that we have here is virtually impossible.

I might add that Professor Rubin not only did statistical analyses of many project to come up with these coefficients that are used to break the total projects into its piece parts, he also said he did apply some judgment, smoothing out some of the results to make in my mind, and probably easier to understand or easier to use, so they were — there were really two phases in what he did.

I believe that these numbers that I have shown on the top here as AD/TEC life cycles reflect both stages of Professor Rubin's research work.

Q What in your opinion explains the correspondence between the coefficients that you have testified to?

A I can think of no other explanation except that they were just copied or lifted from Estimacs and included in the AD/TEC programs" (Tr. at pp. 477-479) (emphasis supplied).

As to a comparison of the data in the IBM manual and the ESTIMACS system, Yaged testified to significant differences, as follows:

"Q Can you draw any conclusions with respect — to a reasonable degree of scientific certainty as to what your testing of the IBM materials shows in comparison with that of Estimacs?

A Yes, I can draw two conclusions.

The first conclusion is that the Estimacs question weights are significantly different from the question weights shown in the IBM slash SRA document.

My second conclusion is that the scaling factors are not the same.

. . . . .

Q Did you have occasion to conduct any tests with respect to the maintenance model of Estimacs 6.0 with any procedures contained in the SRA slash IBM book?

A No.

I could not do any testing of the Estimacs maintenance testing procedure versus the IBM document, because the document has absolutely nothing in it that relates to a maintenance project.

The questions that would have to be asked for a maintenance project are not included in the IBM document, and nothing in the manual even remotely resembles something that can be used to develop an estimate for a maintenance project" (Tr. at pp. 495, 496) (emphasis supplied).

The Court finds Bernard Yaged to be a credible witness and accepts his final conclusions as a finding on the part of the Court, as follows:

"Q Directing your attention in the previous testimony to when you expressed an opinion concerning that the AD/TEC product was a direct descendant of HIQ, have you had an opportunity to form an opinion to a reasonable degree of scientific certainty as to the origin of the HIQ product?

MR. MASTAGLIO: Objection, your Honor.

THE COURT: Overruled.

A Yes.

Q What is that opinion?

A It is my opinion it is a direct descendant of Estimacs incorporating the same formulas and equation, formulas, question weights and scaling coefficients" (Tr at p. 518).

On cross-examination, Yaged conceded that in order to surreptitiously obtain a copy of the AD/TEC product without revealing his true identity, his wife signed his fictitious name on the defendant's evaluation agreement (Defendant's Exh. G). His wife also signed his fictitious name to a license and non-disclosure agreement (Dft's Exh. H and I).

This conduct of Yaged and his wife, although not condoned by the Court, does not constitute the equitable defense of lack of "clean hands." Such action by the Yageds was taken not to copy the AD/TEC system for some competitive advantage but rather to resort to self-help to determine to what extent, if at all, Bryan had copied and was using the CA trade secrets in Estimacs (see Koehring Company v. National Automatic Tool Company, Inc., 257 F.Supp. 282 [S.D.Ind.1966] aff'd 385 F.2d 414 [7th Cir.1967]; Energetec Systems, Inc. v. Kayser, No. 84 C 10611 slip op., 1986 WL 8058 [N.D.Ill. July 17, 1986]).

Moreover, the doctrine of unclean hands is only applicable when the conduct relied on is directly related to the subject matter in litigation—in this case meaning that it would have to be related to the creation or acquisition of the trade secrets themselves (see Mehlman v. Avrech, 146 A.D.2d 753, 537 N.Y.S.2d 236 [2d Dep't 1989]).

SANJAY KUMAR is the Senior Vice-President of Computer Associates and is responsible for the oversight of ESTIMACS. He testified that he is familiar with the trade secrets that the plaintiff is seeking to protect in this lawsuit:

"Q Are your familiar with the trade secrets, the actual trade secrets that Computer Associates is seeking to protect in this lawsuit?

A Yes, I am.

Q Would you identify for the Court what those trade secrets specifically are.

A I would say there are three types. The first are the equations and formulas as developed by Professor Rubin over many years, which permit the product to derive effort estimates for both new product development and maintenance development, and allow the end user of that product to break out that effort estimate into things like staffing, costs and duration for both new product development and maintenance development, and also in various life cycles.

The second is the architecture of the software as has been testified to by, again, Dr. Rubin and Mr. Yaged, which consists of a threshold choice between new development and maintenance development, a variety of features and capabilities for alternatives such as staffing, cost, durations and risk. The capability to assess those alternatives in the context of a life cycle model or estimation model, using various alternative life cycles.

All of those in the—to use a word, all those in terms of an estimation model which requires the user not to know a lot of technical stuff, but rather, to have access to business data which is generally available at the start of a project. The third item is the confidential customer information which CA seeks to protect. Specifically with regard to Estimacs, and specifically the confidential feedback which the customers provide us with regard to enhancements that they desire to Estimacs" (Tr. at pp. 624-626).

Kumar also testified as to the pecuniary and other value of these trade secrets and of ESTIMACS to the plaintiff corporation:

"Q And what is the value of these trade secrets to Computer Associates?

. . . . .

THE WITNESS: I believe you are right.

The amount of the value to that I think is incalculable in many terms. Number one, we have clearly derived great financial benefit from the Estimacs product which amounts to I think millions of dollars which I think we will get into.

However, I think more important than that, we derived significant amount of good will, incalculable good will from having a product as successful as Estimacs.

As Professor Rubin testified Estimacs is installed in over ten times more sites of customers than its nearest competitor. Being able to get a quality product like Estimacs in the door at one of the customer sites, and having the customer being very satisfied with a product like that, and being able to follow it up with other products and being able to get us in the door with more products is incalculable good will associated with that. Directly, we have earned over 17 million dollars in the sale of Estimacs and Planmacs since 1985, since we acquired the marketing rights of the product from MACS.

THE COURT: I said he wouldn't get into dollars. But this is different. He is telling us how much he earned, Mr. Mastaglio.

MR. MASTAGLIO: I agree.

THE COURT: What did you say with respect to that?

THE WITNESS: We had earned 17 million dollars.

THE COURT: Gross?

THE WITNESS: Yes, approximately 17 million dollars gross.

THE COURT: From Estimacs?

A From Estimacs and Planmacs since 1985, at which point we obtained the marketing rights from MACS company" (Tr. at pp. 626-628) (emphasis supplied).

Computer Associates made a substantial monetary investment in ESTIMACS, having paid over four million dollars to Rubin in royalties and consulting fees in addition to "thousands of person hours developing and enhancing the product."

The price of ESTIMACS alone is from $23,000 to $67,000; when sold with PLANMACS, the price is from $34,700 to $133,000. The scope of the business generated by ESTIMACS is substantial. ESTIMACS is used in approximately 1500 sites. Its customers vary in size from IBM and other Fortune 500 and 100 companies to smaller companies and software developers.

Under the standard CA license agreement, title to ESTIMACS remains in Computer Associates and the system is treated in the agreement as a "trade secret." In this regard, the agreement provides, in a section entitled "confidential quality and restrictions", as follows:

"Title to the licensed program remains with CA, and the licensed program is a trade secret and a proprietary property of CA. Licensee and its employees will keep the licensed program strictly confidential, and licensee would not disclose or otherwise distribute the licensed program to anyone other that the licensee's authorized employees. Licensee will not remove or destroy any proprietary markings of CA. Licensee will not permit anyone except its authorized employees to have access to the licensed program. Except for archives purposes, licensee will not make or permit others to make copies of or reproduce any part of the licensed program in any form without prior written consent of CA. In no event would licensee decompile, disassemble, or otherwise reverse engineer any licensed program.

. . . . .

If this agreement should terminate for any reason, licensee shall certify in writing to CA that all copies or partial copies of the licensed program have been either returned to CA or otherwise destroyed and deleted from any computer libraries or storage devices and are no longer in use by licensee" (Tr. at pp. 632-633).

Also, in the ESTIMACS manual, under the title "proprietary and confidential information," the following statement appears:

"This material contains, and is part of a computer software program which is, proprietary and confidential information owned by Computer Associates International, Inc. The program, including this material, may not be duplicated, disclosed or reproduced in whole or in part for any other purpose without the express written authorization of Computer Associates. All authorized reproductions must be marked with this legend" (Tr. at p. 634).

While the Court is not bound by such self-serving statements, these documents are evidence of the rigorous means that the plaintiff instituted and enforced, in a continuous effort to attempt to preserve the confidentiality of its product.

Computer Associates also attempted to protect its trade secrets internally. As a condition of employment, CA employees were required to sign agreements in which they agree not to disclose confidential information either during the employment or anytime thereafter. When an employee leaves he or she signs a "termination agreement" in which the person agrees to keep secret the proprietary or confidential information gained during employment with CA.

The Court finds that Computer Associates took assiduous and appropriate measures to secure the confidentiality of its trade secrets. In its Garden City office all employees are required to wear identification badges containing their name and photograph. Entrance to this building is restricted by magnetic card procedure and certain internal facilities are kept locked. Visitors are required to "sign in" at the front desk, are escorted through the building by a Computer Associates employee and security guards are on duty in the evening hours.

As to the equations and coefficients which are part of the alleged trade secrets, especially tight security is provided. The source code is stored in three locations; at the Garden City headquarters, with the developers who work on the product and at a bonded escrow agent who keeps all the source codes of Computer Associates. The source code is not provided to the licensees. Kumar testified that only Rubin knows the equations and formulas, including the coefficients and weights which form the heart of ESTIMACS.

On cross-examination, Kumar conceded that the Computers Associates customer list was not a trade secret.

JOHN KELLY BROWN is an Assistant Vice-President of Computer Associates and, in 1988, was also a "product owner." He was responsible for the development and support of eight products, including ESTIMACS and PLANMACS. From January, 1988 through September, 1988, the defendant David Bryan was the product manager for ESTIMACS. As such, Bryan attended meetings to develop planning guides and was involved in open-ended discussions regarding the forthcoming ESTIMACS 7.0 product. From June, 1988 through September, 1988, Bryan worked as the product manager in the Garden City headquarters. At that time it was brought to Brown's attention by the two persons responsible for the development and support of ESTIMACS that "Bryan had developed a prototype product which he named HIQ, which looked very much like ESTIMACS" (Tr. at p. 759). Bryan demonstrated the HIQ product to Brown who observed "that it was essentially ESTIMACS with a more modern look and feel" and appeared to do the same thing as ESTIMACS (Tr. at pp. 760, 763).

MICHAEL A. McELROY is an attorney, Vice-President and corporate Secretary of Computers Associates and supervises litigation in which the company is involved. Bryan became an employee of Computers Associates on November 19, 1982 and left its employ on July 26, 1989. In 1982, every prospective employee was required to execute an employment agreement as a condition of being employed and Bryan did so on November 19, 1982 (Plf's Exh. 31).

In addition, defendant Bryan signed a "parting employee agreement" on July 26, 1989 (Plf's Exh. 32), which provides, in part, as follows:

"I acknowledge that I have again been carefully and fully advised of my continuing obligation to preserve as confidential (am not to reveal to anyone or use, for myself or anyone else) any trade secrets, know-how or confidential information learned by me during my employment by Computer Associates; and I confirm my promise to perform the obligations I undertook in the Computer Associates Employment and Confidentiality Agreement, including my agreement not to reveal to my employer (or to anyone else), or to use in any way for my own purposes, any confidential information, trade secrets, business records or other materials or property provided to me, learned by me or disclosed to me during my employment with Computer Associates. Further, I state that I have not retained any Computer Associates materials or personal property including, not by way of limitation, any documents or other written materials, or any items of computer or other hardware, or any software, belonging to, or in the possession of, Computer Associates."

At this point, the plaintiffs rested.

THE DEFENDANTS' CASE

MARK LEFF has been employed by Computer Associates since June, 1986, and is the present product manager for ESTIMACS and PLANMACS. Bryan was his predecessor as product manager of ESTIMACS. He shared an office with Bryan from October of 1988 until May of 1989. He learned of Bryan's HIQ product in May, 1989, when Bryan said to him "take a look at this, this is what ESTIMACS 7.0 should look like ... and it is something I have been working on for a while" (Tr. at pp. 880, 883 and 884). Although Bryan gave him diskettes with the HIQ program, he was never given the source code for that system.

The defendant DAVID WILLIAM BRYAN had studied computers in college. Although he majored in biblical studies and did field work in the ministry, he worked with computers during and after college. Bryan testified that he has been self-employed since August, 1990, as the President and sole shareholder of the defendant Application Development Technologies, Inc. ("ADT"). He was employed by the plaintiff Computer Associates from November, 1982 to July, 1989, in various capacities and at several locations.

In May, 1987, he began to be involved with ESTIMACS and PLANMACS, first as a sales representative and then as marketing manager. It was a product that "intrigued" him. As a sales representative for ESTIMACS he did approximately 150 demonstrations. He became the product manager for ESTIMACS and PLANMACS from January, 1988 to December, 1988, when Mark Leff took over. He tendered his resignation on July 10, 1989 and signed a "parting employee agreement" because of a threat to withhold his accrued vacation pay.

Bryan was then employed by a firm called On-Line Software in Marino Del Rey, California from August 1, 1989 to June 15, 1990 when he formed his ADT company. Thereafter he worked full time for ADT.

Raising a key factual issue, Bryan testified that before he signed his employment agreement on November 19, 1982 he crossed out and initialed an "objectionable" paragraph numbered three, which reads as follows:

"3. Any discoveries, concepts or ideas or expressions thereof, whether or not subject to patent, copyright, trademark or service mark protection concerning any present or future activities of the Company conceived by the Employee, shall be the sole and exclusive property of the Company and the Employee will do all things reasonably requested by the Company in order to vest the entire right, title and interest to any such discoveries, concepts, ideas or expressions thereof in the Company" (Plf's Exh. 31; Dft's Exh. "U").

The Court finds that, contrary to his testimony, the defendant David Bryan did not cross out the provisions of paragraph "3" of the employment agreement before he signed the document. The Court finds that Plfs' Exh. 3 is the unrevised original agreement as signed by Bryan and Computer Associates. This finding is significant, not only as to the substantive result, but also as to the credibility of Bryan in regard to other matters.

Bryan explained that he accumulated the data with which he created his HIQ product and the later AD/VENT product during discussions with various persons in CA, examination of documents at CA, review of articles by another computer scientist named Capers Jones, research done for a lecture, the IBM handbook and article and general research work in the field. He emphasized that he got no information or research assistance from Rubin. In fact, he testified that Rubin refused to help him with his research. According to Bryan, among the research works used by him were the Rayleigh-Norden Curve, the Howard Business School Studies, the Dallas Tire Case and Albrecht data. Bryan stated that he did much of this research to prepare for his lecture at a CA sponsored Users Conference in August, 1988.

Bryan testified about a booklet published by Rubin in 1983 called "Macro-Estimation of Software Development Parameters: the Estimacs System," which lists the 25 questions and 10 of the 25 weights involved, to demonstrate a "publication" of the alleged trade secrets by Rubin. The article also referred to the architecture and life cycles.

After Bryan returned from the User's Conference, he personally ordered and paid for an IBM manual. He states that he saw the questions in the IBM manual and complained to fellow employees that "Computer Associates was being ripped off by Rubin, because in fact there had been no original research" (Tr. at p. 1016).

As to HIQ, Bryan testified that he wrote this computer program at home between December, 1988 and January, 1989, on his own computer. He stated that he started to work on the HIQ estimation program by "accident". In HIQ and AD/TEC he used "prologue" computer language while ESTIMACS uses "basic" language. It took him from January, 1989 to March, 1989, to develop a suitable program. He then approached his boss Steve McManus and told him "by playing around at home I have come up with something that I thought was a valuable product that Computer Associates would be interested in" (Tr. at p. 1038). He also spoke to Don Balducci and demonstrated the product for Kelly Brown, Allen Henriksen and Phil Saper. Kelly Brown then asked to see the source code and he showed it to him. Brown later told him that CA was not interested in HIQ. Although Bryan testified that he obtained the questions and the equations from the IBM manual and the weights from the Allstate Insurance Users' manual, he conceded that the ESTIMACS weights were somewhat different. Further, Bryan stated that he did not have access to the source code for ESTIMACS. Bryan left CA on July 26, 1989.

Bryan then spoke to several computer people about his HIQ program. A woman in Texas was willing to buy his program but he declined saying he was "under a one year noncompete" restriction (Tr. at p. 1080).

On October 6, 1989, Michael McElroy, Vice-President of CA, wrote to Bryan advising him that use of HIQ as a competing system was in violation of his employment agreement and demanding that he cease using HIQ and return the copies to CA (Plf's Exh. 33). Bryan retained counsel in California who wrote to CA denying any violation of the employment agreement and asserting that no trade secrets were involved.

According to Bryan, the first reference to AD/TEC occurred in late June, 1990 when he discussed his estimation system with one Denise Mura, an employee of John Hancock Insurance Company. He contacted Ms. Mura to solicit her help to evaluate his AD/TEC program. Ultimately, by sales of his AD/TEC system, Bryan engaged in direct competition with CA-ESTIMACS and solicited CA licensees.

Bryan described how he developed the AD/TEC system starting in November 1989. He looked at all manuals, he reviewed all old formulas and made "notes and adjustments." Having done this, Bryan testified he came up with entirely new formulas, equations and function points for AD/TEC.

The Court does not credit this testimony and finds that Bryan used the formulas, equations and function points he derived from ESTIMACS, in his HIQ system and, further amplified, in his AD/TEC product.

Bryan further testified that he used preexisting architecture which he bought off a shelf. He developed a "prologue computer language." Also, he explained that his AD/TEC system has some features of its own. For example, it has an "error-trapping" feature and a "mouse", which is a device that can glide across the table top. Further, he stated that ESTIMACS operates under a DOS (disk operating system) while AD/TEC uses an OS/2 more modern system. In addition, AD/TEC has printer support for all commonly used printers, which ESTIMACS does not. Further, the ESTIMACS program contains the Rayleigh-Norden curve which includes certain instructions not present in AD/TEC. Also, there is a difference in the two systems as to staffing levels; the AD/TEC report is a large page while ESTIMACS is a small page; AD/TEC can change life cycle in midstream; and AD/TEC is completely integrated, among the alleged differences in the products.

With regard to architecture, Bryan testified extensively about certain differences between AD/TEC and ESTIMACS. Also, Bryan stated that the AD/TEC and ESTIMACS formulas are not the same, as follows:

"Q And to a reasonable degree of scientific certainty, based upon your empirical testing, are the formulas in Estimacs the same as the formulas in AD/TEC?

A With all 105 tests run and compared and the fact that no matches were ever found it proves beyond a shadow of a doubt that the formula used by AD/TEC NDV isn't the same as used by the CA Estimacs effort module.

. . . . .

Q Can you indicate to the Court what the testing of both the maintenance test which you just referred to and the new development comparisons tell you about the differences, if any, between the formulas and weights in AD/TEC and the formulas and weights in Estimacs?

A Yes, they prove they are different beyond a shadow of a doubt. This testing is conclusive dealing with the complete range of testing there could be no other determination made but that they are different" (Tr. at pp. 1456, 1463) (emphasis supplied).

On cross-examination, Bryan conceded that he lied on the resume he submitted to Computer Associates in that he falsely stated he received a degree from Florida Bible College. He also lied on another Computer Associates' form in which he said he had a degree in computer science from Triton College. When he left Computer Associates, Bryan was employed by On-Line Software and he again lied about receiving a degree in computer science from Triton College. The Court finds he also lied about other facts in the On-Line Software application.

While at Computer Associates, Bryan worked on ESTIMACS for approximately seventeen months and conducted approximately 150 benchmarks (run-throughs) of ESTIMACS. Bryan also conducted research and lectured on the ESTIMACS product. The Court finds he was thoroughly familiar with the ESTIMACS system.

According to Bryan, the HIQ and AD/TEC systems are different in that they have different formulas and internal architecture, among other things. Bryan testified that HIQ took fifteen or sixteen hours to build over a few weekends while AD/TEC was a much more extensively developed system. HIQ, he said, was "a very shoddy prototype."

The Court finds that HIQ was developed by Bryan while working for Computer Associates and by using ideas obtained from ESTIMACS, and that HIQ was a prototype to AD/TEC.

Bryan testified that he developed the ideas for AD/TEC on his own and by use of the IBM manual. However, it is significant that Bryan conceded that he used certain portions of the ESTIMACS system in developing AD/TEC:

"Q How many tests did you run comparing the IBM formula to Estimacs?

A At that time, five or six" (Tr. at pp. 1727, 1728).

In June 1990, while still subject to his one-year restrictive covenant, Bryan contacted John Hancock Life Insurance Company, a Computer Associates licensee. He sold his AD/TEC product to John Hancock for $9500 in mid-October 1990. Bryan also sold AD/TEC to three other licensees of Computer Associates: Arco (Atlantic-Richfield Co.), Electronic Data Systems and Superior Value. In addition, the Court finds that while employed by Computer Associates, Bryan created a data base on his personal computer which contained the names of companies who called Computer Associates indicating that they had a problem with ESTIMACS. Bryan conceded that he used the question set contained in a CA licensee's report to him, as Product Manager of ESTIMACS, on June 14, 1988 (Plf's Exh. 6) both in his HIQ and AD/TEC products.

Interestingly, both ESTIMACS and AD/TEC each represent that it will produce a result "within plus or minus 15 percent of accuracy."

Also, of crucial importance, Bryan knew that the source code for ESTIMACS was on one or both of two computers of two co-employees and that he could access the ESTIMACS source code:

"Q You knew that source code for Estimacs was on one or both of those computers, did you not?

A It should have been.

. . . . .

Q You knew that the source code was available on either Mr. Sapir's or Mr. Henriksen's computers, you knew there was no electronic security on those computers, and you are now telling me that you didn't know where to go and that you could gain access if you had wanted to do it? Is that what you are telling me?

MR. McENTEE: Objection.

THE COURT: Overruled.

A I said the code should be on the machines. I could not verify it. Unless I did a directory of the machine there would be no way, your Honor, for me to know that all the source codes was on those machines. Theoretically the codes should have been on those machines.

THE COURT: Counsel is asking you if it was on there and if the source code was in those machines and if there was no electronic security, you could get access to them?

THE WITNESS: Oh, sure, anyone could" (Tr. at pp. 1962-1964).

MARIAN KAY HARLOW was a sales representative for Computer Associates from August, 1987 to January, 1990. As such, she would send sales literature to prospective customers, including a list of ESTIMACS users. She would also do onsite demonstrations, including discussion of the life cycles and the questions. In addition, some of the licensees developed their own "users manual" which Computer Associates obtained and distributed to prospective customers who were non-licensees. These user manuals contained the twenty-five questions and the weights assigned to those questions and a description of the life cycles. Harlow testified that hundreds of these user manuals were given to potential customers. There was an ESTIMACS 6.0 manual prepared by Computer Associates and "a few" would be sent out to potential customers.

On cross-examination, Ms. Harlow testified she did not know the formulas for ESTIMACS nor does she know the weights which were used.

ALAN ALBRECHT is a self-employed consultant in data processing and estimating software projects. He was formerly employed by IBM from 1949 to 1989. In 1978, he made up the IBM estimating manual which was copyrighted in 1980. The IBM manual of the SRA Management Library, was published by Science Research Associates, Inc., an IBM subsidiary. The IBM manual was intended to be instructional and involved hundreds of projects. He developed and invented "function points" which are used in estimating.

Asked to compare the questions in the ESTIMACS system with the data in the IBM manual, Albrecht stated that the questions were the same but the order of the questions was not the same, in that "it was different from the IBM sequence" and "it made the programming easier of the ESTIMACS estimating tool" (Tr. at pp. 1224, 1225). As to the equations, Albrecht testified that the ESTIMACS order "was in the exact order that the equations were presented in the IBM estimating manual" (Tr. at p. 1227).

Asked to compare the weights in ESTIMACS with the weights in the IBM manual, Albrecht testified that "surprisingly— there were some weight changes between the IBM document and the ESTIMACS document." However, he attributed the weight changes to an undetected "programming error." The Court finds no evidence that the weight differences between the IBM manual and the ESTIMACS system were as a result of a programming error.

While working at IBM during the years between 1974 and 1980, Albrecht developed an estimating system, which was published. The purpose of the IBM estimating system was to help manage the 600 person contract performing group in IBM. The system was published in the IBM manual. The IBM estimating system was a written procedure while the ESTIMACS system is programmed on a computer. Despite this seemingly substantial difference, Albrecht testified that anyone can take the IBM manual and build the ESTIMACS system, as follows:

"THE COURT: So, it is your testimony that anybody can take and buy an IBM manual for $300 and build an Estimacs machine without putting any additional scientific input to it?

THE WITNESS: Yes, sir" (Tr. at p. 1239).

As noted later in this decision, this startling opinion by Albrecht was not correct. In view of the differences in the order of questions, the weights, the formulas and the source code, the Court does not credit this testimony.

While Albrecht explains the differences in the weights as a "programming mistake" or "some sort of analytic process," the Court finds that the differences in the weights were as a result of analytic process by Rubin.

Albrecht also testified that the formulas or equations (an apparently synonymous term) of ESTIMACS and AD/TEC were not the same.

On cross-examination, Albrecht conceded that he was referred to the defendants by Capers Jones, who is Chairman of the company that sells the SPQR system, a major competitor of ESTIMACS. Albrecht also conceded that he is a consultant for the SPQR system, which company would be benefited by an adverse finding against Computer Associates in this action.

Further, on cross-examination, surprisingly, and without any question put to him, Albrecht changed an important prior answer in a material manner, as follows:

"THE COURT: You are still under oath, Mr. Albrecht.

THE WITNESS: Yes, sir.

Your Honor, before we begin, may I clarify an answer I gave in the previous testimony?

THE COURT: YES.

THE WITNESS: You asked me the question whether a person could buy the IBM workbook for $300, and with the information in that workbook, and only the information in that workbook, program an Estimacs system.

I misunderstood the word "Estimacs" and answered the question as if you asked an estimating system or estimating machine, and I answered yes.

I should have answered yes, for the effort estimating piece of Estimacs, because there are a couple of other pieces of Estimacs for estimating the machine time required and the dollars required, that we provided no guidance in the workbook for those two extra pieces.

I am sorry, I misunderstood.

THE COURT: So you could make what, now, an estimating —

THE WITNESS: I answered the question, that, yes, you could make an estimating machine. If in answering your specific question, could you make an Estimacs, I would say yes for the effort estimating piece of Estimacs, which was the gist of the discussion that we all had about it.

. . . . .

BY MS. STEVENS:

Q Mr. Albrecht, with respect to that clarification which you just provided to the Court, and perhaps in further clarification of your prior answer, you are familiar with the Estimacs product, are you not?

A I am familiar with the external aspects of the Estimacs products, nonconfidential material.

. . . . .

Q Just so we are perfectly clear on this, Mr. Albrecht, it is correct that you have now taken back your statement that one could build Estimacs by simply paying $300 to purchase the SRA book, and that no additional scientific input would be needed to build Estimacs other than what was available in Exhibit a?

A I can't answer that yes or no. Could you rephrase it?

Q Well, do you recall —

A I am concerned about the take back, or take it back, taken back.

Q Change your testimony. Are you more comfortable with that?

A Change, yes, I did.

Q And indeed, it is fact that one could not build Estimacs simply from Exhibit A; isn't that the case?

A Part of it.

MS. STEVENS: Move to strike the witness' answer as not responsive.

THE COURT: Motion granted. The answer is stricken.

MS. STEVENS: May I have the question re-read, your Honor?

THE COURT: Yes.

(Whereupon, the court reporter reads the requested material.)

A Yes" (Tr. at pp. 1795, 1796, 1807, 1808).

Further, Albrecht conceded that the IBM manual does not instruct a reader on the "risk analysis of an application project"; or a "risk profile estimate for software development"; or "how to generate a portfolio level risk profile relating to the size, structure and technology risks of the software project to the user's existing portfolio of projects"; or how "to estimate the CPU power, 10 and DASD resources required to insure reasonable service levels for the proposed application project"; or how to estimate the duration of a project; or how to "instruct a reader as to how to perform maintenance estimation"; or with regard to any equations for computing floors and ceilings on effort estimates; or how "provide life cycle models for the prototyping life cycle."

In addition, the IBM manual expressly cautions that it is not intended to address a maintenance estimation. Significantly, Albrecht conceded that no commercial product other than ESTIMACS and AD/TEC used the questions in the IBM manual as a basis for maintenance estimation. Also importantly, Albrecht conceded that IBM itself licenses ESTIMACS and "IBM obviously saw some advantage of that (ESTIMACS) over whatever it was that they had available" (Tr. At p. 1815).

TARI SCHREIDER is a computer consultant who formerly was employed by Computer Associates from February, 1979 to March, 1986. In sales by Computer Associates they used "collateral" which are various sales and promotional products including a manual, user guides, testimonials from customers and also the Computer Associates ESTIMACS manual. However, if a customer buys a tape from Computer Associates, the customer is required to sign a confidentiality agreement. Also, Schreider testified that if someone had the manual and user guides, it would be very difficult to construct a product from such material, "and there was never any concern about that at Computer Associates."

STEPHEN McMANUS is a marketing manager for a main frame software company. He is a friend of the defendant Bryan. During the trial, he picked up the defendant at the airport and Bryan stayed at his home. He was employed by Computer Associates between 1977 and 1990. He supervised Bryan when he was the product manager for ESTIMACS and is familiar with the Computer Associates marketing materials (called "collateral"), including the user manual. He testified that this material was freely distributed to prospective customers.

Although he supervised his employees closely, he never saw Bryan work on HIQ in the office. McManus was familiar with the architecture of both ESTIMACS and AD/TEC and, according to him, the external architecture seems to be very different. He refers to the "mouse" pointing device and the use of pictures rather than text. Also there are differences in the print function capabilities of the two systems. In addition, he said that the two systems have different computer language.

On cross-examination, McManus testified that in November, 1990, he was fired by Computer Associates. Further, he appeared in court with defendant Bryan on almost every hearing date. Also, McManus conceded that he advanced over $20,000 to Bryan. He paid Bryan's prior counsel $15,000, and paid his present counsel $5,000. He also purchased airline tickets for Bryan in the sum of approximately $4,000. In addition, McManus is now employed by a direct competitor of Computer Associates.

McManus admitted that although he himself ran the ESTIMACS system fifty or sixty times, he did not know the internal formulas.

The Court does not credit the testimony of Stephen McManus.

The defendants recalled Dr. Howard Rubin who testified that the following ESTIMACS system parts were trade secrets: the coefficients; the staffing and cost modules; the weights in the effort and maintenance models; and the architecture, including the "look and feel" of the program.

The defendants rested.

In rebuttal, Rubin was again recalled and testified as follows:

"Q Dr. Rubin, Estimacs has a function point model, does it not?

A That is correct.

THE COURT: A what?

MS. STEVENS: A function point model.

. . . . .

Q Has the procedure ever been published to your knowledge?

A Definitely no.

Q Are you aware of any product other than Estimacs and AD/TEC that does this?

A No. I have never seen it done or attempted" (Tr. at pp. 2600, 2601).

BRENDA FEIERSTEIN testified that nothing unusual occurred in the signing of Bryan's employment agreement.

PATRICIA MAGNONE-MITCHELL was employed by Computer Associates from 1980 through 1983. In November, 1982, she was present when defendant Bryan signed the employment agreement at issue. She testified that he did not strike anything from the agreement. The Court credits this testimony.

BRONNE BRUZCO is the Senior Vice-President of Computer Associates. He testified that the policy of Computer Associates is "that the manual is not to be shipped out to any non-licensee of the product." The Court credits this testimony.

MARK J. KAMIYAMA is a systems engineer with Computer Associates. He has sales responsibility for ESTIMACS. In 1989, he worked with Bryan who told him that HIQ was an estimating program that he wrote. The witness found the HIQ program installed in computers in the Los Angeles office. Kamiyama viewed a demonstration of HIQ in June, 1989, and saw many similarities between ESTIMACS and HIQ. In his view "HIQ was a more modern or contemporary version of ESTIMACS" (Tr. at p. 2670).

Kamiyama testified that Bryan told him he was creating a data base file. Kamiyama discovered that the Bryan data base file contained information on at least 451 Computer Associates customers (see Plf's Exh. 78). This information included the customer's name, address, contact person and phone number. This, in effect, was a customer list obtained by Bryan while employed by Computer Associates and was, as Kamiyama testified, "a great marketing edge" (Tr. at p. 2685).

ALAN HENRIKSEN is the Systems Programmer and Project Development Leader of ESTIMACS. He testified that defendant Bryan was very involved in the development of ESTIMACS. He stated that Bryan was aware of the comments and complaints by customers with regard to needed improvements in ESTIMACS. Some of these improvements appeared in AD/TEC and some are, apparently, present in the latest version of ESTIMACS called Computer Associates ESTIMACS 7.0.

BERNARD YAGED was recalled by the plaintiffs in rebuttal. After making certain tests and reviewing many exhibits, Yaged testified that the design factor formula in HIQ and ESTIMACS are the same:

"Q Have you reached an opinion to a reasonable degree of scientific certainty based on all the exhibits you just testified to, 58 A through Q and HHHH, and your knowledge of the HIQ program on Exhibit 71, and your familiar with Estimacs, as to what formula is used in the HIQ program?

A Yes.

Q What is that?

A I believe the design factor formula in HIQ is in fact the Estimacs design formula" (Tr. at p. 2763).

Further and crucial to the key issue in this case, Yaged testified as to a comparison between ESTIMACS and AD/TEC, as follows:

"Q And was your conclusion based upon your initial testing of the Estimacs effort model as compared to the AD/TEC NDV program?

A My initial testing came to an initial conclusion that the AD/TEC program used the same formulas, scaling coefficients and question weights that were in the Estimacs program.

. . . . .

THE COURT: I agree. I think we are coming to one of the important issues in the case. We are finally getting to some of the important opinions. Experts have given opinions in these matters since the courts opened up. And this expert — the weight of his opinion is another matter. But he is entitled to give his opinion in this field.

Objection overruled."

As to the "mouse" feature in AD/TEC, Yaged testified that is a standard enhancement for new programs.

Both sides rested.

ADDITIONAL FINDINGS OF FACT

The Court makes the following additional findings of fact:

1) Dr. Howard Rubin designed, developed and created a software product which estimates the resources required to develop new software or modify existing software, called CA-ESTIMACS.

2) Rubin designed and created CA-ESTIMACS as a result of extensive research, including the use of the IBM manual as a starting point.

3) The resulting CA-ESTIMACS product was a substantial departure from the data contained in the IBM manual in that, among other things, weights were changed, formulas were changed and the entire maintenance system was added.

4) The CA-ESTIMACS product was the first software product which estimated the resources required to develop new software and/or modify existing software.

5) During the times at issue in this case, CA-ESTIMACS 6.0 contained the following four "trade secrets":

(a) The mathematical calculations used by ESTIMACS in its effort and maintenance models to derive total project estimates.

(b) The life cycle coefficients; the coefficients used by ESTIMACS to break apart total estimates into estimates for effort, productivity, duration and staff types at each phase of five alternative life cycles.

(c) The formula used by ESTIMACS to estimate the effect on effort if project duration is shortened, and

(d) The architecture of ESTIMACS.

6) The defendant DAVID W. BRYAN was employed by Computer Associates for seven years. During this period, he devoted approximately eighteen months to working with ESTIMACS as product manager and in liaison with the technical support staff and was thoroughly familiar with all parts of the product. Bryan himself was a systems engineer and was privy to all phases of the ESTIMACS program. He learned of confidential customer feedback information or requested changes and desired improvements of ESTIMACS. Bryan ran the ESTIMACS system no less than 150 times during one period of his employment. He was so well-versed with the ESTIMACS product that he was chosen to lecture at a seminar on the system.

7) Computer Associates took substantial, appropriate and reasonable measures to maintain the secrecy of its trade secrets and to secure and safeguard the confidentiality of its trade secrets. The Court further finds that to the extent materials containing information about ESTIMACS were distributed to non-licensees, at no time did CA publish or reveal the complete mathematical formulas of ESTIMACS and it never permitted recreation.

8) Substantial monies and efforts have been expended by Computer Associates to pay for the development of CA-ESTIMACS and its progency.

9) ESTIMACS and PLANMACS were valuable assets and a successful marketing feature for Computer Associates resulting in sales in excess of $17,000,000 between 1985 and 1991 and involving opportunities to sell their other products, which constituted valuable "good will." The Court further finds that continued marketing and sales by Bryan of AD/TEC will cause more than monetary injury to CA, in that it may lose its competitive edge, which cannot be compensated by money damages.

10) While employed by Computer Associates and privy to the confidential features of ESTIMACS, defendant Bryan developed and created the HIQ system. Bryan used and copied the formulas, equations, function points and other data he derived from ESTIMACS in order to create HIQ, which was a prototype to AD/TEC.

11) Shortly after he terminated his employment with Computer Associates, the defendant Bryan used the HIQ system and the ESTIMACS data, including its trade secrets, to develop and create the AD/TEC system at issue. The ideas were taken by Bryan from ESTIMACS, altered to some degree, and used in AD/TEC. The Court finds that substantially all the ideas, designs and techniques used in AD/TEC were obtained and taken by Bryan from the ESTIMACS system, including the four trade secrets previously stated. Bryan used the formulas, equations and function points he derived from ESTIMACS in his HIQ system and, further amplified, in his AD/TEC product.

12) In sum, the defendant Bryan essentially copied ESTIMACS in a different computer language and he created HIQ and then AD/TEC as an advanced form.

13) Some of the changes in AD/TEC were done solely to camouflage the use of the ESTIMACS trade secrets, so that Bryan's product would have some differences.

14) In addition, while employed by Computer Associates, Bryan created a data base on his personal computer which contained the names of companies who called into Computer Associates indicating that they had a problem with ESTIMACS. This data was used by Bryan to develop AD/TEC.

15) Bryan misappropriated the CA trade secrets listed above in item five.

16) Computer Associates possessed the trade secrets listed above in item five and the defendants are using those trade secrets in breach of two agreements executed by Bryan; and in breach of Bryan's duty as an employee given privy to said trade secrets (see Integrated Cash Mgt. Servs. Inc. v. Digital Transactions, Inc., 920 F.2d 171 [2d Cir.1990]).

17) Prior to and during the hearing, there was a myriad of tests made of ESTIMACS and AD/TEC with numerous graphs and charts as to the similarities and/or differences in the two products. These tests and exhibits en masse did not alter the Court's finding that the defendant Bryan learned of the trade secrets in ESTIMACS while an employee, misappropriated the protected trade secrets, tried them out on HIQ and eventually developed AD/TEC, using these trade secrets.

CONCLUSIONS OF LAW

1) The plaintiff Computer Associates International Inc. has established, by clear and convincing evidence, that the CA-ESTIMACS product contains the trade secrets listed above. Particularly appropriate to this conclusion is the recent statement of the law in this field regarding computer software as stated by the Second Circuit in Integrated Cash Management Services, Inc. v. Digital Transactions, Inc., supra, 920 F.2d at p. 174 (2d Cir.1990), as follows:

"Contrary to defendants' suggestion, the nonsecret nature of the individual utility programs which comprise ICM's product does not alter this conclusion. `[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, design and operation of which, in unique combination, affords a competitive advantage and is a protectable secret.' Imperial Chem. Indus. Ltd. v. National Distillers and Chem. Corp., 342 F.2d 737, 743 (2d Cir. 1965); see Q-Co Indus., Inc. v. Hoffman, 625 F.Supp. 608, 617 (S.D.N.Y. 1985). As the district court found, the architecture of ICM's product, or the `way in which [ICM's] various components fit together as building blocks in order to form the unique whole,' Integrated Cash Management, 732 F.Supp. at 374, was secret. See Dickerman Assocs., Inc. v. Tiverton Bottled Gas Co., 594 F.Supp. 30, 35 (D.Mass.1984) (`[T]he particular combination of procedures used in plaintiff's [computer] system, and the particular features within the system ... are neither obvious nor easily duplicated. They constitute a trade secret.')."

2) The plaintiff Computer Associates International, Inc. has established, by clear and convincing evidence, that the defendant David W. Bryan misappropriated these trade secrets in violation of his agreements and his fiduciary and common law duty.

3) The plaintiff Computer Associates International, Inc. has established, by clear and convincing evidence, both (a) irreparable harm, and (b) a likelihood of success on the merits.

CONCLUSION

Based upon the foregoing, the motion of the plaintiff Computer Associates International, Inc. for a preliminary injunction pursuant to Fed.R.Civ.P. 65(a), is granted, as follows:

The defendants David W. Bryan and Application Development Technologies, Inc. are preliminarily enjoined, pending the outcome of this action, from selling, marketing, leasing or otherwise using the AD/TEC-SYS product or any similar product derived from AD/TEC-SYS. The defendants David W. Bryan and Application Development Technologies, Inc. are preliminarily enjoined, pending the outcome of this action, from disclosing or revealing the nature of the trade secrets of CA-ESTIMACS as contained in the AD/TEC-SYS program, to any other person or entity.

Pursuant to Fed.R.Civ.P. 65(c), the parties are directed to appear for oral argument and/or a hearing on February 4, 1992 at 9:00 a.m., to determine the amount of security to be posted by the plaintiffs for the continuation of this preliminary injunction. At that time, in order to avoid unnecessary damage to the defendants, the Court will set an early trial date, subject to the Court's presently scheduled calendar.

SO ORDERED.