Trade secrets are like pie recipes, says a California appeals court. The baker who steals the recipes might be liable for misappropriating trade secrets, but the customer who buys a pie may eat it without fear of liability.

In Silvaco Data Systems v. Intel Corp., the court compared computer source code to a proprietary pie recipe that may be legally protected. However, executable object code is like the pie itself: Since the consumer does not learn the recipe by eating the pie, he does not share the baker’s liability for its misappropriation.

Silvaco protects as a trade secret the source code of its software used to automate the design of electronic circuits. Some years ago, Silvaco won an injunction barring a competitor from selling a similar product after convincing a judge that the competitor had misappropriated its trade secret.

Silvaco tried to build on this victory with a strategy that most businesses—which typically are averse to alienating potential customers—would avoid: It sued the customers of the misappropriator . . . for misappropriation.

Specifically, the lawsuit charged that Intel had bought and used software from Silvaco’s competitor with knowledge that Silvaco had accused that company with incorporating stolen trade secrets in its program. The case was brought under the Uniform Trade Secrets Act, which 45 states, including California, have adopted.

The act forbids the wrongful “acquisition,” “disclosure” or “use” of a trade secret. The court was quick to reject any notion that Intel ever came into possession of Silvaco’s source code. As a consequence, Intel could not have disclosed what it never acquired. This analysis left only “use” as a possible basis for Intel’s liability.

Silvaco was reduced to arguing that the executable code incorporates the same “information” as the source code from which it is compiled, so that executing it constitutes “use” of the trade secrets embodied in the source code. The company asserted also that its secrets included certain methods and algorithms that were “used” in the executable code.

Silvaco’s case was doomed by its own admission that the object code executed by Intel could not communicate the underlying source code or allow exploitation of its design. The object code “could not, in short, impart knowledge of the trade secret,” said the court.

The kind of “use” that the trade secret act prohibits is the direct exploitation of a secret for one’s own use, for example, by incorporating it into one’s own product or manufacturing technique. “Use” does not extend, says the court, to exploiting “something that was made using the secret.” And here is where the court turns to pie:

One who bakes a pie from a recipe certainly engages in the “use” of the latter; but one who eats the pie does not, by virtue of that act alone, make “use” of the recipe in any ordinary sense, and this is true even if the baker is accused of stealing the recipe from a competitor, and the diner knows of that accusation. . . . The source code is the recipe from which the pie (executable program) is baked (compiled).

Intel, like the baker’s customer, did not acquire knowledge of the recipe by buying or eating the pie.

The California court pungently criticized the contrary reasoning of two courts in other states. A Massachusetts judge had suggested that use of object code alone could somehow constitute misappropriation of a trade secret. The other court, in Utah, had held a software customer liable for misappropriation of source code for knowing that his seller had derived the underlying code through improper means.

The Silvaco court all but labeled the reasoning of these courts slipshod. Executing machine-readable software cannot constitute “use” of the underlying source code, because it provides no way for obtaining the trade secret information. And any wrongdoing of a seller in acquiring a trade secret cannot be imputed to a customer who does not himself obtain knowledge of the secrets.

The California panel cited public policy for its conclusions. If simply running software is deemed use of the source code from which it was compiled, there would be no end of unreasonable risk. If the program were later alleged to be based on stolen code, purchasers would be exposed to liability for misappropriation. The resulting devastation to software sales, and to innovation itself, cannot be accepted as a purpose of the trade secret laws.

The court’s differentiation between baker and pie-eater is explicitly intended to protect competition by restricting liability for misappropriation to persons actually involved in the misappropriation, and to give a pass to commercial customers who purchase products on the open market, even if the customers know the product is accused of incorporating trade secrets.

This is a wise distinction. Unlike a purchaser of goods known to be stolen, the software customer does not actually gain access to the asset that the trade secret owner strives so mightily to protect, namely, the source code underlying the product.

The Silvaco opinion, if its reasoning gains wide acceptance, would limit the legal recourse of developers whose source code is misappropriated. They could now target only those persons who actually acquired or used the source code itself, and must leave the alleged misappropriator’s customers alone.