EASTERBROOK, Circuit Judge.

Both IDX Systems and Epic Systems make software for use in managing the financial side of a medical practice: billing, insurance reimbursement and other collections, and the like. During the 1980s IDX sold this software package to two medical groups that later merged into the University of Wisconsin Medical Foundation, which now comprises more than 1,000 physicians. The Foundation continued to use IDX software until December 2000, when it switched to software developed by Epic. IDX believes that Mitchell Quade and Michael Rosencrance, former employees of Epic who came to manage data processing at the Foundation, not only instigated this change but also used their new positions to transfer valuable information to Epic. According to IDX's complaint, over the course of a year Quade and Rosencrance personally, and with the aid of other Foundation employees, furnished Epic with details about how IDX's software works, enabling Epic to enhance its own package and ultimately take the Foundation's business — and to match up better against IDX in the competition for other customers.

IDX's complaint under the diversity jurisdiction of 28 U.S.C. § 1332 charges the Foundation, Quade, and Rosencrance with stealing IDX's trade secrets and breaking contractual promises of confidentiality; it charges Epic with tortiously inducing the other defendants to do these things. The district court dismissed the tort claims against Epic on the pleadings, observing that Wis. Stat. § 134.90(6)(a) overrides any theory that conflicts with the state's law of trade secrets. Later it pared all contract-based claims out of the case, ruling that the confidentiality agreements are invalid under Wisconsin law (which the parties agree governs) because they do not contain temporal and geographic limitations. Finally, the court granted summary judgment to the defendants on the trade-secret claim, after concluding that IDX had failed to identify with specificity the trade secrets that it accuses the defendants of misappropriating. 165 F.Supp.2d 812 (W.D.Wis. 2001). We shall start where the district court ended: with the trade-secret claim.

Trade secrets are a subset of all commercially valuable information. Wisconsin has followed the Uniform Trade Secrets Act in defining "trade secret" this way:

"Trade secret" means information, including a formula, pattern, compilation, program, device, method, technique or process to which all of the following apply:

1. The information derives independent economic value, actual or potential, from not being generally known to, and not being readily ascertainable by proper means by, other persons who can obtain economic value from its disclosure or use.

2. The information is the subject of efforts to maintain its secrecy that are reasonable under the circumstances.

Wis. Stat. § 134.90(1)(c). Thus to show that particular information is a trade secret, a firm such as IDX must demonstrate that it is valuable, not known to others who might profit by its use, and has been handled by means reasonably designed to maintain secrecy. Like the district judge, we think that IDX failed to do this. It has been both too vague and too inclusive, effectively asserting that all information in or about its software is a trade secret. That's not plausible — and, more to the point, such a broad assertion does not match up to the statutory definition. Reluctance to be specific is understandable; the more precise the claim, the more a party does to tip off a business rival to where the real secrets lie and where the rival's own development efforts should be focused. Still, tools such as protective orders are available to make this process less risky, and unless the plaintiff engages in a serious effort to pin down the secrets a court cannot do its job.

According to IDX, "a 43-page description of the methods and processes underlying and the inter-relationships among various features making up IDX's software package" is specific enough. No, it isn't. These 43 pages describe the software; although the document was created for this litigation, it does not separate the trade secrets from the other information that goes into any software package. Which aspects are known to the trade, and which are not? That's vital under the statutory definition. Likewise, IDX's tender of the complete documentation for the software leaves mysterious exactly which pieces of information are the trade secrets. As we remarked in Composite Marine Propellers, Inc. v. Van Der Woude, 962 F.2d 1263, 1266 (7th Cir. 1992), a plaintiff must do more than just identify a kind of technology and then invite the court to hunt through the details in search of items meeting the statutory definition. See also AMP Inc. v. Fleischhacker, 823 F.2d 1199, 1203 (7th Cir. 1987). What is more, many of the items that appear in the 43-page description, such as the appearance of data-entry screens, are exceedingly hard to call trade secrets: things that any user or passer-by sees at a glance are "readily ascertainable by proper means". Perhaps screen displays could be copyrighted, but no copyright claim has been advanced, and a trade-secret claim based on readily observable material is a bust. Minnesota Mining Manufacturing Co. v. Pribyl, 259 F.3d 587 (7th Cir. 2001), on which IDX principally relies, did not involve such self-revealing information. Other details, such as the algorithms that the software uses to do real-time error checking (a vaunted feature of IDX's software), may be genuine trade secrets, but IDX has not tried to separate them from elements such as its input and output formats. Nor does it contend that the defendants decompiled the object code or otherwise obtained access to the algorithms that power the program; it alleges only that Foundation transferred to Epic those details that ordinary users of the software could observe without reverse engineering.

Because (as what we have already written illustrates) it is hard to prove that particular information qualifies as a trade secret, many producers of intellectual property negotiate with their customers for additional protection. This is a step that Wisconsin permits. See ProCD, Inc. v. Zeidenberg, 86 F.3d 1447 (7th Cir. 1996) (Wisconsin law). Following § 7 of the Uniform Trade Secrets Act, Wis. Stat. § 134.90(6) provides:

(a) Except as provided in par. (b), this section displaces conflicting tort law, restitutionary law and any other law of this state providing a civil remedy for misappropriation of a trade secret.

(b) This section does not affect any of the following:

1. Any contractual remedy, whether or not based upon misappropriation of a trade secret.

2. Any civil remedy not based upon misappropriation of a trade secret.

3. Any criminal remedy, whether or not based upon misappropriation of a trade secret.

IDX and the Foundation (through its predecessors in interest) agreed to the sort of contractual remedy preserved in § 134.90(6)(b)1. The Foundation promised not to allow the software and related materials "furnished by" IDX to be "examined . . . for the purpose of creating another system" and vowed in addition not to "use or disclose or divulge to others any data or information relating to" the system or "the technology, ideas, concepts, know-how, and techniques embodied therein." IDX has evidence (enough to survive summary judgment) that the Foundation, Quade, and Rosencrance broke these promises by describing IDX's system in detail to Epic and helping it duplicate those features that the Foundation liked. Nonetheless, the district judge held, the promises are unenforceable because they are unlimited in temporal and geographic scope, and thus unduly restrain trade.

In reaching this conclusion, the district court relied on decisions requiring restrictive covenants limiting competition between employers and their ex-employees to be reasonable, a limitation that in Wisconsin entails some restrictions on time and scope. See Farm Credit Services v. Wysocki, 243 Wis.2d 305, 627 N.W.2d 444 (2001); Tatge v. Chambers Owen, Inc., 219 Wis.2d 99, 579 N.W.2d 217 (1998); Streiff v. American Family Mutual Insurance Co., 118 Wis.2d 602, 348 N.W.2d 505 (1984); Wis. Stat. § 103.465. Rules limiting the extent of no-compete clauses are based on the fact that they tie up human capital and, if widely adopted, may have the practical effect of preventing horizontal competition in economically significant markets. See Paul H. Rubin Peter Shedd, Human Capital and Covenants Not to Compete, 10 J. Legal Studies 93 (1981). But neither rationale applies to contracts that restrict the use of particular information between businesses that have vertical (supplier-to-customer) rather than horizontal (competitor-to-competitor) relations. See Benjamin E. Hermalin Michael L. Katz, Judicial Modification of Contracts Between Sophisticated Parties: A More Complete View of Incomplete Contracts and Their Breach, 9 J.L. Econ. Org. 230 (1993); Robert Unikel, Bridging the Trade Secret Gap, 29 Loyola U. Chi. L.J. 841 (1998). IDX did not contract for limitations on Epic's ability to compete; contracts between IDX and the Foundation are vertical in nature and protect intellectual property without affecting competition. They may compel rivals such as Epic to do more work to develop software independently, but this promotes rather than restricts competition. Kewanee Oil Co. v. Bicron Corp., 416 U.S. 470, 94 S.Ct. 1879, 40 L.Ed.2d 315 (1974), holds that trade-secret law is compatible with antitrust law; the same can be said for contracts protecting intellectual property that, though not demonstrably a trade secret, is commercially valuable. Rivals such as Epic, as non-parties to the vertical arrangements, remain entitled to discover and use the information independently and to compete vigorously. Nothing in the antitrust laws gives one producer a right to sponge off another's intellectual property, even when the producer of that knowledge has a market share much larger than IDX's. See United States v. Microsoft Corp., 253 F.3d 34 (D.C. Cir. 2001) (en banc).

The parties have not cited, and we have not found, any Wisconsin statute or decision subjecting non-disclosure agreements between suppliers and users of intellectual property to the rules that govern non-competition clauses between employers and employees. To the contrary, Fullerton Lumber Co. v. Torborg, 270 Wis. 133, 139, 70 N.W.2d 585, 588 (1955), tells us that Wisconsin allows "a much greater scope of restraint in contracts between vendor and vendee than between employer and employee." Section 134.90(6)(b)1 implies that contracts about intellectual property are valid, even when they exceed the domain of trade secrets. Restrictions on disclosure may make intellectual property more valuable to its producer, and thus promote both the creation of knowledge and competitions against other firms in the same industry. No one doubts this with physical property: General Motors is entitled to control 100% of its own output of mufflers, without handing any of them over to Ford or Toyota or Volkswagen. Permitting a producer the full return on its investment in mufflers (or any other product) is essential to promote investment in productive assets and rivalry with other producers. Just so with knowledge, an increasingly vital input into production. Why should IDX or any other maker of intellectual property be placed under legal rules that effectively entitle its rivals to a chunk of that asset's value?

No Wisconsin decision of which we are aware requires temporal or geographic limits as a condition to the enforcement of a non-disclosure agreement for intellectual property. It is impossible to understand how a non-disclosure agreement could place "geographical" limits on the dissemination of intellectual property comparable to those restricting the locale where a salesman may try to drum up customers for a new employer. If the Foundation were forbidden to disclose the details to Epic in Wisconsin, but allowed to do so in Indiana, that would be the same thing as permitting disclosure everywhere (and thus nixing all contractual limits) — for Epic could sell worldwide any software derived from what it learned in Indiana. Knowledge does not respect borders. ProCD, a case that the district court did not mention, enforced as a matter of Wisconsin law a contract placing worldwide restrictions on the use of intellectual property embedded in software.

Temporal limitations could make more sense. Perhaps Wisconsin's courts would deem contracts such as those between IDX and the Foundation to cover only information that is not generally known. What would be the point of forbidding the Foundation to talk in public about features of IDX's system that had been the subject of a review in a trade publication? But it is too early in this litigation to decide whether Wisconsin would curb the unqualified scope of this contractual language — and, if some limits would be interpolated into the text, whether these would shelter the actual disclosures that the Foundation made to Epic. (The Foundation contends, for example, that it had customized IDX's software extensively and disclosed to Epic details about its own additions rather than any information "furnished by" IDX and of which Epic was unaware before the disclosures. It also contends that IDX cannot establish damages. But these and other factual arguments must be developed through discovery rather than decided at the complaint stage.)

Because further proceedings must be held on IDX's contractual claims against the Foundation and its employees, we must consider the contention that Epic tortiously induced the Foundation to break its promises (and that Quade and Rosencrance were in cahoots with Epic in this endeavor). Because this claim was dismissed on the pleadings, we assume that Epic did what IDX alleges. The district court held that Epic was free to induce the Foundation to dishonor its promises in light of § 134.90(6)(a), which "displaces conflicting tort law, restitutionary law and any other law of this state providing a civil remedy for misappropriation of a trade secret." That ruling suffers from two problems.

First, paragraph (a) begins with the words "[e]xcept as provided in par. (b)", and paragraph (b)2 carves out of paragraph (a) "[a]ny civil remedy not based upon misappropriation of a trade secret." The tort of inducing breach of a non-disclosure contract (the sort of contract independently protected by paragraph (b)1) is "not based upon misappropriation of a trade secret." It is based on interference with the contract.

Second, even when read apart from paragraph (b), paragraph (a) deals only with " conflicting tort law" (emphasis added). Enforcement of a non-disclosure agreement does not conflict with trade-secret law, and thus preventing third parties from inducing breach of such an agreement does not conflict with trade-secret law. The district court did not cite any Wisconsin case holding that the tort of interference with contract (or interference with economic advantage more generally) conflicts with trade-secret law for purposes of § 134.90(6)(a). In diversity litigation, our task is to implement state law as state courts would implement it, and we cannot think of any reason why the Supreme Court of Wisconsin would find such a conflict. The claims against Epic, Quade, and Rosencrance growing out of the Foundation's breach of its contracts with IDX therefore should not have been dismissed.

The judgment of the district court is affirmed to the extent it granted judgment to the defendants on IDX's trade-secret claims. The remainder of the judgment is reversed, and the case is remanded for further proceedings consistent with this opinion.