In May 1998, Outsource International, Inc. ("Outsource," "OSI," or the "EMPLOYER") filed a temporary restraining order (a "TRO") and preliminary injunction against former OSI employee George Barton and Barton's Staffing Solutions, Inc. ("BSSI") (collectively referred to as the "defendants") based upon Barton's alleged violations of a confidentiality clause and a non-compete clause in the Employment Agreement between Barton and OSI. The district court granted the TRO, and on June 12, 1998, after a two-day evidentiary hearing, the district court entered a modified preliminary injunction against Barton and BSSI. Barton and BSSI appeal from the entry of the modified preliminary injunction order. We affirm.

OSI provides temporary industrial staffing and employment consulting services to industrial customers located throughout the United States, including the Chicago suburban area. OSI has been a prominent fixture in the temporary staffing industry for many years and has developed a strong reputation for its quality and dependable services.

Like other businesses in the temporary staffing industry, OSI's product is temporary workers. OSI attempts to develop and market its product by keeping extensive computerized records on its workers. These records include information such as each worker's previous work environments, pay rates, billing rates, and worker compensation rates. OSI also attempts to provide superior service to its customers by providing more qualified workers, which in turn, makes its product more reliable throughout the temporary industrial labor staffing industry. OSI puts considerable time into developing its employee files and its customer relations and, therefore, attempts to protect this information from outside competitors. It also requires that its staffing consultants enter into certain restrictive covenants, such as a non-compete clause and a confidential information clause. These restrictive covenants are embodied in each consultant's Employment Agreement with the company.

In 1992, Barton became a labor staffing consultant at L.M. Investors, Inc. ("LM"). In 1993, Barton signed an Employment Agreement with LM. The Employment Agreement contained confidentiality and non-compete clauses as conditions of his employment. The agreement also provided that these clauses would remain in effect for one year after the termination of his employment with OSI. Specifically, Barton's Employment Agreement contained the following non- compete clause and confidentiality agreement:

[D]uring the term of the Agreement and for a period of one (1) year immediately following the termination of EMPLOYEE's employment, for any cause whatsoever, so long as EMPLOYER continues to carry on the same business, said EMPLOYEE shall not, for any reason whatsoever, directly or indirectly, for himself or on behalf of, or in conjunction with, any other person, persons, company, partnership, corporation or business entity:

(i) Call upon, divert, influence or solicit or attempt to call upon, divert, influence or solicit any customer or customers of EMPLOYER;

(ii) Divulge the names and addresses or any information concerning any customer of EMPLOYER;

(iii) Disclose any information or knowledge relating to EMPLOYER, including but not limited to, EMPLOYER's system or method of conducting business to any person, persons, firms, corporations or other entities unaffiliated with EMPLOYER, for any reason or purpose whatsoever;

(iv) Own, manage, operate, control, be employed by, participate in or be connected in any manner with the ownership, management, operation or control of the same, similar or related line of business as that carried on by EMPLOYER within a radius of twenty-five (25) miles from EMPLOYEE's home office or within a radius equivalent to EMPLOYEE's defined territory, whichever is greater.

By its terms, the Employment Agreement could be enforced only through injunctive relief. By signing the Employment Agreement, Barton agreed to waive his right to a jury trial if a dispute should arise and he agreed that he would be liable to pay all costs and expenses of the action, including attorney fees.

From 1993 until April 7, 1998, Barton was the exclusive staffing consultant for LM in his territory. In February 1998, OSI acquired LM. On April 7, 1998, Barton resigned from OSI. At the time of his resignation, he was the staffing consultant for four of OSI's Illinois offices: Aurora East, Aurora West, Joliet, and University Park. Barton's home base was the Aurora East office.

Immediately after Barton resigned from OSI, he opened BSSI, a temporary industrial labor staffing company, in West Chicago, Illinois. BSSI's office is approximately 12 miles from OSI's Aurora East office. To staff BSSI, Barton hired former OSI employees that had worked with him while he was at OSI. Within weeks after starting his business, Barton and BSSI had acquired twelve former OSI customers that Barton had serviced while he was employed at OSI.

OSI filed an action against Barton and BSSI seeking immediate temporary and preliminary injunctive relief based on Barton's breach of the non-compete and confidentiality clauses in his Employment Agreement. After a two-day evidentiary hearing on the matter, the district court found that OSI had stated a prima facie case for a preliminary injunction.

On appeal, the defendants admit that Barton violated the restrictive covenant in his Employment Agreement; they argue, however, that the restrictions are unenforceable and, therefore, that the district court erred in entering the preliminary injunction order. They also challenge the scope of the injunction.

Under Illinois law, a preliminary injunction will be granted if the plaintiff can show that: (1) he possesses a clear right or interest that needs protection; (2) an inadequate remedy at law exists; (3) irreparable harm will result if it is not granted; and (4) there is a reasonable likelihood of success on the merits. Office Mates 5, North Shore, Inc. v. Hazen, 599 N.E.2d 1072, 1079 (Ill. App. Ct. 1992) (citations omitted). On appeal, a district court decision preserving the status quo may be overturned only upon a clear showing of an abuse of discretion. VMS Ltd. Partnership Sec. Litig. v. Prudential Sec. Inc., 103 F.3d 1317, 1323 (7th Cir. 1996). A review of the district court's decision requires a re- examination of the court's findings on each of the four relevant considerations. However, the likelihood of success on the merits often serves as a threshold requirement for entitlement to preliminary relief. Reinder Bros., Inc. v. Rain Bird E. Sales Corp., 627 F.2d 44, 49 (7th Cir. 1980). "Nonetheless, in applying the abuse of discretion standard, we do not give equal deference to every aspect of a court's decision. The abuse of discretion standard is used to evaluate the . . . court's application of the facts to the appropriate legal standard, and the factual findings and legal conclusions underlying such decisions are evaluated under the clearly erroneous and de novo standards, respectfully." VMS Ltd. Partnership, 103 F.3d at 1323.

The basic test applied by Illinois courts in determining the enforceability of restrictive covenants is "whether the terms of the agreement are reasonable and necessary to protect a legitimate business interest of the employer." Office Mates 5, 599 N.E.2d at 1080 (citations omitted). This determination "necessarily turns on the facts and circumstances of each case." Id. Illinois courts long have recognized two situations in which an employer has a legitimate business interest to justify enforcement of a covenant not to compete: (1) where the customer relationships are near-permanent and but for the employee's association with the employer the employee would not have had contact with the customers; and (2) where the former employee acquired trade secrets or other confidential information through his employment and subsequently tried to use it for his own benefit. Lawrence and Allen, Inc. v. Cambridge Human Resource Group, Inc., 685 N.E.2d 434, 443 (Ill. App. Ct. 1997); Springfield Rare Coin Galleries, Inc. v. Mileham, 620 N.E.2d 479, 485, 488 (Ill. App. Ct. 1993); Office Mates 5, 599 N.E.2d at 1080. Here, the district court applied both tests and determined that they were alternative grounds that supported its decision to grant a preliminary injunction. We will affirm the district court's decision if we find that either of the two alternative grounds was sufficient to enforce the covenant not to compete.

Illinois courts have developed two tests for determining whether an employer has a near- permanent relationship with its customers or clients. Some courts analyze the near-permanent relationship according to the seven objective factors set forth in Agrimerica, Inc. v. Mathes, 557 N.E.2d 357 (Ill. App. Ct. 1990), while others follow the "nature of the business" test as espoused in Lawrence and Allen or Office Mates 5. In the immediate case, the district court applied the nature of the business test to determine whether OSI had a near-permanent relationship with its customers. Accordingly, we will use the same test in reviewing the district court's analysis and in making our own determination of whether a near-permanent relationship existed. See Springfield Rare Coin Galleries, 620 N.E.2d at 489 (holding that the seven Agrimerica factors need not be applied in every case).

Generally, "the near-permanency test turns in large degree on the nature of the business involved, and certain businesses are just more amenable to success under it." Lawrence and Allen, 685 N.E.2d at 444 (internal citation omitted). For example, "a near-permanent relationship with clients is inherent in the provision of professional services." Id. (citing Springfield Rare Coin Galleries, 620 N.E.2d at 488). On the other hand, a business engaged in ordinary sales generally will not enjoy near- permanent relationships with its customers. Lawrence and Allen, 685 N.E.2d at 444 (citing Springfield Rare Coin Galleries, 620 N.E.2d at 488). Moreover, "a near-permanent relationship is generally absent where the nature of the plaintiff's business does not engender customer loyalty by providing a unique product or personal service and customers utilize many suppliers simultaneously to meet their needs." Id. at 444 (citing Office Mates 5, 599 N.E.2d at 1081-82). The court in Office Mates 5 noted:

The fact pattern typifying these latter cases is the existence of a highly-competitive industry in which customers, through cross-purchasing, satisfy their buying needs. Plaintiffs in these businesses rely heavily on their sales force, who utilize basic sales techniques such as cold calls to make sales. The identity of customers or clients are known by all in the industry.

Office Mates 5, 599 N.E.2d at 1082.

In the present case, the district court considered the nature of OSI's business (the industrial staffing industry) and determined that the customer loyalty OSI enjoyed and the unique product OSI offered in the industry (as opposed to a general sales product) created a near- permanent relationship with its customers. Specifically, the district court stated:

[In this case,] I believe that the existence of multiple suppliers to a single user is not an indication that any one supplier is as good as another. I believe it is because from the user's perspective the user simply does not wish to be in a position of seeking temporary industrial workers and not having them when and where the user wants them.

And I think that this is a key finding because the courts, as I read the prior cases, deal with prior fact situations in terms of--and I paraphrase--"There are many suppliers; ergo, there is nothing unique or special [about the product or service]."

I do not believe that inference can be drawn here and I do not draw it. It is, in fact, I think difficult to consider that relationships are not near[-]permanent, not only for the above reasons, but for all of the facts which show that the need for reliability is paramount in this particular industry; and common sense would lead one to the conclusion that if any suppliers of temporary industrial workers ha[ve] shown to any buyer of these services that their services are reliable . . . [then] the user will continue to go back to that supplier absent some extraordinary incentives in terms of prices.

. . .

So I believe that the near[-]permanent relationships do exist. I find that they do exist in this business.

Trans. of Preliminary Inj. Proceedings, at 239- 40. This finding is supported by record evidence. The record reflects that OSI (and LM, its predecessor company) enjoyed a brand name recognition throughout the industry which gave it a certain level of dependability and prominence. The record also reflects that OSI had strong customer loyalty and that it set itself apart from other staffing businesses in the industry through an elaborate employee screening and customer service system. Thus, the district court did not abuse its discretion in finding that a near-permanent relationship existed between OSI and its customers.

We also must briefly consider the second part of the near-permanence test--we must determine whether "but for" Barton's association with OSI, he would not have had contact with the customers that he obtained from OSI after he left its employ. Lawrence and Allen, 685 N.E.2d at 443. In concluding that the "but for" element of the near-permanence test was met, the district court held that but for OSI's good name and substantial resources, Barton could not have "sold" his customers on his industrial staffing services. Specifically, the court held that:

Outsource added substantial value to the product that Barton was selling in several ways, one of which was for potential customers to agree to meet with Barton. . . . He needed Outsource's resources, not to consummate sales, but to get to the initial contact stage of his deals with companies in need of temporary services.

This factual finding is supported by record evidence as well. Barton testified that he acquired all twelve of BSSI's customers by telephoning the primary contacts he had developed while he was working for OSI. Thus, the district court did not abuse its discretion in finding that but for Barton's association with OSI, he would not have had contact with the customers.

Based on the evidence in the record, the district court's findings regarding the nature of OSI's business and its ability to meet the near- permanency test were within the court's discretion.

The defendants also contend that the district court erred in finding that the confidential information test served as a valid basis for enforcing the restrictive covenant. "A restrictive covenant may be enforced if the employee learned trade secrets or other confidential information while in [the] plaintiff's employ and subsequently attempted to use it for his or her own benefit." Springfield Rare Coin Galleries, Inc. v. Mileham, 620 N.E.2d 479, 485 (Ill. App. Ct. 1993). Here, in making its determination, the district court stated:

The value of the confidential information I think is shown by the speed in which Mr. Barton acquired the business of former [OSI] customers. He may have extraordinary power as a salesperson, but it is doubtful to me that extraordinary power could have--in fact, I find that it is really quite incredible that even the most powerful sales person could have acquired that business with the speed with which he acquired it if he were not assisted at least by cost and price figures and such data as the comp rates.

. . .

[B]ecause he spent so long with Outsource . . . it is impossible to say that any of these clients with which he's dealing are Barton's and Barton's alone.

The record shows that OSI put considerable effort into developing a workforce and keeping data on the workforce secret. Furthermore, OSI maintained classified records on its customers, to which Barton had access. Shortly after Barton left OSI and opened BSSI, most of his staff consisted of former OSI employees. All of BSSI's clients were OSI clients when Barton was employed by OSI. Thus, the district court did not abuse its discretion in determining that Barton took confidential information while in OSI's employ and used it for his own benefit.

The defendants also contend that the geographic and activity restrictions in the restrictive covenant must be modified if we determine that the covenant is enforceable. While we agree with the defendants' general premise that restrictive covenants should be narrowly tailored so as only to protect a legitimate business interest of the employer, see Lawrence and Allen, 685 N.E.2d at 441, in the immediate case, the defendants' argument is underdeveloped and unsupported by law and, therefore, it is waived. See United States v. Brocksmith, 991 F.2d 1363, 1366 (7th Cir. 1993) ("The premise of our adversarial system is that appellate courts do not sit as self-directed boards of legal inquiry and research, but essentially as arbiters of legal questions presented and argued by the parties before them.").

The district court did not abuse its discretion in entering the preliminary injunction order or in finding that the restrictive covenant was enforceable. We Affirm.

Posner, Chief Judge, dissenting.

I regret my inability to agree with the court's disposition of the case, because it is the right disposition from the standpoint of substantive justice. Mr. Barton is an adult of sound mind who made an unequivocal promise, for which he was doubtless adequately compensated, not to compete with his employer within 25 miles for a year after he ceased being employed. He quit of his own volition--quit in fact to set up in competition with his employer. And all the customers whom he obtained for his new company, before the preliminary injunction which the court affirms today put him temporarily out of business, were customers of his former employer. So he broke his contract. But Illinois law, to which we must of course bow in this diversity suit, is hostile to covenants not to compete found in employment contracts. An Illinois court would not enforce this covenant.

There is no longer any good reason for such hostility, though it is nothing either new or limited to Illinois. The English common law called such covenants "restraints of trade" and refused to enforce them unless they were adjudged "reasonable" in time and geographical scope. Mitchel v. Reynolds, 1 P. Wms. 181, 24 Eng. Rep. 347 (K.B. 1711); Horner v. Graves, 7 Bing. 735, 743, 131 Eng. Rep. 284, 287 (C.P. 1831); E. Allan Farnsworth, Contracts sec. 5.3 (3d ed. 1999). The original rationale had nothing to do with restraint of trade in its modern, antitrust sense. It was paternalism in a culture of poverty, restricted employment, and an exiguous social safety net. The fear behind it was that workers would be tricked into agreeing to covenants that would, if enforced, propel them into destitution. Mitchel v. Reynolds, supra, 1 P. Wms. at 193-94, 24 Eng. Rep. at 351; Arthur Murray Dance Studios of Cleveland, Inc. v. Witter, 105 N.E.2d 685, 704 (Ohio C.P. 1952); Harlan M. Blake, "Employee Agreements Not to Compete," 73 Harv. L. Rev. 625, 632-37 (1960). This fear, though it continues to be cited, e.g., Weaver v. Ritchie, 478 S.E.2d 363, 368 (W. Va. 1996), has no basis in current American conditions.

Later, however, the focus of concern shifted to whether a covenant not to compete might have anticompetitive consequences, since the covenant would eliminate the covenantor as a potential competitor of the covenantee within the area covered by, and during the term of, the covenant. E.g., Russell v. Jim Russell Supply, Inc., 558 N.E.2d 115, 123 (Ill. App. 1990); JAK Productions, Inc. v. Wiza, 986 F.2d 1080, 1085-86 (7th Cir. 1993) (interpreting Indiana law); Wilson v. Gamble, 177 So. 363, 365-66 (Miss. 1937); Wells v. Wells, 400 N.E.2d 1317, 1319 (Mass. App. 1980). This concern never had much basis, as recognized in Consultants & Designers, Inc. v. Butler Service Group, Inc., 720 F.2d 1553, 1562-64 (11th Cir. 1983), especially when the covenant was found in an employment contract. It would be unlikely for the vitality of competition to depend on the ability of a former employee to compete with his former employer. So unlikely that it would make little sense to place a cloud of suspicion over such covenants, rather than considering competitive effects on a case by case basis.

At the same time that the concerns behind judicial hostility to covenants not to compete have waned, recognition of their social value has grown. The clearest case for such a covenant is where the employee's work gives him access to the employer's trade secrets. The employer could include in the employment contract a clause forbidding the employee to take any of the employer's trade secrets with him when he left the employment, 765 ILCS 1065/8(b)(1); Abbott- Interfast Corp. v. Harkabus, 619 N.E.2d 1337, 1344 (Ill. App. 1993), as in fact the employer did in this case. Such clauses are difficult to enforce, however, as it is often difficult to determine whether the former employee is using his former employer's trade secrets or using either ideas of his own invention or ideas that are in the public domain. Nalco Chemical Co. v. Hydro Technologies, Inc., 984 F.2d 801, 804 (7th Cir. 1993); Eden Hannon & Co. v. Sumitomo Trust & Banking Co., 914 F.2d 556, 561-62 (4th Cir. 1990). A covenant not to compete is much easier to enforce, id. at 562, and to the extent enforced prevents the employee, during the time and within the geographical scope of the covenant, from using his former employer's trade secrets.

A related function of such a covenant is to protect the employer's investment in the employee's "human capital," or earning capacity. Curtis 1000, Inc. v. Suess, 24 F.3d 941, 947 (7th Cir. 1994); Reddy v. Community Health Foundation, 298 S.E.2d 906, 917 (W. Va. 1982); Paul H. Rubin & Peter Shedd, "Human Capital and Covenants Not to Compete," 10 J. Legal Stud. 93, 96-97 (1981). The employer may give the employee training that the employee could use to compete against the employer. If covenants not to compete are forbidden, the employer will pay a lower wage, in effect charging the employee for the training. There is no reason why the law should prefer this method of protecting the employer's investment to a covenant not to compete.

I can see no reason in today's America for judicial hostility to covenants not to compete. It is possible to imagine situations in which the device might be abused, Watson v. Waffle House, Inc., 324 S.E.2d 175, 177 (Ga. 1985); Farnsworth, supra, sec. 5.3, p. 334, but the doctrines of fraud, duress, and unconscionability are available to deal with such situations. Hilb, Rogal & Hamilton Agency of Dayton, Inc. v. Reynolds, 610 N.E.2d 1102, 1107 (Ohio App. 1992). A covenant's reasonableness in terms of duration and geographical scope is merely a consideration bearing on such defenses ("in a great many instances, [particular covenants not to compete] can be of no use to the obligee; which holds in all cases of a general restraint throughout England; for what does it signify to a tradesman in London, what another does at Newcastle?", Mitchel v. Reynolds, supra, 1 P. Wms. at 190-91, 124 Eng. Rep. at 350). Had Barton signed a covenant in which he agreed that if he ever left the employ of Outsource he would never again work in the business of providing temporary industrial labor anywhere in the world, there would be at least a suspicion that he had been forced or tricked into signing the covenant and therefore that it should not be enforced. There is no suggestion of that here, and so if I were writing on a clean slate I would agree wholeheartedly with the district court's granting a preliminary injunction against Barton's violating the covenant.

But the Illinois courts approach covenants not to compete in a different way, not radically different perhaps but different enough to require a reversal in this case. Their view is that a covenant not to compete that is contained in an employment contract is enforceable in only two circumstances--either where the covenant protects a "near permanent" relationship between the former employer and his customers, or where it protects "confidential information" (that is, trade secrets) of the former employer. Lawrence & Allen, Inc. v. Cambridge Human Resource Group, Inc., 685 N.E.2d 434, 443 (Ill. App. 1997); Curtis 1000, Inc. v. Suess, supra, 24 F.3d at 947 (applying Illinois law); A.J. Dralle, Inc. v. Air Technologies, Inc., 627 N.E.2d 690, 696-97 (Ill. App. 1994); Springfield Rare Coin Galleries, Inc. v. Mileham, 620 N.E.2d 479, 485 (Ill. App. 1993). The latter circumstance is straightforward; as I noted earlier, a covenant not to compete is easier to enforce than a covenant that forbids the former employee to use confidential information.

The reason given for protecting "near permanent relationships" between the former employer and its customers is that if the customer was locked into its relationship with the former employer, the former employee could not have enticed the customer away without using skills or contacts that he had obtained in the course of his employment. E.g., Lawrence & Allen, Inc. v. Cambridge Human Resource Group, Inc., supra, 685 N.E.2d at 444; Audio Properties, Inc. v. Kovach, 655 N.E.2d 1034, 1039 (Ill. App. 1995); Springfield Rare Coin Galleries, Inc. v. Mileham, supra, 620 N.E.2d at 488. This is welcome recognition that a covenant not to compete can be a proper method of protecting an employer's investment in the employee's human capital. The significance of "near" permanent may be to make clear that if the employer and the customer have an actual contract which the employee induces the customer to break, and thus a relationship that is "permanent" for the duration of the contract, the covenant not to compete would be academic; the employee would be liable in tort for inducing the breach of a contract. HPI Health Care Services, Inc. v. Mt. Vernon Hospital, Inc., 545 N.E.2d 672, 676 (Ill. 1989); Lama Holding Co. v. Smith Barney Inc., 668 N.E.2d 1370, 1375 (N.Y. 1996). Without either a contract with the customer or a covenant with the employee, the employee's action in "stealing" the customer would be privileged unless the employee used independently tortious means, such as a violation of fiduciary duty, to accomplish the "theft." Speakers of Sport, Inc. v. ProServ, Inc., 178 F.3d 862, 865-66 (7th Cir. 1999).

The Illinois courts appear to place the burden of proving that the covenant meets one of the two criteria of validity on the employer. Lawrence & Allen, Inc. v. Cambridge Human Resource Group, Inc., supra, 685 N.E.2d at 444; Audio Properties, Inc. v. Kovach, supra, 211 Ill. Dec. 651, 655 N.E.2d at 1037; Springfield Rare Coin Galleries, Inc. v. Mileham, supra, 620 N.E.2d at 485. In effect Illinois requires the employer to prove that the covenant not to compete serves a social purpose. Such a requirement is inconsistent with the idea of freedom of contract, which animates contract law and a corollary of which is that courts do not limit the enforcement of contracts to those the social point of which the court can see. They enforce a contract unless there is some reason to think it imposes heavy costs on third parties, offends the moral code, fails to comply with formal requirements (such as those imposed on some contracts by the statute of frauds), or doesn't embody an actual deal between competent consenting adults.

Still, we must take the Illinois law as we find it, and apply it as best we can to the facts of the case. Burns Philp Food, Inc. v. Cavalea Continental Freight, Inc., 135 F.3d 526, 529 (7th Cir. 1998); Nautilus Ins. Co. v. Zamora, 114 F.3d 536, 538 (5th Cir. 1997). Barton was employed by Outsource as a salesman, soliciting orders for temporary industrial workers that Outsource would supply. These are skilled and semi-skilled factory workers-- packers, assemblers, fork-lift operators, and the like. They are employed by Outsource and in effect "rented" to industrial firms as temporary workers. Barton had been in the business for many years before his employment by Outsource. Deciding to go out on his own, he quit Outsource and quickly obtained business from a dozen customers of Outsource with whom he had dealt. There is no question that he violated the covenant not to compete in his employment contract, which barred him for one year after his employment ended from competing with Outsource in the Chicago area. But there is no evidence that he stole any of Outsource's trade secrets. Like the customer lists in A.J. Dralle, Inc. v. Air Technologies, Inc., supra, 627 N.E.2d at 697; Springfield Rare Coin Galleries, Inc. v. Mileham, supra, 620 N.E.2d at 485, and Curtis 1000, Inc. v. Suess, supra, 24 F.3d at 947-48, Outsource's customer list, which Barton may have used to get customers for himself, was not secret. Nor is it likely that Barton relied on the list; these were people he had been dealing with for years. The wages that Outsource pays its workers are not secret either. Barton did not take the list of workers on Outsource's roster, but obtained workers for his customers in the same way that Outsource does, by radio and newspaper advertisements. Many of these workers had been working for Outsource, but that is no surprise; competing local suppliers of temporary labor hire from the same pool, and temporary workers often register with multiple agencies.

Outsource screens the workers whom it hires and sorts them into different job categories so that it knows whom to dispatch when it receives an order from one of its customers. This information--the list of workers screened for reliability and the jobs that they can do--is a genuine trade secret, and so Outsource would have a strong case if Barton had taken this information with him when he quit and set up on his own. But there is no evidence that he did this. Regardless of that, if Outsource does the screening and sorting function better than other suppliers of temporary industrial labor, it may have established "near permanent" relationships with its customers, but of this there is no evidence either. The only users of temp labor who testified at the preliminary injunction hearing agreed that such users have no sense of loyalty to particular suppliers. Both witnesses used multiple agencies. It was feasible for Barton to use standard selling techniques, rather than any techniques that he had learned from Outsource or information that he took with him when he left Outsource, to get customers for his new business.

The district judge said that Barton "gets credibility in terms of the reliability issue from the testing, screening and transport services that were provided when he was at Outsource." I don't get this. The customers whom Barton obtained when he left Outsource knew that he was no longer with Outsource and that they were dealing not with Outsource but with a start- up. If they trusted him, it was for his personal qualities.

The cases in which Illinois courts (or federal courts applying Illinois law in diversity cases) uphold covenants not to compete found in employment contracts are cases in which the former employer was supplying a specialized, complex product or (more usually) service, often a professional service such as veterinary care. See, e.g., Prairie Eye Center, Ltd. v. Butler, 713 N.E.2d 610, 614 (Ill. App. 1999); Office Mates 5, North Shore, Inc. v. Hazen, 599 N.E.2d 1072, 1081-82 (Ill. App. 1992); Lawrence & Allen, Inc. v. Cambridge Human Resource Group, Inc., supra, 685 N.E.2d at 444; Springfield Rare Coin Galleries, Inc. v. Mileham, supra, 620 N.E.2d at 488 ("a near-permanent relationship with clients is inherent in the provision of professional services"). Those are settings in which the former employee is likely to be using information or training acquired from his former employer. Enforcement is denied when the former employer is supplying a fungible or standardized product or service. As we explained in Curtis 1000, Inc. v. Suess, supra, 24 F.3d at 948 (citations omitted), "the Illinois cases distinguish between sellers of services, especially professional services such as accounting and consulting, and sellers of ordinary goods. In the former class, where the quality of the seller's service is difficult to determine by simple inspection, customers come to repose trust in a particular seller, and that trust is a valuable business asset, created by years of careful management, that the employee is not allowed to take away with him. In the latter class, involving the sale of goods, the element of trust is attenuated, particularly where . . . the good is a simple and common one sold under competitive conditions. In these cases Illinois law does not permit the seller to claim a protectable interest in his relations with his customers. . . . For here current price and quality, rather than a past investment in meeting customers' needs, are the decisive factors in the continued success of the firm, and they of course are not appropriated by the departing employee."

Human beings are not fungible; skilled and semi-skilled workers differ along such dimensions as reliability, experience, know-how, and wage demands. But Barton, who had worked in the business of supplying temporary industrial labor for many years before going to work for Outsource, knew all this. He did not know the specifics about the workers whom Outsource screened and sorted, but he did not take those specifics with him when he left, either. To repeat, as far as this record shows, all he used in signing up customers for his new venture were the standard sales techniques used in this business.

Since the irreparable harm to Barton from the grant of a preliminary injunction to Outsource exceeds the irreparable harm that Outsource would experience from the denial of the injunction (as a start-up, Barton would find it difficult to prove damages from being frozen out of business for a year as a result of the enforcement of the covenant), Outsource must prove not just that it has a better case than Barton but that it has a much better case. Id. at 945; Ty, Inc. v. GMA Accessories, Inc., 132 F.3d 1167, 1172 (7th Cir. 1997); Steakhouse, Inc. v. City of Raleigh, 166 F.3d 634, 637 (4th Cir. 1999). It has a worse case. As a detail, I note that the court assumes that state rather than federal law governs the standard for the grant or denial of a preliminary injunction. Not so; it is federal law. Southern Milk Sales, Inc. v. Martin, 924 F.2d 98, 101-02 (6th Cir. 1991); Ferrero v. Associated Materials Inc., 923 F.2d 1441, 1448 (11th Cir. 1991); Equifax Services, Inc. v. Hitz, 905 F.2d 1355, 1361 (10th Cir. 1990). See generally Mayer v. Gary Partners & Co., 29 F.3d 330 (7th Cir. 1994), and cases cited there.

There is another issue, which the court avoids (through an improper means, as I shall point out), though the issue merits consideration because it comes up a lot and has divided the judges of the district court from which this case comes. Applied Micro, Inc. v. SJI Fulfillment, Inc., 941 F. Supp. 750, 755 (N.D. Ill. 1996); Murry v. Sheahan, 991 F. Supp. 1052, 1054 n. 3 (N.D. Ill. 1998). It is whether a federal court in a diversity case involving an issue to which the state supreme court has not spoken should consider itself bound by the decisional law of the intermediate state appellate court for the region in which the federal district court is located, even though other regional appellate courts of the state disagree, or instead should try to predict what the state supreme court would do if the conflict were brought before it for resolution. If the former approach is the right one, this might strengthen the case for Barton, who contends that the law of the Illinois Appellate Court for the Second Appellate District, the region in which this suit would have been maintained had it been brought in state court, is particularly hostile to covenants not to compete. I can't really see this; the second and fourth districts prefer what they call a "nature of the business" test, Lawrence & Allen, Inc. v. Cambridge Human Resource Group, Inc., supra, 685 N.E.2d at 443-44 (second district); Springfield Rare Coin Galleries, Inc. v. Mileham, supra, 620 N.E.2d at 488-89 (fourth), and the first prefers a seven-factor test, e.g., Audio Properties, Inc. v. Kovach, supra, 655 N.E.2d at 1037, but whether they differ substantively or are just different ways of making the same point (which I would prefer to see stated in terms of the protection of the employer's trade secrets or of his investment in the employee's human capital) is unclear, as implied by both the first and the third districts, which treat the tests as interchangeable. Midwest Television, Inc. v. Oloffson, 699 N.E.2d 230, 232-34 (Ill. App. 1998) (third); Office Mates 5, North Shore, Inc. v. Hazen, supra, 599 N.E.2d at 1082-83 (Ill. App. 1992) (first). This case is in any event too one- sided to make the difference in formulations outcome-determining, but the issue of whether the federal courts should be bound by the decisions of the intermediate state appellate court to which the case would have been appealed had it been litigated in a state trial court is a recurrent one and we might as well lay it to rest and put the contending district judges out of their misery. Rather than do that, or, alternatively, duck the issue as not necessary to decide in this case, my colleagues on this panel merely defer to the district judge's choice between the rival tests of the validity, under Illinois law, of covenants not to compete. But we were told in Salve Regina College v. Russell, 499 U.S. 225 (1991), that we are not to defer to district courts' interpretations of state law.

The argument for the rule urged by Barton that the federal court should be bound by regional precedent is that forum shopping is minimized if the parties know that the law applicable on appeal will be the same whether the suit is litigated in federal court or in state court. Suppose that Barton were suable only in the Second District. If Outsource sued him in state court there, the decisional law of that district would control, while if it sued him in the federal district court, that law would not control if the federal court is not constrained to follow the Second District when there is an interdistrict conflict. This is a good point, but not a decisive one. For suppose that under the Illinois venue rules, see 735 ILCS 5/2-101 et seq., Outsource could sue Barton in state court in either district. Then under Barton's view if Outsource wanted to be in the First Appellate District it would file suit in the federal district court in Chicago (that is, in the eastern division of the Northern District of Illinois), and if it wanted to be in the Second Appellate District it would file suit in DuPage County (in the western division). So there is forum-shopping potential either way.

The argument against Barton's view is that if the case were litigated in state court, the loser, if he lost because of the regional appellate court's position in the conflict between regional appellate courts, could ask the state supreme court to resolve the conflict in his favor, and so if the suit is litigated in federal court he should be allowed to ask the federal appellate court to stand in the state supreme court's shoes and resolve the conflict. Stated differently, an intermediate state appellate court decision is likely to be rather shaky when it is in conflict with the decisions of its sister courts. But this point isn't compelling either, since Illinois permits this court to certify a question to the Supreme Court of Illinois. ILCS S. Ct. R. 20(a); Todd v. Societe BIC, S.A., 9 F.3d 1216, 1221-22 (7th Cir. 1993) (en banc). If a district court felt bound by an intermediate appellate decision that was in conflict with decisions from another appellate district, the losing party could appeal to this court and ask us to certify the issue to the Illinois supreme court, which presumably would be more likely to agree to take the case if there was indeed a conflict at the state intermediate appellate level.

In circumstances of such dubiety, we can do no better than to fall back on the familiar rule, clearly articulated in Commissioner v. Estate of Bosch, 387 U.S. 456, 465 (1967), that decisions by intermediate state appellate courts are not authoritative in federal court. See also Charles Alan Wright, Law of Federal Courts sec. 58, p. 395 (5th ed. 1994). The federal court is to imagine itself the state supreme court rather than an intermediate court of the state. Allen v. Transamerica Ins. Co., 128 F.3d 462, 466-67 (7th Cir. 1997); Wood v. Allstate Ins. Co., 21 F.3d 741, 743-44 (7th Cir. 1994); First National Bank v. Trans Terra Corp. Int'l, 142 F.3d 802, 806 (5th Cir. 1998); 19 Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, Federal Practice and Procedure sec. 4507, pp. 157-161 (2d ed. 1996). We should adhere to that view, since, as I have shown, there is no compelling practical reason to depart from it merely because there is a conflict at the intermediate level.