IV. How Silicon Valley Effectively Abolished Trade Secrets

It is obvious that Silicon Valley's high-velocity labor market could not have grown up if California courts in practice interpreted the Uniform Trade Secrets Act as the Seventh Circuit did in PepsiCo, Inc. v. Redmond, that is, to enjoin employees from competition with their old employers on the grounds that they would "inevitably disclose" some unspecified trade secret. Every day, literally hundreds of Silicon Valley employees make job changes that would be enjoinable at the option of their employer under this standard. Many of these employees who start up a new Silicon Valley firm or move to a competitor have actual knowledge of some tangible program, specifications, or research project, and are not merely, as Redmond was, on the distribution list for strategic plans. Yet we know, from Saxenian and others, that Silicon Valley is built in large part on the ease with which new firms start up and employees move from job to job.(1)

So how can a fluid, high-velocity labor market coexist with the law of trade secrets? The following hypotheses can be tested:

(1) California's law of trade secrets is unusual (like its law of restrictive covenants). Perhaps it (a) uses a restricted definition of "trade secret", or (b) rejects the doctrine of "inevitable disclosure", or © includes a strong policy in favor of employee mobility.

(2) California high-technology employers explicitly or implicitly contract with their employees that intellectual property to discoveries will stay with employees, or that they will be able to exploit what they learned on this job when they move on to the next. Or, perhaps, research "stars" can extract such property rights from employers, even if lesser researchers cannot.

(3) California employers have the same formal rights under trade secret law as their counterparts elsewhere. However, they are unable to enforce these rights effectively for two reasons. (a) Judges and juries in Santa Clara County do not like trade secrets cases and rarely award injunctions or damages, respectively. (b) Firms that litigate in defense of their trade secrets face substantial informal social and economic sanction from other firms (whose cooperation is necessary to accomplish many projects), venture capitalists, and incumbent and prospective employees.

This Part will argue that Hypotheses (1) and (2) are false. California law of trade secrets "on the books" does not differ substantially from other jurisdictions'. Silicon Valley firms normally employ form employment contracts that assert property rights in all firm information. Hypothesis (3), however, is correct. In some highly-publicized cases in the late 1980s, firms that sued departing employees accomplished little and suffered diminished reputations in Valley culture. Until recently, firms were sometimes told by lawyers not even to consider suing departing employees absent very clear evidence of theft of tangible diskettes, documents, or files. Within the last two years, however, firms have become more successful in enjoining departing employees.

The result has until recently been that California law "in action" (as opposed to "on the books") efficiently promoted the endogenous economic growth described by Paul Romer and others. It promised property in information and therefore seemed to offer incentives to firms to invest in its production. Yet, by not impeding employee mobility, it fostered the information spillovers that are essential to economic growth.

The analysis of this Part includes formal analysis of California law and analysis of the journalism surrounding several prominent suits against departing employees, supplemented with unstructured interviews with Silicon Valley corporate managers, venture capitalists, employees, lawyers, and journalistic observers. The interviews are not of a representative sample of anything, but reflect my personal contacts in the area.(2) I have run the conclusions of this Article past many informed professionals in the Valley and believe that they represent an accurate picture of the formal and informal reasons why firms rarely sue departing employees. A formal survey of research directors or company lawyers might yield more systematic information, but would not, I believe, disturb the basic findings.(3)

The "law in action" in California, then, contrary to the law "on the books", is that employees may leave firms to work for competitors or start up their own firms. As a practical matter, there is often little that their old employer can do about it. If a firm depends on particular employees, it must treat them well and keep them happy. This Part shows how this homey truth has been learned in Silicon Valley. Part II has already suggested that this truth will be conducive to rapid economic growth. Part V will consider alternative economic analyses and in particular will show that there is no practical or theoretical impediment to negotiations with particular departing employees. There is thus no reason for law to imply a default rule impeding employee mobility, or even to enforce boilerplate restrictions on mobility that employers extract from employees.

A. Does California's law of trade secrets differ from other jurisdictions'?

Since, as mentioned above, California does not enforce covenants not to compete, its law of trade secrets is uniquely important. In the Route 128 industries in Massachusetts, as Saxenian implicitly suggests, employers often require employees to sign covenants agreeing not to compete with them for a year or two after departure, and courts enforce these agreements.(4) In California, this is not possible, so employers that fear loss of information attempt to use trade secrets law to accomplish the same end.

Trade secrets law is almost sure to be a less-than-satisfactory substitute from plaintiff's point of view. The standards are unclear. Judicial enforcement will be necessary. Covenants not to compete (where enforceable) are often enforced through "bad boy" clauses under which the employee who competes with the former employer will forfeit unvested benefits or other withheld compensation.(5) I have never seen or heard of a "trade secrets bad boy" clause under which compensation is forfeited on disclosure of confidential information. Finally, if this judicial enforcement is sought, the court may require plaintiff to identify and disclose the very information said to be secret. Still, as shown in Part III, trade secrets law, though not as favorable to plaintiffs as enforceable covenants not to compete, can substantially interfere with employee mobility, particularly if the plaintiff is not required to identify particular secrets (the so-called "doctrine of inevitable disclosure," as I am using that term).(6)

There is no obvious reason in the case law of California why trade secrets law could not be used to create an employment market of internal labor markets, like Route 128's. California has the same uniform statute as forty other American jurisdictions. Trade secrets agreements in California are analyzed under the trade secrets statute. They are not treated as covenants not to compete nor analyzed under California's distinctive statute making such covenants not to compete unenforceable.(7)

In fact, California courts do enjoin employees from disclosing trade secrets on their new jobs. There are no prominent California cases limiting the definition of trade secret, or rejecting the doctrine of inevitable disclosure, or stating a firm public policy in favor of information diffusion.(8) Employers do, from time to time, try to use trade secrets law against departing employees, and to expand the utility of that law to plaintiffs. It is not unethical for employers to seek this.

For example, in one front-page but officially unreported case outside the high technology area, a county court ruled that winemaker Jedediah T. Steele could not take a particular winemaking process with him when he left Kendall-Jackson Winery. Steele's wines for Kendall-Jackson mixed a reserve of partially fermented, therefore sweet, wine, thus making a sweet chardonnay. This practice was held to be a trade secret, and Steele enjoined from making use of it elsewhere. Winemakers interviewed unanimously criticized the decision, noting that the injunction extended beyond any specific injunction to enjoin the process, which one winemaker not involved in the litigation described as "centuries old knowledge, not a formula like the recipe for Coca-Cola." Another independent winemaker said: "It would be very sad if this ruling reduced the amount of cooperation and communication between winemakers at various properties. Communication is what's made us skyrocket to the forefront of the world."(9) This Article has tried to generalize these winemakers' two points. Few things that employers claim are secrets really are.(10) Generally, the public and the employers themselves benefit more from rapid diffusion of ideas and information than from firm monopolies on information.

A second, recent case did arise in high technology and enjoined five former employees of Advanced Micro Devices from disclosing information about AMD manufacturing processes, and from working on specified projects at their new employer, Hyundai Electronics.(11) The case later settled; few details of the settlement are public.(12) The injunction has since been described as resting on "inevitable disclosure" both by counsel for defendant Hyundai(13) and for plaintiff AMD. (14) I believe however, based on conversation with counsel for each side, that the injunction rested on "inevitable disclosure" only in the sense that the employees were enjoined from working on particular projects.(15) It does not rest on "inevitable disclosure" in the PepsiCo sense, since, unlike PepsiCo, AMD did identify manufacturing processes that were arguably trade secrets and that defendants were enjoined from disclosing.(16) Finally, in light of the injunction in Advanced Micro Devices v. Hyundai, a spate of recent unreported injunctions against departing employees has been issued by courts in Santa Clara County.(17)

California trade secrets law, in short, is sufficiently malleable so that it could become, or may now be, a useful tool for employers trying to block employees' move to competitors. Unlike the law of restrictive covenants, there is no clear statute or decision that prohibits strong relief against employees and new employers. There are a few minor aspects that could be troublesome for plaintiffs. For example, in the discussion in Part III supra of PepsiCo, Inc. v. Redmond, I noted that the decision rested on a negotiated trade secrets agreement as well as the employee's general obligation under the statute. It is probably an open question under California law whether an employer and employee can by agreement make something a trade secret that wouldn't otherwise have been a trade secret under law.(18) The doctrine of tortious interference with contract may require Employer 1 to be cautious about threatening litigation against Employee and Employer 2.(19) But by and large, there is nothing distinctive about California trade secrets law.(20) Certainly, if a strong consensus of opinion developed among California opinion leaders that the state should become more like Massachusetts, that growth should be slowed and employee mobility dampened, trade secrets law could be made even more useful to employers maintaining internal labor markets without reversing any square holdings.

B. Do Silicon Valley Firms Explicitly or Implicitly Contract that Some or All Employees May Take Firm Information With Them When They Move On?

When I began my interviews, I was particularly interested in learning if there was any practice in the Valley of employees securing assurances that they can take certain ideas or projects with them if they start their own firms or move to another. I know that such negotiations occur between universities and star scientists.(21) I also knew that it has been quite a few years since anyone surveyed the internal practices of U.S. technology companies with respect to their willingness to permit employees to hold intellectual property in their own names.(22) In today's high-velocity labor markets, where employees are not tied to firms and may well be on the market again soon, do firms ever agree in advance that the employee will be relieved from secrecy obligations?

My survey was unscientific, but I was unable to find anyone who knows of this happening at the hiring stage. Conversation with numerous employees, managers, venture capitalists, and lawyers in Silicon Valley has revealed that all firms require new employees to sign form documents stating that everything they are working on belongs to the company; everything they work on after leaving the company for a year or two belongs to the company; no confidential or proprietary information is to be disclosed outside the company. All employees apparently sign these. No one to whom I spoke has ever heard of employees demanding, let along negotiating, anything different; or of firms that use documents that seem any different from anyone else's; or firms that cultivate the reputation of letting employees retain intellectual property rights.(23) (Some firms like IBM and Intel did have a reputation for being particularly likely to sue departing employees). Consequently, although I have not attempted to collect formal written employment contracts in Silicon Valley and compare them with a sample from Route 128, I haven't heard anything that suggests that such a difficult survey would be worth carrying out.

As in Neumeyer's little survey thirty years ago, it appears that all firms formally require invention assignments and confidentiality. In some law and economics models, as we shall see in Part V, this is strong evidence that efficient practice requires that firms, not employees, hold property rights in information and ideas. To my mind, this contracting process more likely reflects some combination of convention;(24) the in terrorem effect of contractual language that company counsel well knows is not fully enforceable; and the inability even of valuable employees to exert much negotiation pressure at the hiring stage, certainly not over issues concerning post-employment. In the next section, I will show why company counsel knows full well that it cannot really enforce those confidentiality clauses in a high-velocity labor market. In Part V, I will discuss the really crucial negotiation: not the hiring negotiation, which I believe is pro forma with respect to intellectual property issues, but the negotiation when the employee wants to leave.

C. The Costs of Enforcing Trade Secret Rights

The actual explanation of Silicon Valley's high-velocity labor market is the third hypothesis. Although employees depart Silicon Valley firms daily for competitors who will learn at least some information that the law calls a trade secret, employers rarely sue these employees or their new employers, for at least three reasons that can be documented. First, a few such highly-publicized suits accomplished little for plaintiffs, as will be seen from reviewing the journalistic coverage of these suits. Second, such suits imposed direct costs on these plaintiffs in reputation, internal morale, and recruiting. This will be confirmed from journalistic and interview accounts. Third, lawyers in the Valley who represent firms and venture capitalists concluded from this experience until recently that the working definition of trade secret in the Valley is narrower than the formal legal definition. They so advised clients, as they told me in interviews conducted mainly in March 1996, and deals are concluded on that basis.

Doubtless there are other reasons that formal trade secret rights are not enforced. Employers are normally both hirers and firers of labor so have interests in informal social norms discouraging all firms from trade secrets litigation. Undoubtedly managers and venture capitalists in the Valley identify its wealth with the kinds of factors that Saxenian discusses. They think everyone benefits in the long run from a fluid labor market. I cannot offer any reliable estimate of how widespread these attitudes are, but they exist.

1. The Peter Bonyhard litigation

One of the earliest trade secret suits against a departing computer scientist involved Dr. Peter Bonyhard, an IBM scientist expert in the technology of magnetic resonance heads for computer disk drives. It is not surprising that IBM should test these waters; by reputation it maintained a particularly traditional internal labor market and rarely hired from outside, particularly at that time, and was therefore more willing to sue departing employees.(25) Bonyhard left IBM in 1991 to work for rival Seagate Technology. IBM's suit attracted a great deal of publicity.(26) IBM succeeded in getting a preliminary injunction from a federal district court in Minnesota enjoining Bonyhard from disclosing confidential information or from working on magnetic resonance heads at Seagate. This injunction was reversed on appeal for failure to identify either the confidential information within its scope or the reasonable time for its effect.(27) The suit contributed to IBM's reputation as a firm willing to sue departing employees, but accomplished little else for IBM. Bonyhard remains at Seagate, and the case eventually settled with a confidential settlement including a patent cross-licensing agreement between Seagate and IBM.(28) As an analyst said of the settlement, "This is a major step in folks agreeing to trade information and work together, which is extremely important for the industry."(29)

2. The Alfred Chan- George Hwang litigation

Similar publicity attended the even more unusual litigation, criminal as well as civil, involving Alfred Chan's departure from Intel.(30) Chan, an engineer, had worked at Intel on their popular 80387 math coprocessor. He left in 1989 to join a new firm, ULSI Technology, Inc., to develop a math coprocessor that would be compatible with other Intel chips (and compete directly with the 80387). At some point Chan showed Hwang, his boss at ULSI, an early target specification for Intel's 387, marked "Intel proprietary."(31) (It remains unclear whether Chan, or another ULSI employee who had been moonlighting at Intel, was the source of the target specification). "It had been superseded by the version that went into manufacturing, and the key information that it contained had now been published by Intel in a trade magazine anyway."(32) Hwang, ULSI's President, called Intel and offered to open his databases and computers to Intel experts to show that he had no Intel proprietary information. Intel declined the offer but instead persuaded the District Attorney's office to prosecute the case as one of criminal theft, including a search of the ULSI offices themselves. The prosecutor had previously worked for Intel's law firm and cooperated closely with them throughout the prosecution, an arrangement common at the time but later criticized by the Supreme Court of California in a different case.(33)

My impression is that the criminal prosecution would not have been brought except for the documents, the nature of which was sharply disputed throughout the proceeding. In any of the prominent trade secrets cases that makes it into the media, there will always be invocations of "cartons of documents" or computer diskettes. Neither from an economic nor a legal point of view should it matter whether a trade secret is on a document or simply in an employee's head. If I am right that the issues are endogenous growth and incentive issues--which allocation of property rights creates the greatest economic growth and incentives for innovation?--it should be irrelevant whether the information is tangible or only in people's heads. However, observers of the field report that juries are much more impressed with documentary theft.(34)

The criminal and civil litigation lasted four years--the preliminary investigation in the criminal matter took up two and a half years of litigation. Intel was never able to demonstrate any theft of a trade secret, but was nevertheless able, by invoking the legal system, to prevent a potential competitor from coming onto the market during that time. Intel was unable to show that the documents contained any Intel trade secrets, and the jury found that neither Chan nor Hwang stole trade secrets. The criminal case ended in acquittal for all defendants. A federal injunction against ULSI for patent infringement was reversed on appeal.(35) Moreover, Intel suffered in internal morale and recruiting. The ULSI litigation apparently "lit up chat groups of engineers all over the Valley." During the litigation, Intel was constantly asked by potential recruits if they, too, would be sued if, as seemed likely, they would someday leave the company.(36) Intel acquired a reputation for being a bully towards its own employees, and may have paid for it.(37)

There is an interesting theoretical point here. "Reputational sanctions" play a large role in the economic literature on "implicit" employment contracts in explaining why such "implicit" contracts don't need legal enforcement. The claim is that an employer who breaches implicit promises to the workforce will have trouble hiring in the next round.(38) Most people who have thought about how labor markets transmit information about employers have concluded that reputational sanctions would not generally be that effective, and I have agreed with that conclusion before.(39) An interesting feature of the labor market for professionals in Silicon Valley is that this might be the exceptional labor market in which reputational sanctions are quite effective, as the Intel story shows. These professionals move around a great deal; have no cognitive disability against recognizing future separation; and are plugged into extremely effective on-line communication networks. I have argued that associations or even labor unions representing employees (not necessarily professionals) in high velocity labor markets should provide web sites and discussion groups for exchange of information about employers.(40)

It might appear from the Bonyhard and Chan tales that Silicon Valley employers would have learned not to assert trade secret rights against departing employees. The press coverage was relentlessly negative, as the headlines alone show, even in business journals.(41) I wanted to find out, though, if this is indeed the advice that lawyers give corporate clients. I also wanted to find out if venture capitalists have become more cautious about funding start-ups where employees have left a firm to go into competition with it. Lawyers and venture capitalists seemed to me, as to others, key figures in the transmission of information about business ventures and the structuring of the Valley.(42) To this end, in March 1996, I interviewed a dozen or so venture capitalists, and lawyers who advise them, about intellectual property issues in their recent start-ups. As I have said, this was not a scientific sample, in the sense that it reflected my contacts in the region. However, there is no reason to suspect any selection bias in those contacts with respect to the issue I was studying. In any case, there was no interesting disagreement among my subjects. The timing, however, was somewhat unfortuitous. One month later, the injunction in Advanced Micro Devices v. Hyundai issued, initiating a wave of litigation against employees.

Every one of my 1996 interviewees had been involved in a start-up where there were questions about the founders' rights to the intellectual property involved, usually, trade secrets. (One said that such issues arise in every start-up). Yet none had actually seen a deal fall apart because the former employer threatened the start-up with litigation over trade secrets, though some claimed to have heard of this happening to others. So how do these venture capitalists make sure that the information, held by the principals of the start-up, is nonrivalrous, and not an excludable asset of the former employer's?

Opinion letters from lawyers are often commissioned, stating that the new firm will have the right to exploit whatever the important information is. The lawyers who draft these letters do not report the kinds of uncertainty to which lawyers cheerfully admit in other contexts. They are confident they can spot potential litigation and have never been surprised. They employ a narrower definition of trade secret than any ever announced by a California (or any other Uniform Act) court. A useful composite of my interviews might be that a trade secret in Silicon Valley consists of a program, protocol, or design blueprint that exists in tangible form, such as a document or diskette, previously clearly marked as confidential and proprietary, never published or revealed to the trade, but nevertheless removed, downloaded, or emailed from the old employer without its consent. (In the Chan litigation, the jury did not find even such a specification to be a trade secret, probably because the important information had since been published). Lawyers do not anticipate litigation problems if the founders will use only information that is: (1) in their head, not tangible; (2) negative (I know what my old employer tried that didn't work) not positive (I know how to); or (3) consists only of feasible targets (I know that it's possible to achieve a 95 percent success rate) without the means of attaining them. Statements like these seem to me an accurate reflection of the practice of Santa Clara County judges and juries, at least as of March 1996. I do not believe, however, that there is any square holding of any California court that says that a trade secret must be tangible, or positive information, or a precise operational method.(43) I believe my informants were reflecting the more important consensus of opinion among Valley business professionals: that Saxenian and Romer are basically right; that the wealth of the Valley is built on the speed with which information that might be considered proprietary diffuses; that such information spread enables competition and rapid product development; and that firms are not dissuaded from new product development even though they know that their employees will soon be working elsewhere, perhaps in competition with them. If these attitudes among lawyers are changing, and there is some indication that they are(44)--there could be more litigation inhibiting competition and serious consequences for Silicon Valley's growth.

D. Aren't trade secrets more important in other industries?

It is possible that this conclusion is correct only for the computer industry in Silicon Valley, California, and for no other industry, in which case Silicon Valley lawyers have done a good thing by developing their narrow working definition of trade secret, but the story has no other legal implications. In other words, it might be that there are no legally-recognized trade secrets in computers, because there are no genuine secrets in the computer industry. The mean time to obsolescence of information is very short. Reverse engineering of products is feasible.(45) Diskettes can be copied. Finally, the labor market is exceptionally high-velocity. Under these circumstances, it may be argued, there really are few secrets--and we can be more certain that litigation nominally cast as trade secrets litigation is not really about protecting secrets, but rather for the illegitimate purpose of preventing competition by tying up a start-up in litigation.

It would be rash to reject this argument out of hand. But I think that an implication of Paul Romer's work is that it would be rash to accept it out of hand either. (Romer's work is by no means limited to software or Silicon Valley). The legal definition of trade secret must be sensitive, whatever the context, to the wealth of nonrivalrous information and the extent to which economic growth always requires the creation and diffusion of such information.

For example, it has been suggested to me that biotechnology might be an industry in which there are meaningful trade secrets in fact, that the law ought therefore to protect. While biotechnology companies hold intellectual property in the form of patents on drugs and microorganisms, those patents will be awarded, if at all, only after long research efforts, with many false leads and wasted efforts. During this long development process, the only body of intellectual property law that protects the company is trade secrets law. The areas of its research, the dead ends and negative information, preliminary findings, might all be trade secrets, and the company might genuinely suffer, it is claimed, if employees could, in the middle of a research project, freely jump to competitors and freely tell everything they know.

Perhaps. There is, however, evidence that even biotechnology makes little effective use of trade secrets. Because of the industry's academic roots, many discoveries that might have been considered trade secrets are instead published in academic journals.(46) While I have been unable to locate figures on mobility of personnel in biotechnology, so cannot compare them with the figures in the computer industry, many observers believe that personnel in biotechnology are quite mobile.(47) If this is so, it is interesting that there are almost no reported cases of biotechnology firms suing to enjoin departing employees from disclosing trade secrets--fewer cases even than the computer industry.(48) Perhaps biotechnology, like computers, illustrates instead that too many trade secrets are antithetical to growth; that more growth can be achieved through publishing information, informal know-how sharing, and diffusion of information through employee mobility and other means, than can be achieved through legally created monopoly.(49)

E. A Proposed Definition of Trade Secret

The legal implications of the economics of endogenous growth generally, and the Silicon Valley story in particular, are that: (1) economic analysis ought to become a standard feature of trade secret litigation; (2) plaintiffs should be required to identify specific information claimed to be a trade secret; (3) the court should, in all cases of proposed trade secret, specifically consider the likely public interest in the diffusion of the information; and, finally, (4) trade secrets should be limited to information that plaintiff can demonstrate would not have been created but for the ability to keep it secret.

By economic analysis, I do not mean only Romer's theories of endogenous technological change. I will examine two alternative approaches in the final Part of this Article. I do mean to shift the focus of litigation, from the somewhat sterile clash of morality plays described by litigators in the field, to questions of economic growth and incentives for the production of information.(50) The recommendation that trade secrets be limited to specific identified information is already the law everywhere that has rejected the doctrine of "inevitable disclosure," which should be rejected for reasons already discussed: it is the functional equivalent of giving every employer an implied covenant not to compete from each employee without the customary analysis of reasonableness, and is incompatible with the economic growth associated with employee mobility and information diffusion. These public and private interests should become a normal part of litigation over trade secrets.

That does not mean that these public interests should automatically trump plaintiff's claim of trade secrets in every case. As discussed in the economics discussion of Part II, the economic literature on nonrivalrous information assumes that there is a trade-off between the gains in encouraging and diffusing such information, and the losses in diminished firm incentives to produce information.

I would suggest that this become the precise inquiry in individual trade secrets cases. Plaintiffs should have to prove that the precise alleged trade secret would not have been created unless it could have been kept secret. It would not be enough to show (in the statutory phrase) that the information was useful to plaintiff in its business; that is true of much of the nonrivalrous information, produced in response to market incentives, that Romer models, such as (in Saxenian's examples) capacities of forthcoming products. Similarly, to return to Evan Chesler's hypothetical,(51) Employer 1's "negative information," that Production Process 1 cannot be significantly improved, would not be a trade secret. Employer 1 produced that information for its own needs and would have done so irrespective of background laws on secrecy. (A court might ask directly: would this information have been produced if it had to be published in a trade journal?) That information having been produced, there is no social advantage to keeping it secret, and considerable social advantage in its diffusion.

Trade secrets should consist only of information that, in a market economy, would not have been produced unless it could have been kept secret. A stock example, offered for a similar purpose by Dean Kronman, is the location of hidden mineral deposits. Society could put all mineral exploration into the exclusive jurisdiction of the Ministry of Mines, but experience (and economic theory) suggest that more information about the mineral wealth of society will be produced by private actors who are permitted to trade on the information they produce and hold secret.(52) A second, legitimate trade secret is a law firm's or investment bank's knowledge of a forthcoming tender offer. The production of either of these legal secrets assumes enforceable secrecy. Either employers anticipate (and create) a labor market with much lower velocity than Silicon Valley's, or, where they can't, can use courts to prevent departing employees from trading on these particularized, identifiable secrets.

A third example of a trade secret even under my definition, if the court got the facts right, might be the formula for a particular poultry vaccine. Two employees left the company that dominated that market and enabled a rival to enter it by replicating their former employer's production process. The court found that the production process was a trade secret because not generally known, though individual parts of it were, and awarded the first employer an injunction and damages. In my proposed analysis, the result may not change though the analysis differs. There is always a public interest in competition among producers and diffusion of knowledge. However, it is quite possible that the first employer could show that the vaccine would not have been created at all if it had to be shared.(53) Judicial infallibility is not to be expected under prevailing legal definitions or economic analysis, but, as the legal system starts to focus on the public interest in information diffusion, and the rough trade-off with private rights in information, it will develop a greater repertoire of examples and may come to enrich the economic analysis.

1. See also Joseph Bankman, The Structure of Silicon Valley Start-Ups, 41 UCLA L.Rev. 1737, 1739 (1994)("For the most part, Silicon Valley ventures are founded by employees of existing Silicon Valley companies."); Mark Charles Suchman, On Advice of Counsel: Law Firms and Venture Capital Funds as Information Intermediaries in the Structuration of Silicon Valley (PhD dissert., Stanford Univ., Sociology, 1994).

2. 59 I received my A.B. from Stanford University in 1972. Many of my closest college friends remained in that pleasant area, and nearly all of them work in or around high-technology industry. They were able to arrange many of the interviews.

3. 60 After versions of this Article had been presented at professional meetings and distributed on-line, I was able to benefit from the interesting study by Bui-Eve, supra n.52, which is largely consistent with my understanding of the practice of hiring.

4. 61 Gilson, supra n.38.

5. 62 See supra n.41.

6. Supra n. 57.

7. 64 Gordon v. Landau, 321 P.2d 456 (Cal. 1958)(agreement making customer lists confidential is not a "contract by which anyone is restrained from engaging in a lawful profession, trade, or business" under Cal.Bus.& Prof.Code 16600, the statute that makes such covenants unenforceable). The basic proposition that trade secrets agreements fall outside 16600 has been reaffirmed many times, though most frequently in dicta, e.g. Fowler v. Varian Associates, Inc., 241 Cal.Rptr. 539, 545 (Ct.App. 1987)("agreements designed to protect an employer's proprietary information do not violate section 16600."). The statement in Edmund W. Kitch, The Law and Economics of Rights in Valuable Information, 9 J.Leg.Stud. 683, 710 n.74 (1980), that there is no exception to 16600 "for restrictions reasonably necessary to protect a trade secret", only "dictum to the contrary," is simply wrong: Gordon v. Landau is a square holding construing the statute not to apply at all to trade secrets agreements.

8. 65 There is no California analog to the often-anthologized, rarely-followed defense of information diffusion expressed almost forty years ago by the Pennsylvania Supreme Court. Greenberg, an industrial chemist, was employed to analyze and duplicate competitors' cleaning compounds. He left to join a sales competitor which, for the first time, set up a manufacturing arm, making products with formulas identical to those made by Greenberg's earlier employer. The court expressed concern with any limitation on the mobility of skilled technical employees. It noted that the formulas were developed by Greenberg himself, not "disclosed" by Employer 1. Finally, it noted the need to balance incentives to employers to invest in research, with effects on employee mobility, and, unusually, noted a public interest: "[S]ociety suffers because competition is diminished by slackening the dissemination of ideas, processes, and methods." Wexler v. Greenberg, 160 A.2d 430 (Pa. 1960).

9. 66 Lawrence M. Fisher, The Winery, Not the Winemaker, Owns the Secrets, a Court Rules, N.Y. Times, July 2, 1992, at A1, col. 5. An old example of use of trade secret law against a departing employee with exceptionally harsh relief is By-Buk Co. v. Printed Cellophane Tape Co., 329 P.2d 147 (Cal.App.1958), ordering Employer 2 to dismantle machines that it had built after Employee had told them how Employer 1 had set up its machines. Employer 2 was liable for an accounting for all sales, and punitive damages. While the case predates the adoption of the Uniform Act and has not been heard from in some time, it has never been overruled. If it were still good law in California, there could hardly be a start-up in the Valley. For the broadest definition of "trade secret" employed in a departing employee case, by a California court, that I could find (in a pre-Act, common law case), see Sinclair v. Aquarius Electronics, Inc., 116 Cal.Rptr. 654, 658 (Ct.App. 1974)("a trade secret in the broad sense consists of any unpatented idea which may be used for industrial and commercial purposes.").

10. 67 A recent tale that sounds as if it came from "Dilbert" had Dow Chemical Co. suing General Electric, accusing them of "systematically and aggressively" hiring employees privy to trade secrets. The key "secret", alleged to have escaped, was a confidential marketing study, written for Dow, with estimated demands for various plastics. It quickly emerged that the departed employee, who had written this study while at Dow, had obtained nearly all the material from a public presentation at a trade conference--by a General Electric manager. William M. Carley, Don't Read This! There is Stuff Here That's Confidential, Wall St. J., April 7, 1997, at A1, col.4. Dow then dropped its suit without any financial compensation. Kenneth N. Gilpin, Dow and G.E. Resolve Suit on Theft of Trade Secrets, N.Y. Times, April 19, 1997, at D2, col.5.

11. 68 Advanced Micro Devices, Inc. v. Hyundai Electronics America, No. CV752679, California Superior Court, Santa Clara County, Apr. 5. 1996 (Turrone, J.). The order is not available on-line or in any publication that I have found. Thanks to Ian Feinberg, Esq. for sending me a copy.

12. Dan Goodin, AMD, Hyundai Resolve Litigation Over Trade Secrets, The Recorder, Nov. 19, 1996, at 1 (available in LEXIS, News data base); Crista Hardie, AMD, Hyundai Settle Flash Squabble, 42 Electronic News, Nov. 25, 1996, at 16 (available 1996 WL 13933861).

13. Personal communication, Ian Feinberg, Esq., Aug. 26, 1998; see also Goodin, supra n. 69 (quoting Hyundai press release).

14. 71 Terrence P. McMahon, Gary E. Weiss & Sean A. Lincoln, Inevitable Disclosure: Not So Sure in the West; California's Laws Don't Bar Use of Inevitable Disclosure Theory, 19 Nat'l.L.J., May 12, 1997, at C35, col.3; Gary E. Weiss, Sean A. Lincoln & Erin M. Farrell, Protecting Trade Secrets: A Primer for California Start-ups, in Documentation for Start-up and Emerging Companies (Education Committee, Business Law Section, State Bar of California, 1997).

15. Supra n. 57.

16. The PepsiCo case is discussed supra nn. 52-56.

17. Ian C. Ballon, Keeping Secrets, IP Magazine, March 1998, available on-line at <http://www.ipmag.com/98-mar/ballon.html>, reporting on GTE Government Systems Corp. v. Patrick, No. 770111; Ian N. Feinberg, Inevitable Disclosure of Trade Secrets: A New Problem for Companies Hiring Experienced Technical Employees, available on-line at <http://www.gcwf.com/articles/interest/interest_5.html>, reporting on IBM v. Read-Rite, Inc.; Gary E. Weiss, personal communication to author, August 21, 1998, reporting on GTE v. Patrick, supra, and National Semiconductor v. Allouche).

18. 75 A distinguished California jurist suggested years ago in dicta that Gordon v. Landau, supra n.62, the case exempting trade secrets agreements from the statutory prohibition on covenants not to compete, was limited to agreements that make confidential what is already a statutory trade secret. It does not authorize employers and employees to make confidential some information not already a statutory trade secret and then authorize injunction against the disclosure of this "contractual trade secret." State Farm Mut. Auto Ins. Co. v. Dempster, 344 P.2d 821, 825-26 (Cal.D.Ct.App. 1959)(Tobriner, J.)(customer lists). I have been unable to locate any cases speaking to the issue and, frankly, given the breadth of the current statutory definition of "trade secret", don't think it could arise.

19. 76 In Herzog v. "A" Co., Inc., 188 Cal.Rptr. 155, 158 (Ct.App. 1982), the court held that Employer 1 could be sued for tortious interference with contractual relations if the effect of its letter to Employer 2 was to induce Employer 2 to renounce an offer of employment, and if Employee can show that the job at Employer 2 would not necessarily have put him in breach of the secrecy provisions of his employment contract with Employer 1. The "offending letter facially exceeds any legitimate purpose when it threatens Herzog and his future employers with suit based merely upon the fact of employment." There do not, however, appear to have been many successful cases by employees of this type.

20. 77 Recent cases refusing to enjoin departing employees cite grounds that in theory are recognized everywhere. See, e.g., Hilb, Rogal and Hamilton Insurance Services of Orange County, Inc. v. Robb, 39 Cal.Rptr.2d 887, 891-92 (Ct.App. 1995)(employee's announcing job move to customers of former employer is not solicitation or improper use of trade secret); Metro Traffic Control, Inc. v. Shadow Travel Network, 27 Cal.Rptr.2d 573, 577-80 (Ct.App. 1994)(employer failed to show that radio announcers were hired for "information related to them" as opposed to "personal qualities": "[A] stable of trained and talented at-will employees does not constitute an employer's trade secret.").

21. 78 Supra n.34.

22. 79 Neumeyer studied nine large U.S. companies in the late 1960s (General Motors, US Steel, IBM, Westinghouse, RCA, TRW, Gulf Oil, Polaroid, and Bell Laboratories) and discovered that all required employees to assign all their inventions to the company, using very similar documents. Fredrik Neumeyer, The Employed Inventor in the United States: R & D Policies, Law, and Practice 85-95 (1971). The survey of 650 research and development managers published as Richard C. Levin, Alvin K. Klevorick, Richard R. Nelson, & Sidney G. Winter, Appropriating the Returns from Industrial Research and Development, 3 Brookings Papers on Economic Activity 783 (1987) inquired about propensities to patent and the costs and benefits of patenting, but did not ask any "inside the firm" questions about compensation or other contractual relations with employees.

23. 80 For example, Neumeyer, supra n.79 at 149, reported that Polaroid granted all requests to modify the formal contract that ostensibly required assignment to it of inventions developed within two years of employment and forbade employment by, or advice to, a competitor during that period. I could not learn of any Silicon Valley employer with this kind of reputation.

24. 81 In labor economics, it is repeatedly discovered that firms rarely administer their compensation plans for maximum economic advantage. They are much likelier to do what everyone else in their industry does. Allen Cheadle, Explaining Patterns of Profit-Sharing Activity, 28 Indus.Rel. 387 (1989); Susan Chaplinsky & Greg Niehaus, The Tax and Distributional Effects of Leveraged ESOPs, 19 Fin.Mgmt. 29, 36 (1990).

25. 82 See also International Business Machines Corp. v. Zachariades, 1993 WL 443409 (N.D.Cal. 1993), aff'd in part, rev'd in part 70 F.3d 1278 (table), 1995 WL 697210 (text)(9th Cir. 1995)(employee patents in his own name alleged to be breaches of invention assignment and confidential information clauses of written employment agreement; some, not all, claims held time-barred); IBM v. Read-Rite, Inc., discussed in Feinberg, supra n.74.

26. 83 See, e.g., James Bennet, Who Owns Ideas, and Papers, Is Issue in Company Lawsuits, N.Y. Times, May 30, 1994, at 1, col. 5; Victoria Slind-Flor, More Trade Secrets Wars, Nat.L.J., March 22, 1993, at 1, col. 2; Michael W. Miller, Workplace: IBM Sues to Silence Former Employee, Wall St. J., Jul. 15, 1992, at B1.

27. 84 Seagate Technology, Inc. v. International Business Machines Corp., 962 F.2d 12 (table), 1992 WL 96271 (unpublished text)(8th Cir. 1992)(lifting preliminary injunction and remanding to district court), on remand 941 F.Supp. 98 (D.Minn. 1992)(denying preliminary injunction).

I've noticed that company counsel, and credulous courts, continue to cite the original district court opinion in Seagate, enjoining Bonyhard's supposed inevitable disclosure of IBM secrets, without noting either that (1) that injunction was reversed on appeal, for failure to identify any specific trade secret, or (2) the district court, on remand, refused to enter any injunction at all. See, e.g., Merck & Co. v. Lyon, 941 F.Supp. 1443, 1460 (M.D.N.C. 1996)(citing original district court opinion without noting reversal on appeal); PepsiCo Inc. v. Redmond, 1996 WL 3965, *18 (N.D.Ill. 1996)(citing original district court opinion; misdescribing reversal in Court of Appeals as "remanded for more specific statement of relief"; failing to note second district court opinion, denying any relief).

28. 85 International Business Machines: Trade-Secret Dispute Settled with Seagate Technology, Wall St. J., Nov. 16, 1994, at B16.

29. 86 40 Electronic News, Nov. 21, 1994, available on Westlaw PCNEWS database, 1994 WL 2830269.

30. 87 Bennet, supra n. 83; Slind-Flor, supra n.83; Victoria Slind-Flor, Charges Fly in Chip War, Nat.L.J., Apr. 12, 1993, at 1, col. 2.

31. 88 The legend "Intel Proprietary" was not necessary of any significance. "[Advanced Micro Devices] discovered , in one of its many court battles against [Intel] during the 1980s, that Intel routinely kept stacks of paper next to its photocopiers preprinted with the legend 'Intel confidential'." Tim Jackson, Inside Intel 406 (1997).

32. 89 Jackson, supra n.88, at 287.

33. 90 Compare People v. Eubanks, 927 P.2d 310 (Cal. 1996)(county district attorney's office should have been disqualified from prosecuting alleged criminal theft of trade secrets by departing employees, where alleged victim paid investigation costs); Lawrence M. Fisher, Charges Dismissed in Silicon Valley Trade Secrets Case, N.Y. Times, Nov. 23, 1996, at 33, col. 3.

34. 91 Bennet, supra n.83; Slind-Flor, supra n.83.

35. 92 Third Blow for Intel, Nat. L.J., June 28, 1993, at 19, col.1; Brian Fuller, Cleared, ULSI's Hwang picks up pieces, Elec.Eng.Times, Sept. 20, 1993 (available on Westlaw, ALLNEWSPLUS database), Jackson, supra n. 88, at 284-93.

36. 93 This information was given to me by someone with access to higher levels at Intel, under strict instructions that I not disclose the source. My source has concluded that Intel in the future will just "let people go," citing the departure, recent at the time of the interview in March 1996, of two vice-presidents. I have not seen any litigation by Intel against departing employees since the ULSI litigation ended.

37. 94 See, e.g., Don Clark, Intel Lawyer Commands Chip War--Rivals Say Dunlap is a Darth Vader Guarding a Microchip Empire, San Francisco Chronicle, June 28, 1993, at E1 (review of Intel's large quantity of mostly unsuccessful litigation; charges against it of stealing exhibits). Today, one can visit a web site maintained by disgruntled present and former Intel employees, reviewing litigation -mostly several years old now--between Intel and former employees, <http://www.igc.org/faceintel>. For discussion of the legal issues raised by such a form of "employee organization", see Alan Hyde, Employee Caucus: A Key Institution in the Emerging System of Employment Law, 69 Chi.-Kent L.Rev. 149, 155-71 (1993), reprinted in The Legal Future of Employee Representation 146, 152-69 (Matthew W. Finkin ed. 1994); Elena N. Broder (student author), (Net)workers' Rights: The NLRA and Employee Electronic Communications, 105 Yale L.J. 1639 (1996).

38. 95 See, e.g, Sherwin Rosen, Implict Contracts: A Survey, 23 J.Econ.Lit. 1144 (1985); Edward P. Lazear, Why Is There Mandatory Retirement?, 87 J.Pol.Econ. 1261, 1271 (1979).

39. 96 Paul C. Weiler, Governing the Workplace: The Future of Labor and Employment Law 73-78 (1990); David Charny, Nonlegal Sanctions in Commercial Relationships, 104 Harv.L.Rev. 373, 417-20 (1990); Susan Rose-Ackerman, Progressive Law and Economics--and the New Administrative Law, 98 Yale L.J. 341, 355-57 (1988); Carl Shapiro & Joseph E. Stiglitz, Equilibrium Unemployment as a Worker Discipline Device, 74 Am.Econ.Rev. 433, 442 (1984); Alan Hyde, In Defense of Employee Ownership, 67 Chi.-Kent L.Rev. 159, 179 n.59 (1991). Indeed, the locus classicus of the theory of relational sanctions, if read carefully, would normally exclude labor markets. Benjamin Klein & Keith B. Leffler, The Role of Market Forces in Assuring Contractual Performance, 89 J.Pol.Econ. 615 (1981).

40. 97 Alan Hyde, Employee Organization in High Velocity Labor Markets, 50 Proc.NYU Conf.Lab. (1997)(forthcoming).

41. 98 See also Peter Burrows, A Nest of Software Spies?, Business Week, May 19, 1997, reporting a criminal indictment alleging that departing employees of Cadence Design Systems stole software. "[L]itigation is considered bad form in Silicon Valley, where the honorable way to win market share is through innovation. During the fight, [Cadence CEO Joseph B.] Costello was lambasted in the media as a bully and his board, customers, and friends urged him to give up the chase."

42. 99 Suchman, supra n.58.

43. Indeed, two California decisions have stated that negative information, such as potential customers who have never ordered products, might be a trade secret, although neither case seems to turn on the point. Morton v. Rank America, Inc., 812 F.Supp. 1062, 1073-74 (C.D. Cal. 1993); Courtesy Temporary Service, Inc. v. Camacho, 272 Cal.Rptr. 352 (Ct.App. 1990).

44. Supra n.74.

45. 102 "Reverse engineering" starts from an existing product and figures out how it was made. "It is well recognized that a trade secret does not offer protection against discovery by fair and honest means such as by independent invention, accidental disclosure, or by so-called reverse engineering, that is, starting with the known product and working backward to divine the process." Sinclair v. Aquarius Electronics, Inc., 116 Cal.Rptr. 654, 661 (Ct.App. 1974)(citation omitted)(pre-Act, common law). This statement continues to be true under the Uniform Act, which excludes from its definition of trade secret information "readily ascertainable by proper means", UTSA 1(4). An Official Comment states that reverse engineering is "proper means". While California's version of the Act omits the quoted language, it also specifically provides that "reverse engineering or independent derivation alone shall not be considered improper means." Cal.Civ.Code 3426.1.

Not only is reverse engineering neither illegal nor immoral, a plaintiff who fails to "reverse engineer" may not have mitigated contract damages in California. Advanced Micro Devices, Inc., v. Intel Corp., 885 P.2d 994 (Cal. 1994) affirmed a complicated commercial arbitration growing out of a collapsed technology exchange agreement. The arbitrator found that Intel had breached the agreement by refusing to let AMD be a second source for Intel's popular 386 chip and by refusing, in bad faith and without disclosing to AMD what it was doing, to take any AMD products. (If Intel took no AMD products, it would not be obliged to share its technology with AMD). The arbitrator awarded AMD a permanent, nonexclusive and royalty-free license to any Intel intellectual property embodied in the 386 chip that AMD eventually produced. However, the arbitrator declined to award AMD any lost profits, on the specific grounds that AMD unnecessarily delayed its independent entry into the 386 chip market. Having "inferred by mid-1985 that Intel was not going to accept the AMD parts that could earn AMD the 80386, AMD should have sought arbitration at that time or immediately begun reverse engineering the 80386 when it became available in July 1986." 885 P.2d at 998. The Supreme Court of California enforced both the arbitrator's award (which was much more complicated than this footnote conveys) and the finding that AMD's failure to reverse engineer denied it lost profits, although the opinion rests in part on a deferential standard for review of arbitral remedies. The controversial issue in the Supreme Court was not the denial of lost profits, but the award to AMD of royalty-free property in the 80386. This exceeded any rights that AMD had under the contract, which, if performed in good faith by Intel, would at most have permitted AMD the right to "earn" the 80386, and pay royalties on it, in exchange for some AMD technology taken by Intel.

46. 103 Rebecca S. Eisenberg, Proprietary Rights and the Norms of Science in Biotechnology Research, 97 Yale L.J. 177 (1987).

47. 104 See, e.g., Josh Lerner, Patenting in the Shadow of Competitors, 38 J.L.& Econ. 463, 472 (1995).

48. 105 A rare plaintiff's victory was Salsbury Laboratories, Inc. v. Rhone Merieux Labs, Inc., 908 F.2d 706 (11th Cir. 1990). A defendant's victory was Stratagene, Inc. v. Huse, 935 F.2d 280 (Fed.Cir.), cert. denied 502 U.S. 941 (1991). Thanks to Josh Lerner.

49. 106 Walter W. Powell, Inter-Organizational Collaboration in the Biotechnology Industry, 152 J.Inst.& Theoret.Econ. (1996); Henry Hansmann, What Determines Firm Boundaries in Biotech?, 152 J.Inst.& Theoret.Econ. (1996). My contacts in biotechnology confirm that information sharing is widespread. They have suggested that this partly reflects those very long research times. You can be confident, if you answer a competitor's question or even share a cell culture, that they truly are using the information only for their own purposes. Even if they know what's happening in your lab, they will be too far behind you to be able to scoop you. Curtiss Jameson, e-mail to author, March 1997.

50. 107 Published, and my private, interviews with trade secrets litigators reveal the moralistic strategies followed by both sides. Trade secret lawyers are not, by and large, patent lawyers, and the cases are not the technical inquiries into originality that typify the patent field. Trade secrets lawyers describe their victories as moralized narratives. The plaintiff company wins if the employee can be made to seem an ungrateful, sneaky, lying little thief, biting the hand that fed him for so long. The defendant employee wins if the plaintiff employer can be made to seem an overbearing, feudalistic gorilla, interested only in insulating itself from the normal market competition to which the rest of us are subject, and heedless of the livelihoods of individuals whom it crushes in its illegitimate drive to prevent competition. "Many experienced intellectual property litigators say that while judges and juries often are overwhelmed by the technical intricacies of a patent or copyright infringement claim, they seldom have the same degree of difficulty with a trade secret claim. 'I just have to show something was taken improperly,' says Michael E. Epstein, chairman of the technology group at Weil, Gotshal. 'That's very different from an infringement analysis that is so very technical.'" Victoria Slind-Flor, More Trade Secrets Wars, Nat.L.J., March 22, 1993, at 1, 34. A well-known example of such moralistic litigation is Roberts v. Sears, Roebuck, 573 F.2d 976 (7th Cir.), cert. Denied, 439 U.S. 860 (1978), on remand, 471 F.Supp. 372 (N.D. Ill. 1979)(rescinding employee's assignment of patent and awarding restitution of profits), vacated, 617 F.2d 460 (7th Cir. 1980)(election of fraud damages precludes restitutionary relief), cert. denied 449 U.S. 975. While the cases often suggest that employers secure individual negotiated assignment of intellectual property rights from employees, this case gave close scrutiny to, and set aside, just such an assignment. Roberts, when an eighteen-year-old sales clerk at Sears, designed and constructed a socket wrench from which Sears made a great deal of money. The court found that Sears obtained Roberts' assignment by fraud, giving him inaccurately pessimistic assessments of the invention's patentability and commercial potential. The tale is a comfortable morality play in which the big corporation is rebuked in its attempt to fleece the naif. There is no economic analysis, even in the Seventh Circuit. Since Sears was obviously a better marketer of wrenches than Roberts, it seems likely to me that Roberts would have assigned the invention to Sears even if more accurately informed of the invention's potential. The case may rest on an inarticulate guess that the public will benefit if employees like Roberts get more incentives for creativity, and conversely that the public would suffer if employees believe they will be swindled and never create in the first place.

51. 108 Supra TAN 29, n.56.

52. 109 Anthony T. Kronman, Mistake, Disclosure, Information, and the Law of Contracts, 7 J.Leg.Stud. 1 (1978).

53. 110 Salsbury Laboratories, Inc. v Merieux Laboratories, Inc., 735 F.Supp. 1555 (M.D. Ga. 1989), affirmed as modified, 908 F.2d 706 (11th Cir. 1990).