A number of telephone companies formerly owned by the American Telephone and Telegraph Company, and still known as the Bell operating companies, have petitioned us under 28 U.S.C. Sec. 2342(1) (which vests the courts of appeals with exclusive jurisdiction to review certain orders issued by the Federal Communications Commission, see 47 U.S.C. Sec. 402(a), including orders promulgating rules, see Columbia Broadcasting System v. United States, 316 U.S. 407, 419, 62 S. Ct. 1194, 1201, 86 L. Ed. 1563 (1942)) to set aside an order made by the FCC at the conclusion of a notice-and-comment rulemaking proceeding. The order promulgated a rule that forbids the divested Bell operating companies after July 1, 1984, to sell or lease telecommunications equipment to their customers, or to sell certain specialized services to them, other than through separate subsidiary corporations created for this purpose. FCC, Furnishing of Customer Premises Equipment, Enhanced Services and Cellular Communications Services by the Bell Operating Companies, 49 Fed.Reg. 1190 (Jan. 10, 1984). The Commission's power to adopt such a rule as an incident to its authority to regulate the interstate telecommunications industry under Title II of the Communications Act of 1934, as amended, 47 U.S.C. Secs. 201-224; see 47 U.S.C. Sec. 154(i), is not questioned. Nor is it questioned that the standard that should guide our review is whether the rule is arbitrary or capricious. See Administrative Procedure Act, 5 U.S.C. Secs. 553, 706(2)(A); FCC v. National Citizens Comm. for Broadcasting, 436 U.S. 775, 802-03, 98 S. Ct. 2096, 2115-16, 56 L. Ed. 2d 697 (1978).

The rule has a long history behind it (on which see generally Brock, The Telecommunications Industry: The Dynamics of Market Structure (1981)) without which it would be thoroughly incomprehensible. At one time AT & T completely dominated the provision of telephone handsets and other terminal equipment used by customers of its regional subsidiaries--Illinois Bell, New England Telephone and Telegraph Company, and the other, and today free-standing, Bell operating companies. If you lived in one of the areas served by a Bell operating company (and more than 80 percent of the nation's telephone subscribers did and still do) and you wanted telephone service, you had to get it from the operating company. The company would run a line from your premises to a central office containing switching equipment. Other lines owned by the company connected up all the central offices in the city or other locality, while AT & T's Long Lines Division provided intercity ("long distance") service. AT & T's Western Electric subsidiary manufactured the telephone handsets (as well as the other transmission, switching, and terminal equipment used in or attached to the telecommunications network) and sold them to the operating companies which in turn leased them to the subscribers as part of basic telephone service. The tariffs under which telephone service was sold forbade the subscriber to attach anything to the line that was not provided by the operating company ("foreign attachments"), or to service the telephone himself--the operating company took care of that too, and would replace the phone if it broke down, wore out, or became obsolete. Thus the Bell System manufactured, owned, installed, maintained, and replaced the telephones and other terminal equipment used by its subscribers. Rates for telephone service were of course regulated but Western's prices to the operating companies were not--not directly, anyway.

This thoroughgoing integration of equipment and service was praised as a source of cost savings and guarantee of service quality but it was also, and as time went by increasingly, criticized. At first it was criticized on two grounds: that it might enable AT & T to transfer profits from a regulated to an unregulated activity by having Western Electric charge the operating companies an inflated price for the equipment they bought from Western (some state regulatory authorities responded to this danger by regulating Western's prices to the operating companies, see, e.g., Illinois Bell Tel. Co. v. Illinois Commerce Comm'n, 55 Ill. 2d 461, 483-84, 303 N.E.2d 364, 376 (1973)); and that it might enable AT & T to exclude Western's competitors from the enormous equipment market represented by the Bell operating companies. Later it was also criticized because the Bell System was thought to be slow in introducing new kinds of terminal equipment made possible by advances in communications and computer technology and in otherwise responding to customer needs. These criticisms began to persuade the FCC and the Justice Department's Antitrust Division in the late 1960's. The FCC rescinded AT & T's "foreign attachments" tariff in 1968. Use of Carterfone Device in Message Toll Telephone Service, 13 F.C.C.2d 420, on reconsideration, 14 F.C.C.2d 571 (1968). AT & T responded by requiring subscribers who wanted to attach their own terminal equipment to lease from it an expensive device linking the equipment to the telephone line, called a "protective connecting arrangement," which many believed was an unnecessary hindrance to foreign attachments. Eventually the Commission rescinded this tariff as well. See Proposals of New or Revised Classes of Interstate and Foreign Message Toll Telephone Service, 56 F.C.C.2d 593 (1975) (First Report ), 58 F.C.C.2d 736 (1976) (Second Report ) aff'd under the name of North Carolina Utilities Comm'n v. FCC, 552 F.2d 1036 (5th Cir.), modified, 64 F.C.C.2d 1059 (1977); for additional detail on the history of AT & T's efforts to discourage foreign attachments see Litton Systems, Inc. v. AT & T, 700 F.2d 785, 790-98 (2d Cir.1983), and Brock, supra, ch. 9.

Working parallel to a Commission increasingly bent on prying open the telephone industry to competition, the Justice Department in 1974 brought the most ambitious antitrust suit in history, seeking nothing less than the dismemberment of the world's largest corporation, as AT & T then was. While the Department thus sought to change the structure of the industry by breaking up its principal firm, the FCC--wanting to protect the new competitors that it had midwifed by rescinding various exclusionary Bell tariffs--imposed increasingly stringent restrictions on AT & T's ability to compete. Among these was a requirement that AT & T establish a separate subsidiary to handle the marketing of terminal equipment to Bell System subscribers. See Second Computer Inquiry, Final Decision, 77 F.C.C.2d 384, 398, on reconsideration, 84 F.C.C.2d 50 (1980), 88 F.C.C.2d 512 (1981), aff'd under the name of Computer & Communications Industry Ass'n v. FCC, 693 F.2d 198 (D.C.Cir.1982). The subsidiary was to be separate not merely in the arid formal sense of having a separate corporate charter but in the practical sense of having a completely different staff and different business premises from AT & T and the operating companies. (The requirements for a separate subsidiary, as they will be applicable to the divested companies under the order reviewed here, are set forth at 47 C.F.R. Secs. 64.702(c), (d).)

Before the new subsidiary could be formed, AT & T, after a year of trial of the Justice Department's antitrust suit, agreed in 1982 to an elaborate consent decree settling the suit. United States v. AT & T, 552 F. Supp. 131 (D.D.C.1982), aff'd without opinion under the name of Maryland v. United States, 460 U.S. 1001, 103 S. Ct. 1240, 75 L. Ed. 2d 472 (1983). The decree allowed AT & T to keep Western Electric and also its interest in the two Bell operating companies (Cincinnati Bell Telephone Company and Southern New England Telephone Company) in which it was a minority shareholder, but required it to spin off the other 22 Bell operating companies. These were to remain separate corporations but were also to be grouped into seven geographically contiguous systems each owned by a holding company known as a Regional Bell Operating Company. The customer-equipment part of the decree (discussed at 552 F. Supp. 189-93) transferred the ownership of all terminal equipment owned by the operating companies to AT & T and forbade the operating companies to manufacture equipment. But it did not forbid them to reenter the terminal-equipment market by selling or leasing to their subscribers or others equipment that the operating company could buy from Western Electric (still owned by AT & T, but no longer affiliated with the operating companies) or any other manufacturer, as it chose. The decree also forbade the operating companies to provide "information services," that is, computer processing of data transmitted over the telecommunications network. But an operating company can get a waiver of this prohibition by persuading the court that "there is no substantial possibility" that it will be able to "use its monopoly power to impede competition." Id. at 195. Several requests for waiver are now pending before the district judge who approved the decree. He considered requiring the operating companies to provide terminal equipment if at all only through separate subsidiaries, but in the end decided this was a matter best left to the judgment of the FCC and other regulatory and legislative bodies.

The Commission on March 29, 1983, took up the invitation and issued a notice of proposed rulemaking, and after considering comments from 31 interested parties issued on January 10 of this year the order that the operating companies have asked us to set aside. The order requires each of the seven regional holding companies to form a separate subsidiary if it wants to sell or lease "customer premises" equipment (telephone handsets and other terminal equipment), "cellular services" (a telephone service for automobiles), or "enhanced services" (described in due course). We shall describe the coverage of the order as "customer equipment," though this term is imprecise as applied to "enhanced services." The order is not challenged insofar as it relates to cellular services.

The operating companies attack the order on several grounds. The first is that the Commission has changed regulatory direction without explanation. In Second Computer Inquiry, supra --the decision that before the divestiture of the Bell operating companies required AT & T to form a separate subsidiary to market equipment to Bell System subscribers--the Commission had declined to impose the separate-subsidiary requirement on any of the non-Bell operating companies, the "independent" telephone companies as they are called. The largest independent is General Telephone and Electronics, a vertically integrated firm like AT & T before the divestiture but much smaller (serving fewer than 10 percent of the nation's telephone subscribers). At first the Commission had decided to require GTE to form a separate subsidiary but it later changed its mind. See 84 F.C.C.2d at 72-73. It explained that it had decided to cut off the requirement of a separate subsidiary just below AT & T upon consideration of four things: nationwide control of bottleneck facilities (the telephone lines, without which customer equipment is useless); incentive to use earnings in one market to subsidize sales in another ("cross-subsidization"); degree of vertical integration; and cost of operating through a separate subsidiary. The first factor in particular distinguished AT & T from the rest of the industry. AT & T was the only telephone company operating nationwide, not only in the sense that the Bell operating companies as a group served most of the nation's telephone subscribers but also in the sense that AT & T, through its control of long-distance service (which it provided to the independent telephone companies as well as to the Bell companies), fused the entire nation into a unified network. The control of long-distance service also made AT & T more highly integrated than any independent, including GTE. And of course AT & T had much greater resources than any independent for creating and operating a separate subsidiary to market customer equipment.

The operating companies argue that the final decision in Second Computer Inquiry clearly implied that if they were divested they (or regional groupings of them) would not have to form separate subsidiaries. Neither the operating companies nor even the regional entities would have nationwide control of bottleneck facilities and they would in fact be less integrated than GTE and on average no larger. Since GTE had been spared having to form a separate subsidiary they should be spared too. They concede as they must that the Commission is allowed to change its mind, see, e.g., American Trucking Ass'ns, Inc. v. Atchison, Topeka & Santa Fe Ry., 387 U.S. 397, 416 (1967), but argue that it must give a good reason for doing so, and did not. As a matter of fact, the Commission did not acknowledge that it had changed its mind.

If the Commission had decided in Second Computer Inquiry that the Bell operating companies, should they be divested, would not be required to furnish customer equipment through separate subsidiaries, then it would be obliged by principles recently reaffirmed in Motor Vehicle Mfrs. Ass'n v. State Farm Mutual Automobile Ins. Co., 463 U.S. 29, ----, 103 S. Ct. 2856, 2866, 77 L. Ed. 2d 443 (1983)--the case invalidating the rescission of an agency rule requiring automobile manufacturers to install airbags--to justify its about face in the present proceeding. An unexplained reversal of policy is arbitrary. But that is not what happened in the present case. At the time of the Second Computer Inquiry the Bell operating companies had not been divested; nor was it clear they would be. The timing and terms of the settlement came as a surprise to almost everyone but the negotiators. The Commission did not allude in its decision to the possibility of divestiture, let alone decide what it would do if the possibility materialized. When it said it would require only AT & T to set up a separate subsidiary for marketing customer equipment it did not mean AT & T and not its Bell operating company subsidiaries; it meant AT & T and not GTE and the other independents.

What is true is that the analysis in Second Computer Inquiry provides the operating companies with arguments against requiring the regional Bell operating companies to form separate subsidiaries. Unlike the pre-divestiture AT & T, they do not (considered individually) have nationwide control of bottleneck facilities, and they are not so highly integrated as AT & T was (and is) or even as GTE is. But this just means that the Commission's decision in the present case is not entirely consistent with a previous decision in a related matter. Although "patently inconsistent application of agency standards to similar situations lacks rationality and is arbitrary," Contractors Transport Corp. v. United States, 537 F.2d 1160, 1162 (4th Cir.1976), there is no principle that enjoins perfect consistency. Perfection is not an attainable goal of government. A sequence of judicial decisions will not exhibit perfect consistency from case to case; no more should a sequence of administrative decisions be required to do so.

When a court (or administrative agency) decides a case, it does so by applying a rule or standard, such as the four-factor test of Second Computer Inquiry. Because a rule or standard is by definition worded at least somewhat generally (a standard more generally than a rule), it necessarily embraces--at least in the sense that its words describe--other cases besides the case before the court or agency, including cases that have never arisen and that because of the limitations of human foresight the court or agency may not be thinking about at all and at any rate is not thinking about clearly when deciding the case before it. When the unforeseen case arises it may lead the court or agency to modify the previous standard, often without acknowledging that it is doing so. That is what happened here. The Commission, describing its previous decision, said, "we did not mean to imply that nationwide control was a prerequisite to employing separate structure as a condition on other carriers' offer of CPE or enhanced services." 49 Fed.Reg. at 1196. But that is certainly what it had seemed to imply. However, the inconsistency that arises from inability to foresee all possible applications of the general language in an opinion is not equivalent to an agency's changing its mind in the same case--first deciding that airbags should be required in cars, then deciding they should not be. The burden of explanation is greater when the agency changes its mind in the same case than when, in the process of deciding a new case, having new facts, it departs from the standard it used to decide a previous case. The conceptual untidiness that results when an agency, taking one problem at a time, implicitly modifies a substantive standard as it goes along has never been an automatic ground for reversing an administrative decision. We must therefore consider the Commission's order in this case on its own terms.

The basis of the order is a concern with what the Bell operating companies, even though they not only have been divested by AT & T but also have been deprived of the customer equipment they had owned before they were divested, still might be able to do to interfere with regulation or competition or both if allowed to compete in the equipment market with no additional restrictions. The operating companies retain control of access to the tele-communications network. It is still the case, no less than before the divestiture, that if you want to have telephone service and are located in an area served by a Bell operating company you must get your access line from that company. Hence the operating companies have monopolies of basic telephone service in many important markets and continue to be regulated both by state regulatory commissions and by the FCC. If the companies were allowed to condition access to the network on the subscriber's purchasing his terminal equipment from them, they might be able to charge an exorbitant price for the equipment and so defeat regulation of their telephone rates. See Indianapolis Airport Authority v. American Airlines, Inc., 733 F.2d 1262, 1268-69 (7th Cir.1984); 1 Kahn, Economics of Regulation 28 (1970). The charge would be nominally for the equipment but really for the service. Of course they are no longer allowed to condition access to their lines on the customer's agreeing to buy his telephone from them and if they did so they would be violating not only the divestiture decree but the antitrust laws, and thus could be sued for treble damages as well as being punished for contempt. See Jefferson Parish Hospital District No. 2 v. Hyde, --- U.S. ----, ----, 104 S. Ct. 1551, 1560, 80 L. Ed. 2d 2 (1984); Otter Tail Power Co. v. United States, 410 U.S. 366, 377, 93 S. Ct. 1022, 1029, 35 L. Ed. 2d 359 (1973); Northern Pacific Ry. v. United States, 356 U.S. 1, 78 S. Ct. 514, 2 L. Ed. 2d 545 (1958); International Business Machines Corp. v. United States, 298 U.S. 131, 56 S. Ct. 701, 80 L. Ed. 1085 (1936); Litton Systems, Inc. v. AT & T, supra. But since they retain their privileged access to the customer for equipment (because he is their customer for the line), conceivably they might disparage competitors' equipment or might make it difficult for that equipment to work properly on their lines.

Another concern is that the operating companies might sell terminal equipment below cost to weaken or destroy their competitors, hoping to recoup these short-term losses right away by jacking up their rates for transmission and switching. Of course below-cost selling with intent to weaken or destroy competitors ("predatory pricing") violates the antitrust laws, see, e.g., Barry Wright Corp. v. ITT Grinnell Corp., 724 F.2d 227, 230-36 (1st Cir.1983), and of course rate increases to telephone subscribers require regulatory approval. But as it is frequently difficult to allocate costs between such closely related activities as providing the telephone line and providing the telephone, the companies might be able to reclassify costs, in ways difficult to detect, from the equipment to the line part of the service, and by thus making the equipment look cheaper than it was and the line more costly than it was might hinder antitrust and regulatory efforts to prevent the underpricing of the equipment and the compensating overpricing of the line. See, e.g., Lavey & Carlton, Economic Goals and Remedies of the AT & T Modified Final Judgment, 71 Geo.L.J. 1497, 1508-10 (1983).

It might seem that the company would have to choose between the two abuses we have described--that it could not charge below-cost prices in the equipment market in order to expand its share of that market and, at the same time, charge supracompetitive prices for equipment (thus reducing its market share, since price and output are inversely correlated) as a way of placing profits made in its monopoly markets beyond the reach of the regulators of those markets. But, conceivably if improbably, the two forms of abuse could be practiced together. By shifting costs from the equipment market to the service market at no cost to itself (because the telephone ratepayer would absorb these costs), the operating company could make money in the equipment market so long as it was selling the equipment at a price higher than the costs that it had not shifted to the service market. Yet that price might be below the true economic cost of the equipment. The company might be able to increase its equipment profits further by overpersuading the customers for its regulated service to become its equipment customers as well. Any profits it made in the equipment market would be enabled by its monopoly of basic telephone service but would be outside the effective authority of the regulators.

Admittedly these dangers are less acute in the case of the operating companies than in the case of the pre-divestiture AT & T. AT & T had relatively little (though rapidly growing) competition in the equipment market, and the weakness of that competition made it more attractive than it would otherwise have been to pursue a predatory strategy. The divested Bell operating companies, in reentering the equipment market, are competing with powerful firms--not the least of which is AT & T itself. It seems highly unlikely that they could carry to fruition a predatory strategy that to succeed would have to do in AT & T. Also, the hugeness of AT & T before the divestiture gave rise to widespread concern that neither the state commissions nor the FCC could regulate AT & T or its subsidiaries effectively. And there was some, though not great, evidence in the government's antitrust case against AT & T--evidence of which the FCC was well aware--that it really had pursued a predatory strategy in the equipment as in other markets. See United States v. AT & T, supra, 552 F.Supp. at 160-63, 174; United States v. AT & T, 524 F. Supp. 1336, 1349-50, 1380 (D.D.C.1981). The Second Circuit recently upheld a large damages judgment against AT & T for predatory behavior in that market. See Litton Systems, Inc. v. AT & T, supra, 700 F.2d at 814-16.

But we cannot say that the danger that the Bell operating companies will, unless restricted, engage in predatory conduct, speculative though that danger is, is too slight to have moved a rational Commission to action. The divestiture occurred six months ago and probably it will be years before turnover of personnel will create a different business culture in the operating companies from the one they shared with AT & T. What the parent did, the children can, for a time at least, be expected to continue to do--or so at least the Commission could rationally fear. Remember, too, that the divestiture decree grouped these companies in regional systems each of which is a major enterprise having billions of dollars in sales and assets, operating in several states, and enjoying as we have said monopoly control of access to the telephone network for all customers in its service area. Although they face a most formidable competitor in the equipment market--AT & T, which took over from them all the installed customer-premises equipment at the date of divestiture--they have the advantage even over AT & T of being the only firms in the equipment market that control their potential customers' access to the telecommunications network. The basic source of AT & T's monopoly power was not the manufacture of telecommunications equipment or even the ownership of the nation's long-distance lines; it was the operating companies' control of access to the telecommunications network. The inheritors of the Bell monopoly are the operating companies rather than AT & T. This is why the Commission's emphasis in Second Computer Inquiry on nationwide control of bottleneck facilities was misplaced. The worst bottlenecks in this industry are local.

Just how acute the danger posed by the Bell operating companies in their post-divestiture form is to the successful implementation of regulatory and antitrust policies is a matter of judgment; and naturally we defer to the views on this question of the FCC, which has struggled with the problems of telephone regulation for many decades, and of the Justice Department, which has been active in recent years in exploring the antitrust problems of the regulated industries in general and the telecommunications industry in particular and which supports the Commission's order fully. If despite all they said the danger seemed to us phantasmal we would have to set aside the Commission's order. But the order makes a plausible though not overwhelming case for concern with the regulatory and antitrust problems that might arise if the divested Bell companies could participate in the equipment market without special restraint. The technically complex nature of the regulated activity makes us especially reluctant to substitute our guesses for the Commission's.

Of course it is not enough that the agency has been able to identify a problem; the solution must be reasonably appropriate to it. As the FCC admits, forcing the operating companies to create separate subsidiaries does not eliminate the incentive to engage in the kind of regulatory and antitrust evasion that we have described, since the subsidiaries' profits accrue to the parents. But no one is arguing to us that the FCC did not go far enough (except an intervenor who, as we shall see, lacks standing to argue so). And it is not as if the separate-subsidiary device cannot make any contribution to solving the problem. The danger that the operating companies might use their monopoly of access to the telecommunications network to subvert competition in the equipment market is heightened by the difficulty of allocating costs between these sectors. The FCC once tried using accounting rules to ensure a proper allocation, but it makes a persuasive showing that this approach did not work. The separate-subsidiary device solves the problem of cost allocation by forcing the companies physically to separate the capital and human assets used in the service and equipment markets respectively. The separate subsidiary cannot share staff or facilities with the operating company; it must have its own staff and facilities. Anything it needs from the operating company it must pay for at the tariffed rate--that is, the rate that any stranger would pay. We cannot describe this as an inapt or irrational solution.

But its efficacious features are also the source of its costs; and a solution that costs much more than any reasonable estimate of its benefits is, as the Commission acknowledges, unreasonable and therefore arbitrary. By forbidding the operating companies to use existing staff and facilities to provide equipment to their subscribers, the separate-subsidiary device eliminates the economies (if any) of joint provision of service and equipment; and it could be argued that the potential for such economies is the major reason for allowing the operating companies to reenter the equipment market in the first place. Although the sacrifice of these economies is the principal cost of the Commission's approach, neither the Commission nor any party made an effort to estimate them. The complete absence of quantitative data gives rather a hollow ring to the Commission's claim to have based its order on a comparison of the "costs" and "benefits" of the separate-subsidiary approach. It is no doubt true that data on the expected benefits to consumers of making it harder for the operating companies to avoid regulatory and antitrust constraints would be unobtainable in even a remotely reliable form and must therefore be left largely to the Commission's judgment. But some data on the economies of joint provision should be obtainable; this kind of question (not this exact question, though) has been studied extensively by economists. See, e.g., Meyer, et al., The Economics of Competition in the Telecommunications Industry (1980); Eldor & Sudit, Alternative Specifications of Returns to Scale and Joint Estimation of Factor Demand and Production Functions in Telecommunications, 18 Rev.Bus. & Econ.Research 15 (1982); Evans & Heckman, Multiproduct Cost Function Estimates and Natural Monopoly Tests for the Bell System, in Breaking Up Bell: Essays on Industrial Organization and Regulation 253 (Evans ed. 1983); Raj & Vinod, Bell System Scale Economies Estimated from a Random Coefficients Model, 34 J.Econ. & Bus. 247 (1982). Even when the benefit side of the balance cannot be quantified, it is important to the formulation of rational policy to be able to measure the cost side. One may not know what preventing regulatory and antitrust abuse is worth but one would still like to know, before deciding whether prevention is worthwhile, whether it would cost $1 million or $1 billion.

But we do not think the Commission's order can be overturned because of the absence of cost data. The pace of the proceeding was dictated by the divestiture timetable rather than by the Commission. The courts did not approve the settlement agreement between AT & T and the Justice Department until 1983. The notice of proposed rulemaking in the present case was issued in March 1983 and it is not argued that the FCC dawdled in issuing it. With the divestiture scheduled to take place on January 1, 1984, the Commission had only a few months to obtain evidence. Naturally it had to rely heavily on the comments of the parties. The Bell operating companies have both greater resources than the Commission for conducting research and greater access to the data necessary for estimating the economies from providing telephone service and customer equipment jointly. The fact that they presented no estimates suggests to us that they doubted that such studies would come up with impressive numbers.

There are other indications that the economies of joint provision are not enormous. Customer-premises equipment usually is marketed and serviced by separate staffs from those that minister to the line itself; and the economies from housing the separate staffs in the same buildings cannot be very great. So the major claim of economies must relate to "enhanced services"--communication services that involve either changing the signal as it flows through the telephone network so that what emerges is different from what entered the network, or storing it so as to delay its arrival. (If the signal emerges in the same form as it went in, and without delay--as a telephone call does, for example--the service is "basic." "Basic service" is the heart of the operating companies' business.) Equipment for providing enhanced services is often sold as terminal equipment, but not always; sometimes it is installed in the telephone companies' central offices rather than on the subscribers' premises. The divested Bell operating companies will not be allowed to do this. Their separate subsidiaries will have to put the equipment somewhere else and lease lines from the operating company to connect the equipment to the local central offices. This follows from the fact that an operating company may not provide any service to the subsidiary other than on the normal tariffed basis and may not share premises with the subsidiary. The operating companies argue that small customers will be hurt, as only large ones will find it economical to buy or lease terminal equipment for enhanced services.

Besides a complete lack of evidence concerning the needs of small customers there are two problems with this argument. First, it is unclear to what extent the operating companies will be allowed to provide enhanced services, quite apart from anything the Commission may do. The divestiture decree as we have said forbids the operating companies to provide "information services," and they seem to be defined so broadly as to embrace most enhanced services. Although the parties agree that certain rather simple forms of enhanced service probably would not be classified as information services, it is not clear how big the permitted zone is; and the smaller it is, the less will be the social costs of the separate-subsidiary device. Second, a number of the operating companies have asked the divestiture court to waive the prohibition against providing information services. They must convince the court that there is no danger of monopolistic abuse, and have sought to sweeten their applications for waiver by proposing to provide information services through separate subsidiaries, just like the separate subsidiaries required by the order under review. This is not their first choice; but evidently they do not think the costs of the device so great that it would prevent them from competing effectively with providers not so burdened.

The operating companies also argue that even if the course that the Commission has taken with regard to their reentry into the customer-equipment market has a favorable benefit-cost ratio when the divested Bell operating companies are considered in isolation from the rest of the nation's telephone companies, it creates an irrational distinction between the divested companies on the one hand and on the other hand GTE and the two Bell operating companies (Cincinnati Bell and Southern New England Telephone) in which AT & T was allowed to retain stock. The Commission has not required GTE, the other independents, Cincinnati Bell, and SNET to form separate subsidiaries for the provision of customer equipment.

By far the most persuasive defense that the Commission makes of the difference in treatment is one of timing. The divestiture order presented it with an urgent need to decide whether and under what conditions the divested companies would be allowed to operate in the customer-equipment market. There was no similar urgency with regard to the independents and the retained Bell companies; the Commission could deal with them at its leisure. Since each telephone company has its own service area, which it monopolizes, there is unlikely to be much competition between telephone companies in the customer-equipment market; they sell to different groups of customers. Conceivably the regional Bell operating companies established by the divestiture decree might try to market customer equipment outside their service area or GTE might try to market customer equipment outside of its service area; but putting these possibilities to one side as wholly speculative on the record before us, we cannot see how the fact that the divested Bell operating companies are subject to more onerous restrictions than the independents and the AT & T-retained companies can place the divested companies under a competitive handicap. And insofar as they are merely complaining that the Commission should not have acted against them before acting against other similarly situated firms, they are complaining about an exercise of prosecutorial discretion or about agency inaction--areas where judicial review is extremely limited. See, e.g., Dunlop v. Bachowski, 421 U.S. 560, 572-73, 95 S. Ct. 1851, 1860, 44 L. Ed. 2d 377 (1975); Clayco Petroleum Corp. v. Occidental Petroleum Corp., 712 F.2d 404, 409 (9th Cir.1983) (per curiam); Rockford League of Women Voters v. Nuclear Regulatory Comm'n, 679 F.2d 1218, 1222-23 (7th Cir.1982); Sunstein, Deregulation and the Hard-Look Doctrine, 1983 Sup.Ct.Rev. 177, 201-02. The Commission still may order GTE and perhaps others to form separate subsidiaries if they want to continue marketing customer equipment. It just has not got around to that yet, and in the circumstances it did not have to.

The Commission did, however, seek to bolster its refusal to enlarge this proceeding to take in GTE and other telephone companies by showing that there were substantive differences between them and the divested Bell operating companies. This was also a way of showing that its approach was not inconsistent with Second Computer Inquiry, where it had decided not to force GTE to form a separate subsidiary. This part of the Commission's opinion is not fully persuasive. Although most of the independent companies, apart from GTE, are quite small firms that might find it very costly to establish separate subsidiaries to market customer equipment, GTE not only is large but has a history much like AT & T's of trying to hinder foreign attachments. See, e.g., Phonetele, Inc. v. AT & T, 664 F.2d 716, 725-26 (9th Cir.1981); see also International Telephone & Telegraph Corp. v. GTE, 518 F.2d 913 (9th Cir.1975). True, GTE is in a somewhat different position because unlike the Bell operating companies it still owns installed customer equipment, and to transfer all that equipment to a new subsidiary and put the staff that markets and maintains it in separate facilities could be quite costly, whereas all the customer equipment owned by the Bell operating companies was transferred to AT & T. But this point is easily overstated. The Commission's order allowed the divested companies to sell or lease customer equipment during the six-month period between the issuance of the order and the deadline (July 1, 1984) for forming the separate subsidiaries, and they have taken up this dispensation. At the end of this time they will have to transfer all the equipment that they have leased or otherwise own, plus the staff that markets and services this equipment, to the separate subsidiaries, and they will have to find separate premises for that staff and equipment. Presumably the cost of this transfer is no big deal, or the companies would either have deferred reentering the customer equipment business until they had to form the separate subsidiaries or have activated them before reentering.

Cincinnati Bell and SNET, like most of the independents, are rather small companies. They are big enough, surely, to be able to create separate subsidiaries without undue travail; and size of company has nothing to do with ability to abuse a monopoly position. But size does affect the number of customers that the company can hurt, and is therefore a good reason at least for postponing a final decision on whether to apply the separate-subsidiary requirement to the other companies. But we are greatly puzzled by the Commission's repeated references to the independents and the two retained Bell operating companies as serving mainly rural areas, as if the residents of such areas were less deserving of protection from monopoly abuses than city dwellers. The Commission tries to dress up this argument by pointing out that the demand for enhanced services is likely to be concentrated in urban areas. This is true; but because the divestiture decree forbids the Bell operating companies to provide information services (and hence most enhanced services), the separate-subsidiary requirement probably will have its smallest impact on enhanced services. But the fact that we can find some holes in the Commission's analysis does not make its order arbitrary, especially as we think the timing ground for not imposing the separate-subsidiary requirement on the other companies in this proceeding is convincing and would have led to the same result even if the Commission had not gotten tangled in what seems an irrelevant distinction between urban and rural areas. See Illinois v. ICC, 722 F.2d 1341, 1348-49 (7th Cir.1983), and references cited there.

Similarly, while we think the Commission should have owned up to the fact that the emphasis it had placed in Second Computer Inquiry on nationwide control of access to the telephone network and on extensive vertical integration as preconditions to engaging in cross-subsidization and transferring profits from regulated to unregulated markets was misplaced, as the Justice Department urged in its comments in this rulemaking proceeding, the Commission's lack of candor in acknowledging a departure from the earlier standard is not a serious enough blemish to warrant our setting aside its rule. We can see what the Commission was driving at; it took account of all relevant factors; its goal was reasonable; and the means chosen for attaining that goal were reasonable. That is enough to require us to sustain the Commission's order under the deferential standard that not only the Administrative Procedure Act but also the technical nature of the subject matter impose on us. See Motor Vehicle Mfrs. Ass'n v. State Farm Mutual Automobile Ins. Co., supra, 103 S.Ct. at 2865-67; Bowman Transportation, Inc. v. Arkansas-Best Freight System, 419 U.S. 281, 285-86, 95 S. Ct. 438, 441-42, 42 L. Ed. 2d 447 (1974); Illinois v. ICC, supra, 722 F.2d at 1348. The failure to notify the regulated firms of an impending change of course, which led us to reverse in Aero Mayflower Transit Co. v. ICC, 699 F.2d 938, 942 (7th Cir.1983), is not a problem here; the notice of proposed rulemaking gave the petitioners adequate warning.

The petitioners have moved to strike a brief filed by the North American Telecommunications Association, which intervened in this review proceeding. The Association, a trade association of manufacturers of customer equipment, fears competition by the Bell operating companies and thinks the Commission's order does not do enough to rein those companies in; it particularly dislikes a provision in the order which allows each company, for four years, to refer telephone subscribers who inquire about equipment to its own subsidiary. The Association was entitled to intervene in this proceeding because it has an interest in the outcome. See 28 U.S.C. Sec. 2348. But it was not entitled to intervene for the purpose it did, that is, to challenge rather than defend the order. If it wanted to challenge the order it had to file a petition for review, or a cross-petition within 14 days after the filing of the first petition for review; it could readily have done either, since it was a party to the proceeding before the agency. This is the rule for appeals from district courts, see Fed.R.App.P. 4(a)(3); Martin v. Hamil, 608 F.2d 725, 730-31 (7th Cir.1979), as well as for review of decisions of administrative agencies, see Fed.R.App.P. 1(a); Bath Iron Works Corp. v. White, 584 F.2d 569, 573 n. 2 (1st Cir.1978); 9 Moore's Federal Practice p 204:11 (2d ed. 1983). It is the orderly, as well as the prescribed, procedure. By filing its brief as an intervenor at the same time as the Commission and the other parties (various manufacturers of telecommunications equipment) who were defending the order, the Association deprived the Commission of a chance to respond, in its brief in response to the other challengers, to the Association's challenge.

The motion to strike NATA's brief is therefore granted, but the Commission's order of January 10 requiring the Bell operating companies to form separate subsidiaries if they wish to remain in the customer equipment business after July 1, 1984, is affirmed.