This case concerns the manner in which certain local telephone companies are reimbursed for providing their subscribers with access to long-distance common carriers. In the order under review, the Federal Communications Commission approved a modified system of reimbursement proposed by the National Exchange Carrier Association, the entity obligated under FCC regulations periodically to examine the formulas for compensation and to suggest revisions to those formulas.
Petitioners are twelve small “average schedule” companies, an appellation that we shall presently explain. They mount a two-pronged attack on the decision. First, they fault the Commission for failing to follow the notice-and-comment rulemaking procedures of the Administrative Procedure Act, 5 U.S.C. § 553 (1982). Second, they contend that approval of the reimbursement modifications proposed by NECA in the face of significant methodological flaws in NECA’s approach constituted arbitrary and capricious decisionmaking in violation of the APA, Id. § 706(2)(A). We decline to reach petitioners’ procedural objections for the simple reason that they were not raised before the Commission and thus cannot be advanced here. As to their second line of attack, however, we are constrained to conclude that the agency’s approval of NECA’s proposal was arbitrary and capricious. We therefore grant the petition for review.
I
The path that an interstate telephone call takes as it wends its way from one subscriber to a user at the other end of the line raises two regulatory issues of significance to this case. To state the obvious, a caller uses some of the same facilities to telephone the neighbor next door as are necessary to place a call across the continent. And thus the first issue: because interstate calls originate and terminate over local telephone company (or “exchange company” 1) facilities that also carry intrastate calls, the question arises as to how to apportion the costs of these exchange facilities (known as “local plant”) between intrastate and interstate jurisdictions. This apportionment is not of mere academic or internal bookkeeping interest but must be performed because the FCC enjoys jurisdiction over interstate rates, whereas the several States reign supreme over intrastate rates. See 47 U.S.C. §§ 151,152(b) (1982 & Supp. III 1985); Louisiana Public Service Commission v. FCC, 476 U.S. 355, 106 S.Ct. 1890, 1894, 90 L.Ed.2d 369 (1986); National Association of Regulatory Utility Commissioners v. FCC (NARUC), 737 F.2d 1095, 1104-05 (D.C.Cir.), cert. denied, 469 U.S. 1227, 105 S.Ct. 1224, 84 L.Ed.2d 364 (1984).2 Second, because transmission of an interstate call involves more than one carrier, revenues from subscribers must be divided among the carriers in a manner that allows them to recover the interstate costs of local plant.
To address the first issue, the FCC has devised the rather bureaucratic nomenclature of “jurisdictional separation” procedures. See MCI Telecommunications Corp. v. FCC, 750 F.2d 135, 137 (D.C.Cir. 1984); see also MCI Telecommunications Corp. v. FCC, 712 F.2d 517, 523 n. 4 (D.C. *94Cir.1983). To address the second, it has developed a “settlement process.” Because of their bearing on this case, each warrants what we shall try to limit to a modest description.
A
For purposes of “jurisdictional separation” procedures, the Commission distinguishes traffic sensitive costs from non-traffic sensitive costs. As their names suggest, these terms refer respectively to exchange company costs that vary with the extent of phone usage and those that do not. See NARUC, 737 F.2d at 1104.3 In general, the FCC has separated traffic sensitive costs of local plant attributable to the interstate jurisdiction on the basis of relative minutes of use. FCC Brief at 4-5.
Allocation of non-traffic sensitive (“NTS”) costs of local plant, in contrast, has presented a thornier problem. See MCI, 750 F.2d at 137. Beginning in 1970, the Commission adopted for NTS separations a method based on the “subscriber plant factor” (“SPF”). See id. at 137-38 & n. 2. See generally Prescription of Procedures for Separating & Allocating Plant Investment, 26 F.C.C.2d 247 (1970). Whatever its benefits, this approach had the “effect of assigning approximately 3.3 percent of the non-traffic sensitive costs of subscriber plant equipment to the interstate jurisdiction for every 1 percent of interstate calling.” MCI, 750 F.2d at 137 (citation omitted). This skewing of cost allocation permitted exchange companies to recover through interstate revenues NTS costs properly recoverable through intrastate rates. See id. at 138. As an interim remedial measure, the Commission in 1982 froze the percentage of NTS costs allocable to the interstate jurisdiction at 1981 average levels. See id. at 139, aff’g Amendment of Part 67 of the Commission’s Rules & Establishment of a Joint Board, 89 F.C.C.2d 1 (1982). In 1984, the Commission finally abandoned the SPF factor method entirely in favor of assigning to the interstate jurisdiction a flat 25% of the NTS costs of each exchange company. Amendment of Part 67 of the Commission’s Rules & Establishment of a Joint Board, 96 F.C.C.2d 781 (1984), reconsid. denied, FCC No. 85-56 (Jan. 30, 1986), pet’n for review pending sub nom. Rural Telephone Coalition v. FCC, No. 84-1110 (D.C.Cir. filed Mar. 22, 1984).
Recognizing that the shift from SPF to a fixed allocatur would work a “major change[],” id. at 802, the Commission decided to delay implementation until January 1, 1986, two years from the date of its decision. In addition, the FCC determined that no exchange company subject to the new separation procedures would lose more than 5% of its interstate allocation per year. See MTS & WATS Market Structure & Establishment of a Joint Board; Amendment, 50 Fed. Reg. 939, app. A (1985) (amending part 67 of the Commission’s rules, 47 C.F.R. pt. 67); see also 47 C.F.R. § 67.124(d)(7) (1986). By virtue of the fact that since 1971 exchange companies employing the SPF separations procedure had been able to assign no more than 85% of their NTS costs to the interstate jurisdiction, this decision meant that some companies would take as long as twelve years to reach the much-reduced 25% limit. See NECA Modification of Average Schedules, app. B (Sept. 16, 1985), J.A. at 63-65; see also Petitioners’ Brief at 8-9 & n. 18.
*95Now all this assumes the availability of accurate cost data, the basic ingredient needed for the FCC’s recipe. But this assumption is not borne out by reality, at least in respect of smaller telephone companies. Here is why. To determine interstate costs accurately, it is necessary to begin with reliable estimates of total intrastate and interstate costs, known collectively as “unseparated” costs. Most local telephone (or “exchange”) companies obtain data on their unseparated costs by periodically performing lengthy, expensive cost studies, thereby earning the name “cost companies.” However, the Commission traditionally has recognized that “[p]recise determination of a local company’s costs in all relevant areas may require extensive data collection, analysis, reporting and auditing, which can be a difficult and costly burden for small telephone companies.” NARUC, 737 F.2d at 1127. Accordingly, the FCC permits certain small exchange companies to estimate costs through use of an average schedule that “adopts generalized industry data to reflect the costs of a hypothetical exchange company.” Id.; see also MTS & WATS Market Structure: Average Schedule Companies, 103 F.C.C.2d 1017, 1019 (1986) [hereinafter Average Schedule Order], Joint Appendix (“J.A.”) at 1, 2.
This, then, is the genesis of the term “average schedule companies,” a group which consists, in essence, of local telephone companies that are typically smaller than the sometimes behemoth members of the “cost company” species. It is NECA’s efforts to adapt the changes in separations procedures that we have just discussed for application to these smaller, “average schedule” companies that form one aspect of the dispute now before us.
B
Before turning to the particulars of this dispute, however, we are well advised to dscribe — briefly—the settlement process by which revenues for interstate service are apportioned among participating carriers. Prior to the AT & T reorganization,4 exchange companies recovered their costs— both traffic sensitive and non-traffic sensitive — through charges imposed on interstate callers. NARUC, 737 F.2d at 1104-05. These charges were, in general, usage-sensitive, increasing with the distance and duration of calls. Id. AT & T was responsible for pooling these revenues and distributing them to exchange companies in proportion to their total (traffic sensitive and NTS) interstate investment. Id. at 1104-05 n. 5.
The Commission reexamined this traditional settlement process in light of technological advances and AT & T’s impending divestiture. It determined that recovery of interstate NTS costs through usage-sensitive charges required heavy interstate users to pay more than their fair share of interstate NTS costs and therefore impeded growth and development in the nationwide communications network.5 Id. at 1105-10.
The Commission’s reexamination led to the present system of “access charges.” Id. Under this new regime, exchange companies recover a large part of their interstate NTS costs through flat charges per access line 6 billed to end users.7 To enable exchange companies to calculate access charges, the Commission promulgated de*96tailed regulations that classify the cost components of providing interstate access into access elements and prescribe methods of calculating charges for each element. See generally 47 C.F.R. pt. 69 (1986). Those charges designed for recovery of the NTS costs of local plant that is used to carry interstate and intrastate messages are known as “common line” charges. See Access Charges Order, 93 F.C.C.2d at 268-69, 279-80.
In adopting access charge regulations, the Commission addressed the anomaly of traffic sensitive charges for NTS costs that was embedded in average schedule formulas. This anomaly resulted from the “Average Revenue Per Message” (“ARPM”) methodology then in effect, which tied average schedule company settlements, covering NTS as well as traffic sensitive costs, to the .number of interstate messages carried. See Average Schedule Order, 103 F.C.C.2d at 1019, 1023 n. 31, J.A. at 3, 8 n. 31. The Commission promulgated a rule requiring as follows:
Payments shall be made in accordance with a formula approved or modified by the Commission. Such formula shall be designed to produce disbursements to an average schedule company that simulate the disbursements that would be received ... by a company that is representative of average schedule companies.
47 C.F.R. § 69.606(a) (1986). This rule, in effect, required revision of average schedules to reflect the change in the way cost companies are reimbursed, which for NTS costs is no longer usage-sensitive.
Under section 69.606, AT & T was responsible for submitting average company schedules for 1984. Id. § 69.606 (b) (1984). Since AT & T’s role in the industry was to be drastically altered following divestiture, however, the Commission provided that after 1984, average schedules would be filed by an association of exchange companies. Id.8 Elsewhere the Commission prescribed the composition and duties of the association, now incarnated as the National Exchange Carrier Association. See id. § 69.-601-.603.9
C
The controversy before us arose when in September 1985 NECA filed for Commission approval proposed modifications to average schedules to take effect June 1,1986. The filing took the form of a forty-five page document in which NECA stated its determination that the usage-sensitive ARPM system for recovery of NTS costs should be abandoned.10 In its stead, *97NECA proposed to fix settlements for interstate NTS costs of local plant that carried interstate and intrastate calls (known as “common line settlements,” see supra note 9) at $8.06 per access line. See NECA Modification at 12, app. A, at 14, J.A. at 32, 57.11
NECA also outlined a three-part transition plan for implementation of its average schedule revisions. See id., app. A, at 12, J.A. at 55. First, the new formulas would apply immediately to exchange companies whose settlements would increase under the revision. Id. Second, under a so-called “flash cut” feature of the plan, companies that were recovering common line settlements that exceeded 85% of their unseparated (i.e., interstate and intrastate) common line revenue requirements would immediately be limited to 85%. This 85% cap, NECA claimed, was analogous to the 85% SPF limit that had been placed on cost companies in 1971. Id. at 18-19, J.A. 38-39. The third feature of the transition plan covered other companies whose settlements decreased because of the revisions; it provided that reductions would amount to no more than $1.25 per access line per month each year over a four-year period, resulting in a maximum reduction over four years of $5.00 per line per month. Id.12
D
Following the procedure it had employed the previous year for reviewing AT & T’s average schedule filing, the FCC released public notice of NECA’s proposal on October 9, 1985; it did not, however, publish this notice in the Federal Register. See FCC Mimeo 0189 (Oct. 9, 1985), J.A. at 67 (notice of NECA filing); see also FCC’s Brief at 40 n. 51. The notice described the filing simply as “revised formulas and supporting data that update the schedules that are associated with average schedule company interstate settlement disbursements.” See Notice, J.A. at 67. The Commission invited comments within 30 days, with reply comments due 30 days thereafter.13
Soon after notice was given, two average schedule companies asked the Commission to order NECA to release additional information on its proposal.14 They sought, in particular, the data underlying NECA’s composition of an “average” schedule company and further details on the calculations leading to NECA’s flat rate common line settlements figure. Before the Commission ruled on the production requests, NECA volunteered to produce additional material and on November 18, 1985 released some 900 pages of data.15 NECA prefaced this additional submission by admitting that “certain deliberative steps” were never reduced to writing and could not be documented; likewise, it explained that some “critical calculations” had not been preserved at the time they were performed and had only been reconstructed in response to the request for more data. See NECA Data Submission, notes 2, 4 (Nov. *9818, 1985), J.A. at 402. The Commission extended the comment period by 30 days— through December 6, 1985 for initial comments — to permit further examination of the released information. MTS & WATS Market Structure: Average Schedule Filing, CC No. 78-72, Mimeo No. 0757 (Nov. 7, 1985).
Dissatisfied with NECA’s further efforts, petitioners renewed their motion for production of data and sought further extension of the comment period.16 They complained generally that “[t]he data wholly fail[ed] to provide the information necessary to ascertain whether NECA’s proposed revisions have any reasonable relationship to average schedule costs or revenue requirements.” Id. at 3, J.A. at 85. To make matters worse, they complained, NECA’s data contained many obvious mathematical errors that, tb their mind, called into doubt the reasonableness of the results. Id. at 6-7, J.A. at 88-89. The Commission, however, was unmoved and denied the motion. MTS & WATS Market Structure: Average Schedule Filing, CC No. 78-72, FCC Mimeo No. 1319 (Dec. 6, 1985), J.A. at 111.
Petitioners’ comments detailed their criticisms, faulting NECA’s proposed revision for what it included as well as what it omitted. They noted several instances in NECA’s data where basic calculations had been incorrectly performed; in several, for example, NECA had multiplied non-zero numbers by a factor of zero to arrive at non-zero results.17 Petitioners objected, furthermore, to NECA’s omission of critical analytic steps in its derivation of the proposed flat rate figure for interstate common line settlements. Id. at 19-27, J.A. at 137-45. Adding insult to injury, in petitioners’ view, NECA had not even accurately informed average schedule compunies what, if any, reductions they would suffer under NECA’s revisions. According to petitioners, the “impact statements” furnished in July 1985 were flawed. See supra note 13. In an effort to correct what it admitted were “logical inconsistencies,” NECA had sent out a second set of impact statements in November 1985, but this second set, petitioners contended, was also inaccurate. Id. at 15-16, J.A. at 133-34; see also NECA Reply to Production Requests at 2, J.A. at 79. As a result, petitioners concluded, many companies could not meaningfully comment on the revisions.
Finally, petitioners attacked two aspects of NECA’s transition plan for implementing the revisions. First, petitioners asserted, in proposing a “flash cut” NECA had not justified the manner in which it translated the 85% limit on cost companies’ allocation of interstate NTS costs into a measurement that could be applied to average schedule companies. Id. at 27-28, J.A. at 145-46. Second, NECA’s proposal to limit reductions in settlements to $1.25 per access line per month over four years was, they charged, harshly discriminatory compared to the twelve-year period applicable to cost companies under the transition to a 25% fixed allocatur. Id. at 11-12, J.A. at 129-30.
NECA responded by submitting additional documents to correct the obvious errors that petitioners had discovered.18 In reply to charges that it had omitted important calculations and analytic steps from prior submissions, NECA chastised petitioners for holding the Association to an “unrealistic” standard of statistical precision.19 “[Ejxpert judgment,” NECA explained, necessarily played a significant role in formulating the modifications and was not susceptible to rigorous statistical support. *99Id. at 9-10, J.A. at 270-71. NECA defended its revised formulas as “what realistically could be accomplished in the time frame and with the resources available.” Id. at i, J.A. at 259.
NECA maintained, moreover, that its revisions represented a significant improvement over existing average schedules. For one thing, they were based on more recent data than existing schedules. See id. at 23-24, J.A. at 284-85. For another, the revisions took into account recent changes in separations procedures by classifying data according to the twenty-one categories of investment, expense, and income adjustments prescribed for cost companies in part 67 of the Commission’s rules. See id. at 14, 24, J.A. at 275, 285; see also 47 C.F.R. § 67.1(c) (1986). Finally, NECA asserted, the revisions reflected the logic of the system of access charges outlined in part 69 of the Commission’s Rules. See NECA Reply Comments at 24, 28, J.A. at 285, 289.
NECA also defended its transition plan. To translate the 85% SPF applicable to cost companies into a measurement applicable to average schedule companies, NECA had determined that “the average point at which an average schedule company’s common line settlement recovery exceeds 85% of its unseparated NTS costs is 15.9 messages per line.” Id. at 19, J.A. at 280 (footnote omitted). NECA dismissed petitioners’ charge that this equation was unjustified, contending that the two measurements were highly correlated. Id. NECA likewise waved off attacks on the four-year implementation of settlement reductions. This transition, NECA explained, was no harsher than that to which cost companies were subject, because average schedule formulas were based on cost company data and would necessarily incorporate the more gradual transition applied to cost companies. “Therefore, the transition of the average schedule companies ... is occurring on the same schedule as the cost companies.” Id. at 21, J.A. at 282.
E
The Commission approved NECA’s modifications to average schedule formulas in a decision released April 18, 1986.20 It acknowledged at the outset that “improvements are possible in the studies, methodologies, and data that NECA employed in developing new average schedules.” Id. at 1023 (footnote omitted), J.A. at 8. Nonetheless, the Commission noted, revisions were necessary so that the average schedules reflected changes in the separation procedures and settlement process applicable to cost companies. Id. at 1024, J.A. at 8. Thus, the pertinent inquiry in the Commission’s view was whether NECA’s revisions were an improvement over existing schedules. The FCC concluded that they were, for “reasons ... stated in the supporting comments.” Id. (footnote omitted). Its reasoning is set out in full:
[T]he NECA revisions: (1) more closely simulate the principles that are contained in Part 69 of our rules; (2) take more fully into account the effects of changes in jurisdictionally separated costs under Part 67 of our rules that have occurred since the last revision to the average schedules; (3) use more recent data; and (4) employ a residue ratio that more closely parallels the residual risk sharing that occurs under cost company pooling.
Id. (footnote omitted). Thus, the first three reasons mirror those presented by NECA. The fourth refers to the process by which revenues that remain in the interstate pool after initial distribution of expenses (i.e., the residue) are distributed to companies in proportion to their share of the risk of the overall enterprise. FCC Brief at 30 n. 38; see also 47 C.F.R. § 69.607-.610 (1986).21
*100The Commission addressed the transition plan with similar dispatch. The “flash cut” was reasonable, it found, because it affected only those companies that were receiving more than 100% of their total NTS revenue requirements from the interstate jurisdiction through average schedule settlements. Average Schedule Order, 103 F.C.C.2d at 1026, J.A. at 11. Those companies, in the Commission’s view, were in no position to complain about immediate reductions. More generally, any average schedule company that believed itself adversely affected by the revisions could elect to be treated as a cost company and thereby recover its full costs. “No company,” the Commission observed, “has a legal or equitable right to obtain more than its full costs.” Id. at 1027, J.A. at 12.
F
In the wake of the Commission’s decision, several parties sought relief or modification of the order in various respects.22 Three exchange companies sought waiver of the application of the average schedule revisions.23 The FCC denied two of the three requests and delayed ruling on the third pending further investigation.24 The Commission did decide, however, to grant limited relief to “small” average schedule companies.25 Specifically, it permitted average schedule companies with fewer than 5,000 local subscriber access lines to waive application of the revisions through July 31, 1986. In addition, small average schedule companies that elected by December 1, 1986 to become cost companies could choose to be compensated on a cost basis retroactively from August 1, 1986. Id. None of these subsequent rulings, it should be noted, is presently before us.
Shortly after the Commission's order on reconsideration, petitioners filed this petition for review. NECA, among others, has intervened.
II
Petitioners’ challenges to the Commission’s decision fall into two broad categories. First, they argue that the proceedings were procedurally defective when measured against the requirements prescribed by the APA for informal rulemaking. See 5 U.S.C. § 553. Second, they contend that the FCC’s decision was substantively flawed. In petitioners’ view, NECA’s description of the methodology underlying its proposal was so incomplete and the data supporting its revised formulas so riddled with errors and omissions that the Commission acted arbitrarily and capriciously in approving the revisions without addressing these shortcomings and without *101considering alternatives to NECA’s methodology outlined in comments submitted to the Commission.
A
Petitioners present a litany of objections to the manner in which the FCC conducted its proceedings. They fault the Commission because it did not publish notice of NECA’s filing in the Federal Register. See 5 U.S.C. § 553(b). In the notice it did provide, they complain, the Commission adequately described neither “the terms or substance” of the filing nor “the subjects and issues involved.” Id. § 553(b)(3). Notice of NECA’s filing was also deficient, according to petitioners, because the Commission did not disclose its views on the proposal or the alternatives it was considering. See, e.g., Home Box Office, Inc. v. FCC, 567 F.2d 9, 35 (D.C.Cir.), cert. denied, 434 U.S. 829, 98 S.Ct. 111, 54 L.Ed.2d 89 (1977). Finally, petitioners maintain that by failing to order NECA to produce more information on how it derived the average schedule revisions, the Commission ran afoul of the principle that an agency must disclose the data upon which a proposed rule is based. See, e.g., Connecticut Light & Power Co. v. NRC, 673 F.2d 525, 530-31 (D.C.Cir.), cert. denied, 459 U.S. 835, 103 S.Ct. 79, 74 L.Ed.2d 76 (1982).
We decline to rule on petitioners' procedural arguments, for they were not raised before the Commission. Our decision flows from section 405 of the Communications Act, which in relevant part provides:
The filing of a petition for reconsideration shall not be a condition precedent to judicial review of any ... order, decision, report, or action, except where the party seeking such review (1) was not a party to the proceedings resulting in such order, decision, report, or action, or (2) relies on questions of fact or law upon which the Commission ... has been afforded no opportunity to pass.
47 U.S.C. § 405 (1982). This court has construed section 405 to “codify the judicially-created doctrine of exhaustion of administrative remedies.” Washington Association for Television & Children v. FCC (WATCH), 712 F.2d 677, 681-82 (D.C.Cir. 1983) (interpreting § 405(2)); accord Office of Communication of the United Church of Christ v. FCC, 779 F.2d 702, 706-07 (D.C.Cir.1985) (interpreting § 405(1)); see also North Texas Media, Inc. v. FCC, 778 F.2d 28, 33-34 (D.C.Cir.1985). As incorporated in section 405, this doctrine “requirefs] complainants, before coming to court, to give the FCC a ‘fair opportunity’ to pass on a legal or factual argument.” WATCH, 712 F.2d at 681 (quoting Alianza Federal de Mercedes v. FCC, 539 F.2d 732, 739 (D.C.Cir.1976)).26 It is clear that in this case the Commission has been seised of no such opportunity.
Although we have recognized the salutary, commonsense notion that the exhaustion doctrine is to be applied flexibly, none of the traditional exceptions to the requirement avails petitioners. Their objections do not fall within that class of issues which, “by their nature could not have been raised before the agency.” WATCH, 712 F.2d at 682. On the contrary, we have previously deemed procedural objections premised on the APA to be precisely the sort appropriately raised before the Commission in the first instance. American Radio Relay League, Inc. v. FCC, 617 F.2d 875, 879 n. 8 (D.C.Cir.1980).
Nor do we find any evidence to suggest that it would have been “futile” for petitioners to lodge their procedural complaints in the agency proceedings, another basis for withholding application of the exhaustion requirement. See WATCH, 712 F.2d at 682 & n. 9. Nowhere in the administrative proceedings did the Commission hint that it was wedded to the procedures that it employed. Cf. Action for Children’s Television v. FCC, 564 F.2d 458, 469 (D.C. *102Cir.1977) (exhaustion not required where Commission’s general views on issue were already known). Nor has the Commission previously had the opportunity to decide the applicability of Section 553 to average schedule revisions.27 Cf. Great Falls Community TV Cable Co. v. FCC, 416 F.2d 238, 239-40 (9th Cir.1969) (exhaustion not required where issue had been specifically addressed in prior rulemaking). See generally 4 K. Davis, Administrative Law Treatise § 26.11 (2d ed. 1983) (discussing futility exception). Finally, this is not a situation in which another party to the proceedings voiced the objection now championed by petitioners. See WATCH, 712 F.2d at 682 & n. 10; see also L. Jaffe, Judicial Control of Administrative Action 457-58 (1965).28
Assuming arguendo our discretion to craft an exception to the exhaustion requirement to accommodate petitioners, see WATCH, 712 F.2d at 683, we would decline to do so. For one thing, petitioners themselves are at least partly responsible for the agency’s failure to address whether notice-and-comment rulemaking procedures were necessary, cf. id. at 683; their motions and comments emphasized NECA’s alleged failure to produce sufficient data rather than the FCC’s role in the proceedings. For another, the question of the applicability of notice-and-comment procedures of the APA does not yield an obvious answer. Respect for the proper court-agency relationship, see Alianza, 539 F.2d at 739, therefore counsels against permitting petitioners to bypass the agency with respect to this issue.
B
1
Before addressing the particulars of petitioners’ substantive challenge, we take an obligatory pause to set forth the legal standard governing our review. In a familiar passage, the APA directs us to “bold unlawful and set aside agency action, findings, and conclusions found to be arbitrary, capricious, an abuse of discretion, or otherwise not in accordance with law.” 5 U.S.C. § 706(2)(A). This statutory mandate clearly applies to the Commission’s decision, whether it constituted informal rulemaking or, as the Commission argues, merely implementation of a prior rule, namely section 69.606, 47 C.F.R. § 69.606. Under this “arbitrary-and-capricious” standard, courts are to conduct an inquiry that is “ ‘searching and careful,’ yet, in the last analysis, diffident and deferential.” Natural Resources Defense Council, Inc. v. SEC, 606 F.2d 1031, 1049 (D.C.Cir.1979) (McGowan, J.) (quoting Citizens to Preserve Overton Park, Inc. v. Volpe, 401 U.S. 402, 416, 91 S.Ct. 814, 823, 28 L.Ed.2d 136 (1971)) (footnote omitted).29
*103We have recently had occasion to observe the “wide range of reasons why agency action may be judicially branded as ‘arbitrary and capricious.’ ” FEC v. Rose, 806 F.2d 1081, 1088 (D.C.Cir.1986). Here, the focus of petitioners’ arguments is that the Commission failed to demonstrate that an adequate basis in the administrative record existed rationally to conclude that NECA’s revisions represented “an improvement over the existing schedules.”30 Average Schedule Order, 103 F.C.C.2d at 1024, J.A. at 8. To withstand this assault, the FCC must demonstrate a “rational connection between the facts found and the choice made.” Farmers Union Central Exchange, Inc. v. FERC, 734 F.2d 1486, 1499 (D.C.Cir.) (quoting Burlington Truck Lines v. United States, 371 U.S. 156, 168, 83 S.Ct. 239, 246, 9 L.Ed.2d 207 (1962)), cert. denied, 469 U.S. 1034, 105 S.Ct. 507, 83 L.Ed.2d 398 (1984); accord Aeron Marine Shipping Co. v. United States, 695 F.2d 567, 577 (D.C.Cir.1982) (quoting Sierra Club v. Costle, 657 F.2d 298, 323 (D.C. Cir.1981)). We must be able to discern this connection in the record and the agency decision. See SEC v. Chenery Corp., 318 U.S. 80, 87-88, 63 S.Ct. 454, 459, 87 L.Ed. 626 (1943); Vermont Department of Public Service v. FERC, 817 F.2d 127, 139 (D.C.Cir.1987). Post hoc rationalizations advanced to remedy inadequacies in the agency’s record or its explanation are bootless. See National Coalition Against the Misuse of Pesticides v. Thomas, 809 F.2d 875, 882-83 (D.C.Cir.1987). Upon review of the proceedings before the Commission, we are convinced that, in an admittedly imperfect world, the FCC’s decision nonetheless fails muster.
2
NECA’s initial submission adumbrated a three-step process for arriving at average schedule formulas. See NECA Modification at 14, J.A. at 34. Because petitioners allege that deficiencies pervade the whole process, each step, unfortunately, requires some further description than was possible in our earlier recitation of the background underlying the present controversy.
First, NECA developed “access element allocation factors.” Id.; see also 47 C.F.R. § 69.101-.115 (1986). This mind-numbing turn of phrase can best be described by borrowing NECA’s own words. NECA conceived these factors as a means of “allocating total average schedule company amounts of major categories of expense, investment, and income adjustments to access categories,” which are the categories into which access elements are grouped, namely common line (comprising NTS costs), traffic sensitive, and billing and collection. NECA Modification, app. A, at 1-3, J.A. at 44-46; see supra note 11; see also NECA Data Submission, J.A. at 767-85. To derive these factors, NECA began by selecting 255 cost companies that it considered representative of average schedule companies. NECA Modification at 14-15, *104app. A, at 1, J.A. at 34-35, 44. For every company, it calculated what percentage of total company amounts the amounts in each access category (segregated into the 21 expense, investment, and income adjustment accounts) represented. Id., app. A, at 3, J.A. at 46. Then it found the average percentage for each category among the companies studied. It tested these factors for correlation with “various demand, investment and expense variables,” id., and adjusted the factor for one investment category, “outside plant,” in light of these tests. Id.
Second, the Association compiled “a representative selection of average schedule investment and expense data that was distributed to the access elements using the access element allocation factors.” Id. at 14, J.A. at 34. It collected data from 489 average schedule companies for the period 1981 to 1983, segregating this data into the same 21 investment, expense, and income adjustment accounts into which average schedule data had been classified in step one. Id. at 16, J.A. at 36. It projected 1983 data to reflect estimated growth for average schedule companies through 1986. Id. Then it applied the allocation factors to the data in each category. Through this calculation, NECA hoped to obtain investment and expense data in each access category that would, for an average schedule company, resemble the results that would be obtained through cost studies. Id.
Finally, NECA developed formulas “to relate the selected average schedule company costs results [obtained in the second step] to appropriate access demand quantities.” Id. at 17, J.A. at 37. For non-traffic sensitive, or “common line costs,” the demand quantity selected was the access line. Id.31 Based on these formulas, NECA determined that average schedule company common line settlements should be fixed at $8.06 per access line per month. Id., app. A, at 14, J.A. at 57.
3
Having elaborated on the abstruse but important arcana of NECA’s average schedule methodology, we may now turn to the particulars of petitioners’ complaints. Petitioners criticized every aspect of the first step of NECA’s methodology. To begin with, they complained, NECA did not show how it selected the 255 cost companies studied.32 Although NECA described its selection criteria in its reply comments, this explanation was fatally inadequate in petitioners’ view because it came too late for them to comment. Even more importantly, they maintain, nothing in NECA’s data submission shows that the criteria spelled out in NECA’s reply comments were actually employed.33 Petitioners also faulted NECA's calculation of “access element allocation factors.” They pointed out that while NECA generally described these factors as based on ratios calculated for 255 cost companies and then averaged, it did not show these calculations, nor could petitioners reproduce them based on the scanty information NECA provided.34 Similarly, petitioners faulted NECA for not disclosing the various “correlation tests” it claimed to have performed for each allocation factor.35 The result, petitioners summarized, was a set of allocation factors that is grossly inaccurate or, at any rate, apparently drawn from the ether.36
*105Petitioners were similarly thorough and scathing in their evaluation of the second step of NECA’s methodology. They criticized NECA for treating 489 average schedule companies as representative of all such companies without justifying this homogenizing treatment.37 They also took issue with NECA’s estimating average companies’ growth trends on the basis of data that included data from cost companies as well as from the former.38
Finally, petitioners maintained that NECA improperly failed to reveal the source of the demand data employed in the third stage of its revision process. They added that flaws appeared on the face of the data provided.39 Perhaps most damning, to petitioners’ mind, was that in describing this third step NECA did not disclose the calculations performed in proceeding from the data tó the specific average schedules.40
In the face of these apparently substantial charges, NECA was largely unresponsive. It provided no data in addition to that supplied in November 1985. On December 12, 1985, however, after the deadline for filing initial comments had passed, it replaced a sét of tables containing obvious errors with a corrected set.41 Still later, in its reply comments, NECA described how it had selected the 255 cost companies studied in the first step of its methodology. It also defended at that late stage its inclusion of cost company data in calculating average schedule company growth rates in the second step. It furnished no data to verify these explanations, which, in any event, came too late for petitioners to question them.42
4
When we look to the Commission’s decision for its assessment of petitioners’ thoroughgoing assault, we find only the following unedifying remark: “[T]here is little doubt that improvements are possible in the studies, methodologies, and data that NECA employed in developing new average schedules.” Average Schedule Order, 103 F.C.C.2d at 1023, J.A. at 8. This is understatement in the extreme. But in any event, in the face of this acknowledgment the Commission arrived at the conclusion that NECA’s revisions represented “an improvement over existing schedules” and accordingly approved them. Id. at 1024, J.A. at 8. Yet, rather than reveal the path it took to reach this conclusion by explaining why the shortcomings pointed out with such scorching heat by petitioners did not lead to seriously flawed results, the FCC directs the onlooker to comments in support of NECA’s proposal. Id. But this leads us into a cul de sac. The only detailed supporting comments are those of NECA itself, which, as we have seen, do *106not even respond directly to most of petitioners’ criticisms.43
This will not do as reasoned decisionmaking. It may well be that the concepts undergirding NECA’s proposal — for example, shifting from a usage-sensitive basis for common line recovery to one based on access lines — were reasonable means of enhancing the accuracy of average schedules. Nonetheless, the undisputed omissions in data and methodology appear to have made it impossible to reproduce how NECA translated these concepts into hard figures.
Obviously, the FCC could reasonably have accepted lacunae in data and methodology and even have forgiven ancillary analytic weak points, in light of the difficult time constraints within which NECA had to work, the undisputed need for revisions, and the formidable nature of the task of adapting average schedules to the sea changes in separations procedures and the settlements process. Cf. Public Citizen Health Research Group v. Tyson, 796 F.2d 1479, 1493-96 (D.C.Cir.1986); MCI Telecommunications Corp. v. FCC, 627 F.2d 322, 343 (D.C.Cir.1980). We dare not overlook the realities and exigencies of regulatory life in an imperfect world and demand from our lofty perch that which could not reasonably be delivered. What is more, we would face a much different situation as a reviewing court if, besides acknowledging these shortcomings frankly, the Commission had explained why these were merely start-up problems resolvable in future revisions and did not undermine the basic improvements in NECA's overall approach. Cf. National Cable Television Association v. Copyright Royalty Tribunal, 689 F.2d 1077, 1091 (D.C.Cir.1982).
Lacking any indication of such a reasoned determination, however, we are forced to conclude that the FCC acted irrationally in glossing over gaping holes, especially in light of errors and anomalies evident in what NECA did submit. As it is not the task of a reviewing court to “rummage” through the record in search of a basis for upholding the Commission’s conclusion, Connecticut Power & Light Co., 673 F.2d at 534-35, we are obliged to conclude that the FCC’s approval of NECA’s revisions was, on this record, arbitrary and capricious. See Celcom Communications Corp. v. FCC, 789 F.2d 67, 71 (D.C.Cir. 1986) (“[T]he agency must consider the relevant evidence presented and offer a satisfactory explanation for its conclusion.”); Aeron Marine, 695 F.2d at 577-80 (finding arbitrary and capricious agency decision that lacked adequate factual predicate); see also Office of Communication, 779 F.2d at 707 (reasoned decisionmaking requires agency not employ means that undercut its ends).
C
In addition to asserting that the FCC did not articulate a rational basis for its decision, petitioners maintain that the Commission improperly failed to consider alternatives to the methodology employed by NECA. We are persuaded that petitioners advanced at least one alternative sufficiently detailed and of ample significance to merit the Commission’s consideration.
As exhibits to their reply comments, petitioners submitted “the suggestions of two distinguished average schedule experts as to how revisions could be derived in a scien*107tific and accurate manner.” ICORE Reply Comments at ii, J.A. at 207; see also supra note 36. Mr. Henry Coo outlined three alternative bases for average schedule revisions: (1) a full-scale cost study of a scientifically selected group of average schedule companies; (2) data from cost companies that, through matching of salient characteristics, would function as surrogates for average schedule companies; and (3) adaptation of current average schedules to a demand base reflective of the new access charges environment, which, among other things, would reimburse companies for NTS costs on an access line basis.44 Dr. Robert Brousseau, a second expert relied on by petitioners, elaborated on the first alternative proposed by Mr. Coo.45 The merits of a full-scale cost study alternative were also trumpeted in AT & T’s comments. See Comments of AT & T at 4-5, J.A. at 184-85.
It is well settled that an agency has “a duty to consider responsible alternatives to its chosen policy and to give a reasoned explanation for its rejection of such alternatives.” Farmers Union, 734 F.2d at 1511 (footnote omitted).46 Of course, as the Commission emphasizes in its defense, this duty extends only to “significant and viable” alternatives, Farmers Union, 734 F.2d at 1511 n. 54, not to “every alternative device and thought conceivable by the mind of man ... regardless of how uncommon or unknown that alternative may have been,” id. (quoting Vermont Yankee Nuclear Power Corp. v. Natural Resources Defense Council, Inc., 435 U.S. 519, 551, 98 S.Ct. 1197, 1215, 55 L.Ed.2d 460 (1978)). But with that sensible caveat, the fact remains that “[t]he failure of an agency to consider obvious alternatives has led uniformly to reversal.” Yakima Valley Cablevision, Inc. v. FCC, 794 F.2d 737, 746 n. 36 (D.C.Cir.1986); see also National Black Media Coalition v. FCC, 775 F.2d 342, 357 (D.C.Cir.1985); International Ladies’ Garment Workers’ Union v. Donovan, 722 F.2d 795, 815-18 (D.C.Cir.1983), cert. denied, 469 U.S. 820, 105 S.Ct. 93, 83 L.Ed.2d 39 (1984). This court has been particularly reluctant to blink at an agency’s ignoring ostensibly reasonable alternatives where it admits, as the Commission has here, that the choice embraced suffers from noteworthy flaws. Farmers Union, 734 F.2d at 1511.
In our view, the FCC has breached this duty in failing to consider the approach calling for a full-scale cost study of scientifically selected average schedule companies. This proposed alternative was certainly “significant.” More than suggesting minor improvements to subsidiary aspects of NECA’s submission, it constituted an altogether different methodological approach. It was also sufficiently “obvious” to warrant the Commission’s attention. Not only was it prominently featured in Mr. Coo’s submission, it was discussed further both by Dr. Brousseau and AT & T, the latter of the two having interests that sharply diverged from petitioners’ in many important respects. Finally, we cannot agree with the Commission’s argument that none of these alternatives was sufficiently described to be viable. FCC Brief at 35. It is clear that the full-scale cost study alternative, at least, was adapted from prior revisions to average schedules and therefore (presumably) familiar to the Commission. See J.A. at 237; see also AT & T Comments at 4, J.A. at 184. This *108alternative, then, had sufficient detail and support, in light of its familiarity, to permit evaluation and warrant attention. Cf. National Citizens Committee for Broadcasting v. FCC, 567 F.2d 1095, 1114-15 (D.C. Cir.1977) (requiring agency on remand to evaluate proposed alternative details of which had not been included in initial proceedings), cert. denied, 436 U.S. 926, 98 S.Ct. 2820, 56 L.Ed.2d 769 (1978).
Besides criticizing the proposed alternatives as insubstantial, the FCC advances two other arguments to justify its failure to discuss them. Neither, upon examination, is persuasive. First, the Commission asserts that the proposals were not worthy of mention because petitioners did not present them until they submitted their reply comments. This, the FCC contends, came too late in the day. While we recognize full well that the timing of presentation of alternatives directly affects the agency’s ability (and hence its duty) to consider them, we cannot fault petitioners under the circumstances at hand. Petitioners were afforded a mere eighteen days after NECA’s 900-page data submission to submit initial comments. We credit petitioners’ explanation that this scant period barely enabled them to evaluate NECA’s proposal and pinpoint their objections, much less formulate alternatives to it. Second, the Commission now belittles the alternatives as impractical, expensive, and time-consuming. This argument, however, posed for the first time in the Commission’s briefs to this court, is the latest in an undistinguished line of post hoc rationalizations that the Supreme Court has clearly taught does not excuse failure to consider significant alternatives at the agency level. State Farm, 463 U.S. at 49-50,103 S.Ct. at 2869-70.
It should go without saying that our determination that petitioners’ proposed alternative merited the Commission’s consideration in no wise prevents the agency on remand from rejecting it as unworkable. We hold only that, in light of the admitted shortcomings of NECA’s revisions, the Commission fell into the forbidden zone of arbitrary and capricious conduct in failing even to consider the proposed alternative.
D
Petitioners aim their final salvo at the transition plan NECA proposed to implement average schedule revisions. Specifically, they target: (1) the “flash cut,” under which companies deemed to be recovering over 85% of their total NTS costs in common line settlements suffered immediate reductions; and (2) the limit of $1.25 per access line per month for four years placed on reductions otherwise experienced by companies whose settlements decreased under NECA’s revisions. Petitioners contend that both features subject average schedule companies to harsher treatment than cost companies have experienced under analogous limitations. In addition, they argue that the “flash cut” was improperly calculated and consequently affects companies that are not in fact over-recovering.
We need not tarry long over petitioners’ charges of unfair treatment. To the extent NECA’s “flash cut” affects companies that recover more than 85% of their unseparated (i.e., combined intrastate and interstate) common line revenue requirements in the form of interstate common line settlements, average schedule companies are being treated the same as cost companies.47 That some average schedule companies suffer greater reductions than cost companies does not, as petitioners would have it, betoken unfairly discriminatory treatment. It seems to us, rather, an indication that these average schedule companies have been overrecovering to a greater extent than cost companies. We also reject petitioners’ argument that the $1.25 reduction limit spread over four years is more draconian than the analogous limit, spread over fourteen years (including a two-year grace period), applicable to cost companies. In fact, as the Commission pointed out, average schedule companies *109will reap the benefit of the fourteen-year transition because average schedules under NECA’s revisions are based on cost company data. Since that data will reflect the fourteen-year phase-in, cost companies and average schedule companies will proceed to the 25% allocatur proposed for jurisdictional separations at the same pace.48
We do find merit, however, in petitioners’ concern over the way NECA translated the 85% cost company SPF into a measurement applicable to average schedule companies. As we have seen, NECA premised its “flash cut” on the determination that average schedule companies carrying more than 15.9 messages per line per month were transgressing the 85% SPF limit.49 The only relevant data submitted, however, belies this assertion. The data consisted of a graph plotting messages per line per month against subscriber plant factor. Of the thirteen companies shown to be exceeding 85% SPF, ten have fewer than 15.9 messages per line per month. Moreover, of six companies shown to have more than 15.9 messages per line per month, only three exceed 85% SPF.50 Although NECA claimed in its reply comments that the 15.9 figure accurately classifies 97% of average schedule companies, it points to no data to support this claim.51 The Commission, consistent with its approach to other anomalies described by petitioners, failed to address this issue.
Once again, our function as a reviewing court is not to canvass the record to resolve the dispute over the accuracy of NECA’s surrogate measurement for an 85% SPF. See Home Box Office, Inc. v. FCC, 567 F.2d at 36. This is the Commission’s job, not ours. Id. Wanting an indication that the Commission resolved this dispute in a reasoned fashion, we have no alternative but to hold its approval of this aspect of NECA’s revisions arbitrary and capricious.
III
In summary, we conclude that the FCC acted arbitrarily and capriciously in failing to demonstrate a rational basis for approving NECA’s revisions to average schedules. The Commission also violated the requirements of reasoned decisionmaking by not considering one óf the alternatives to NECA’s methodology brought to its attention by petitioners, namely a full-scale cost study of selected average schedule companies. Finally, it improperly ignored petitioners’ argument that the “flash cut” proposed by NECA was grounded on an inaccurate figure.
We reject petitioners’ argument that, under the APA, the Commission must adhere to informal rulemaking procedures in passing on NECA’s revisions. Any such argument should have been advanced before the Commission prior to seeking judicial review; we see, moreover, no justification to excuse compliance with this sensible (and indeed vital) principle of administrative law.
In remanding this case to the Commission, we leave to its sound discretion to what extent, if any, it should reopen the record to satisfy the concerns we have articulated. See Bowman Transportation, Inc. v. Arkansas-Best Freight System, Inc., 419 U.S. 281, 294-95, 95 S.Ct. 438, 446, 42 L.Ed.2d 447 (1974); Camp v. Pitts, 411 U.S. 138, 143, 93 S.Ct. 1241, 1244, 36 L.Ed.2d 106 (1973) (per curiam); see also Environmental Defense Fund, Inc. v. Costle, 657 F.2d 275, 285 (D.C.Cir.1981). Far be it from us even to pretend to understand the technical intricacies that loom so large in this case. We also forebear from ordering the reassessment of settlement payments made under the average schedules proposed by NECA and approved by the Commission. That sort of order at this late stage would disrupt the settlement process and would, among other things, cause economic hardship to many compa*110nies that are not parties to the petition for review.52 See Rodway v. Department of Agriculture, 514 F.2d 809, 817-18 (D.C.Cir. 1975); Indiana & Michigan Electric Co. v. FPC, 502 F.2d 336, 343-48 (D.C.Cir.1974), cert. denied, 420 U.S. 946, 95 S.Ct. 1326, 43 L.Ed.2d 424 (1975); see also Western Oil & Gas Association v. EPA, 633 F.2d 803, 813 (9th Cir.1980). Accordingly, we leave to the Commission’s judgment in the first instance how best to accommodate these various interests in light of the proceedings on remand.
The petition for review is
Granted.
. See Smith v. Illinois Bell Tel. Co., 282 U.S. 133, 148, 51 S.Ct. 65, 68, 75 L.Ed. 255 (1930); see also United States v. American Tel. & Tel. Co., 552 F.Supp. 131, 141 (D.D.C.1982) (describing “exchange areas” in which former Bell Operating Companies would operate as independent companies), aff’d mem. sub nom. Maryland v. United States, 460 U.S. 1001, 103 S.Ct. 1240, 75 L.Ed.2d 472 (1983).
. See also 47 U.S.C. § 410(c) (1982) (providing for a Federal-State Joint Board to address certain issues of allocating local plant costs between interstate and intrastate jurisdictions); Smith, 282 U.S. at 148-51, 51 S.Ct. at 68-69 (percentage of local plant costs used for interstate service is within federal jurisdiction).
. For a full discussion of this distinction, see NARUC, 737 F.2d at 1104; see also MCI, 750 F.2d at 137; MTS & WATS Market Structure, 93 F.C.C.2d 241, 268-69, reconsid., 97 F.C.C.2d 682 (1983), further reconsid., 97 F.C.C.2d 834 (1984), aff’d in part, NARUC, 737 F.2d 1095 [hereinafter Access Charges Order]. An example of a non-traffic sensitive cost is that of installing a line, or "loop,” 47 C.F.R. § 67.701 (1986), between a subscriber’s premises and a local telephone office; the cost is the same regardless of how many calls the subscriber places. The cost of the switching equipment at a local office that routes calls from one line to another, in contrast, tends to be traffic sensitive. Larger, more expensive switches become necessary as usage increases. NARUC, 737 F.2d at 1104; see also FCC Brief at 4. But cf. Access Charges Order, 93 F.C.C.2d at 269 (indicating that a portion of local dial switch is classified as non-traffic sensitive for separations purposes in order to segregate cost of terminating a line in the switch from costs of switching).
. See United States v. American Tel. & Tel. Co., 552 F.Supp. 131.
. Correlatively, subscribers who neither placed nor received interstate calls provided no revenues for cost recovery, even though exchange companies incurred interstate NTS costs in providing those subscribers with access to an exchange system that carried interstate as well as intrastate calls. Id. at 1108.
. An "access line” essentially comprises the transmission facilities between a subscriber’s premises and the central office of an exchange company. See 47 C.F.R. § 67.701 (1986). Thus, the number of access lines that a company serves is approximately equal to the number of subscribers it serves. Petitioners’ Brief at 16 n. 51.
.In addition to a flat charge per access line imposed on end users, the FCC has provided for recovery of the interstate NTS costs of local telephone plant through two additional charges: (1) interexchange (i.e., long-distance) carriers pay a "carrier’s carrier” charge; and (2) users of private lines pay a surcharge. NARUC, 737 F.2d at 1110.
. 47 C.F.R. § 69.606(b) (1984) provided in relevant part:
The association shall submit a proposed revision of the formula [governing 'average schedule compensation] for each annual period subsequent to May 31, 1985 or certify that a majority of the directors of the association believe that no revisions are warranted for such period on or before November 30 of the preceding year.
In. 1985, the FCC amended § 69.606(b) to. provide that any "proposed revision of the [average schedule] for each annual period subsequent to December 31, 1986" must be submitted “on or before June 30 of the preceding year.” 47 C.F.R. § 69.606(b) (1986).
. In addition to its duty to propose average schedule revisions, NECA also has broad responsibilities for preparation of access charge tariffs and collection and distribution of access charge revenues. See 47 C.F.R. § 69.603. The Access Charges Order required all exchange carriers to recover the "carrier common line charge" levied on interexchange carriers through NECA distributions. See Access Charges Order, 93 F.C.C.2d at 327-30; see also 47 C.F.R. § 69.3; NECA Brief at 5 n. 8. An exchange company’s common line revenue requirements are also recovered through common line charges levied upon end users, known as “the end user common line charge." See Access Charges Order, 93 F.C.C.2d at 328. Exchange companies are not required to recover these charges through NECA. Id.
. See NECA Modification of Average Schedules at 9-10 (Sept. 16, 1985), J.A. at 29-30. To be specific, the usage-sensitive ARPM method in effect prior to the NECA modifications was used to compute settlements under the schedule designated A-1. See id. at 7, 10, J.A. at 27, 30. The A-1 schedule compensated average schedule exchange companies primarily for interstate common line costs, which are interstate costs of exchange facilities that carry both interstate and intrastate messages. See id. at 7, J.A. at 27; see also Average Schedule Order, 103 F.C.C.2d at 1019, J.A. at 2. In addition to the A-1 schedule, numerous other schedules covered various categories of interstate costs. Nonetheless, average schedule settlements as a whole may be considered usage-sensitive and ARPM-based be*97cause A-1 settlements represented 90% of average schedule settlements. Comments of the ICORE Local Exchange Companies on the Proposed Revisions to the Average Schedule Formulas, exh. 6, at 2-3 (letter from Mr. Henry Coo to Mr. Jan Reimers), J.A. at 152-53; see also NECA Modification at 5-9, J.A. at 25-29; Average Schedule Order, 103 F.C.C.2d at 1019, J.A. at 3.
.The dispute before us primarily concerns NECA’s decision to base common line settlements on access lines. The Association also proposed modification to settlements for traffic sensitive costs and billing and collection costs, the other two categories of access elements. See NECA Modification at 10-13, J.A. at 30-33; see also Average Schedule Order, 103 F.C.C.2d at 1022, J.A. at 6; ALLTEL Comments at 4, J.A. at 294 (summarizing NECA’s overall approach).
. In addition, NECA stated that it would continue to study those companies that stood to. lose more than $5 per line after enduring the four-year cuts. Id., app. A, at 12, J.A. at 55.
. In addition to the notice provided by the Commission in October 1985, NECA sent "impact statements” to each average schedule company in July 1985 to apprise them of the effect the revisions would have on each. See NECA Modification at 1, J.A. at 21.
. Motion of Jefferson Telephone Co. and Northwest Iowa Telephone Co. for Leave to File Request for Production of Data and Studies Supporting the Revised Average Schedule Formulas (filed Oct. 28, 1985), J.A. at 69.
. NECA’s Reply to Request for Production (Nov. 7, 1985), J.A. at 77.
. Renewed Motion of ICORE Companies for Production of Data and an Extension of Time to File Comments (Nov.. 27, 1985), J.A. at 83. ICORE is a consulting firm that represents independent local telephone companies. Its members include the twelve petitioners. See Petitioners’ Brief at 3 & n. 2.
.Comments of the ICORE Local Exchange Companies on the Proposed Revisions to the Average Schedule Formulas at 22-23 (Dec. 6, 1985), J.A. at 113, 140-41.
. See Letter from G.R. Evans, Director of Tariff & Regulatory Matters, NECA, to Wm. Tricarico, Sec’y, FCC (Dec. 12, 1985), J.A. at 195.
. NECA Reply Comments at i (Jan. 6, 1986), J.A. at 257, 259.
. Average Schedule Order, 103 F.C.C.2d 1017, J.A. at 1. The Commission also ruled on two aspects of its access charge rules that were remanded to it for further consideration in NARUC, 737 F.2d at 1147. These rulings are not before us for review.
. This fourth reason cited by the Commission was briefly mentioned as an improvement in comments submitted by ALLTEL Corp. See Reply Comments of ALLTEL Corp. at 8 (Jan. 6, 1986), J.A. at 298.
. See MTS & WATS Market Structure: Average Schedule Companies, CC No. 78-72, Phase I (Commission Memorandum Opinion & Order dated Sept. 30, 1986) [hereinafter Reconsideration Order], J.A. at 375.
. Id. at 14-22 (discussing waiver applications of Northwest Iowa Telephone Co. and Jefferson Telephone Co.), J.A. at 388-96; Petition of Brindlee Mountain Telephone Co. for Waiver of the Average Schedule Revisions (filed Sept. 25, 1986).
. In its Reconsideration Order, the FCC denied Northwest Iowa's waiver petition, which was based on asserted "extreme hardship.” The Commission concluded that the company had received adequate notice of the revisions to protect itself against the hardship of which it complained. Id. at 16-17, J.A. at 390-91. The Commission did not, in contrast, finally dispose of Jefferson’s waiver request. Jefferson sought relief from the "flash cut” called for in NECA's transition plan. The company claimed that the “flash cut” should not be applied to it because, contrary to NECA's estimates, Jefferson was not recovering more than 85% of its unseparated NTS costs; in fact, Jefferson maintained, it was recovering only 60%. Id. at 18-19, J.A. at 392-93. The Commission found data and analysis submitted by Jefferson too speculative to support a waiver, but sufficiently detailed to justify further study. Id. at 20-22, J.A. at 394-96.
The third petition, filed by Brindlee Mountain, was based on the assertion that Brindlee had initially received an impact statement from NECA indicating that it would benefit from the revisions, but was informed in a later impact statement, received after the comment period ended, that it would actually suffer reduced settlement payments. Petitioners' Brief at 13 n. 39, 24. The Commission denied this petition in an order dated March 30, 1987. Petitioners’ Letter Advising of Additional Authority (filed Apr. 14, 1987) (citing Memorandum Opinion & Order, FCC No. 87-72 (Mar. 30, 1987)).
.Reconsideration Order at 23-24, J.A. at 397-98.
. See also United States v. LA. Tucker Truck Lines, 344 U.S. 33, 37, 73 S.Ct. 67, 69, 97 L.Ed. 54 (1952):
Simple fairness ... requires as a general rule that courts should not topple over administrative decisions unless the administrative body not only has erred but has erred against objection made at the time appropriate under its practice.
. This was only the second time average schedule revisions have been submitted under Commission rule 69.606; the first set of revisions, submitted by AT & T, were essentially unopposed and thus did not present the opportunity to rule on the procedural issues now raised. See FCC Brief at 12-13 (quoting MTS & WATS Market Structure, FCC Mimeo No. 83-509, at ¶ 8 (Nov. 14, 1983)).
. We acknowledged in WATCH that other exceptions to the general rule of exhaustion may exist, 712 F.2d at 682-83 & n. 11, but none of the examples cited in that case applies here.
. The Commission and NECA point out that petitioners do not, as part of their substantive challenge, allege that under NECA’s revisions they will be unable to recover their costs of providing interstate service. On this basis, NECA and the FCC mount two related legal arguments against judicial review. Neither of them has merit.
Their primary contention is that Federal Power Comm'n v. Hope Natural Gas Co., 320 U.S. 591, 64 S.Ct. 281, 88 L.Ed. 333 (1944), precludes judicial review. Hope articulated an "end result” test under which courts may not review the method by which rates are set; they may only determine whether the results are unreasonable. Id. at 602, 64 S.Ct. at 287. Since petitioners have not impugned the end result of NECA’s revisions, the argument runs, they may not take issue with the methods employed. See FCC Brief at 25-27; NECA Brief at 21-27. This argument fails for several reasons. First, the Commission itself classifies average schedule companies as such because it has determined that they have insufficient resources to conduct cost studies; they would therefore hardly seem to be in a position to determine whether NECA’s revisions permit recovery of full costs. Second, the Hope doctrine applies to ratemaking; NECA's revisions affect the division of revenues among carriers, not rates. Third, assuming arguendo NECA and the FCC negotiated these *103logical hurdles, they would run up against City of Charlottesville v. FERC, 661 F.2d 945 (D.C.Cir. 1981). In that case we held that the "end result” standard of review applicable to ratemaking had "evolved” since Hope was decided. "Experience has taught that a determination of whether the result reached is just and reasonable requires an examination of the method employed in reaching that result.” Id. at 950.
In addition to advancing the “end result" rationale, NECA cites a line of precedent concerning review of ICC approval of industry-wide railroad rate increases. NECA Brief at 17-20. These cases, upon examination, are wholly inapposite. For one thing, like the "end result” line of precedent, these ICC cases concern ratemaking, not division of revenues. Furthermore, these decisions involve rate changes initiated by carriers; in contrast, the revisions before us for review were proposed by NECA, an entity created and shaped by Commission regulation. See, e.g., Arizona Elec. Power Co-op, Inc. v. ICC, 675 F.2d 303, 305-06 (D.C.Cir. 1982). Thus, while it is true that the courts have determined that the ICC’s decision whether to suspend or investigate carrier-initiated rate increases is "committed to ICC discretion and therefore ... not judicially reviewable,” id at 306, that precedent simply cannot be stretched to fit the different factual and legal setting of this case.
. Petitioners do not seriously dispute that the FCC’s goal of adopting average schedules that reflect the shift in reimbursement of cost companies was lawful and reasonable. See Oral Argument Tr. at 15-16 (Mar. 27, 1987). Indeed, the rule calling for such a shift, section 69.606, was part of the Commission’s Access Charges Order and was not judicially challenged. See 93 F.C.C.2d 241, app. A, at 362.
. The demand unit on which revised schedules for traffic sensitive costs were based was the access minute. See NECA Modification at 12, J.A. at 32. The unit of demand on which revised billing and collection schedules were based was the message. Id. at 13, J.A. at 33. See also id., app. A, at 6, 15-18, J.A. at 49, 58-61; supra note 11.
. See ICORE Comments at 20-21, J.A. at 138-39.
. See Oral Argument Tr. at 7. Compare NECA Reply Comments at 13, J.A. at 274 with NECA Data Submission, J.A. at 719.
. See ICORE Comments at 23-24, J.A. at 141— 42; ICORE Reply Comments at 17-18, J.A. at 224-25.
. See ICORE Comments at 24, J.A. at 142.
. To buttress this contention, petitioners submitted the calculations of Mr. Henry Coo, an analyst who had overseen the development of average schedules for AT & T before the responsibility for average schedule revision was shifted to NECA. See Petitioner’s Brief at 21; see also infra text at 1168-69. Mr. Coo purported to demonstrate that when one attempted to work "back*105wards" from the allocation factors to the figures of the individual cost companies on which the factors were based, one obtained results that were woefully inadequate in replicating the actual costs reported by the cost companies. See ICORE Reply Comments, app. B, J.A. at 234-38.
. ICORE Comments at 24, J.A. at 142.
. Id. at 25 & n. 54, J.A. at 143 & n. 54.
. Id. at 26, J.A. at 144; ICORE Reply Comments at 19, J.A. at 226.
. ICORE Comments at 26, J.A. at 144.
. The tables to which petitioners drew NECA’s attention early in the proceedings contained common line allocation factors multiplied by average schedule financial data that had been broken down into 21 major investment and expense accounts. For several account items to which NECA assigned allocation factors of zero, it multiplied the zero factor by the account item to produce non-zero products. Conversely, a number of multiplications of non-zero factors by non-zero account items yielded products of zero. See ICORE’s Renewed Motion for Production of Data at 6 & exhs. 3-4, J.A. at 88, 106-07. Having received no response to this discovery, petitioners renewed their objections to these math errors in comments filed December 6, 1985. ICORE Comments at 22-23, J.A. at 140-41. On December 12, 1985, NECA submitted new tables “which corrected] prior submissions for Allocation Factor Development” and announced that “no further data [would] be forthcoming." Letter from G.R. Evans, J.A. at 195.
.Petitioners did attempt to respond to these explanations of NECA. See Motion of ICORE for Leave to File Response to Reply Comments (Feb. 7, 1986), J.A. at 324. The Commission denied the motion in its Reconsideration Order. See Reconsideration Order at 24, J.A. at 398.
. Besides NECA, three parties to the proceedings below filed comments in support of the revisions. One of the supporters, the Rural Telephone Coalition, did not directly address the specifics of petitioners' complaints; rather, it more generally supported the "concept" underlying the revisions. See Comments of RTC at 3-4 (Dec. 6, 1985), J.A. at 188, 190-91; see also Reply Comments of RTC (Jan. 6, 1986), J.A. at 304. The ALLTEL Corporation made similarly generalized comments on NECA’s methodology and data, asserting, for example, that "the methodology ... is in keeping with past practice— The data used was reasonably representative and reasonably reliable.” Reply Comments of ALLTEL Corp. at 5 (Jan. 6, 1986), J.A. at 291, 295. It is worth noting, however, that ALLTEL implicitly recognized some cogency in petitioners’ criticism when it opined that this criticism "should be used as a guide to developing future average schedules [but] not as a reason to delay the implementation of those porposed [sic] by NECA now.” Id. at 6, J.A. at 296. Southwestern Bell Telephone was the third NECA supporter; it expressed its approval in two and a half pages. See Reply Comments of Southwestern Bell Telephone Co. (Jan. 6, 1986), J.A. at 300.
. ICORE Reply Comments, exh. B. at 4-6, J.A. at 237-39.
. Id., exh. C, at 3-5, J.A. at 251-53.
. We note that the duty to consider alternatives in this case does not depend on whether the Commission was engaged in promulgating a substantive rule, as petitioners maintain, or merely implementing a pre-existing rule, section 69.606, as the Commission contends. We do not resolve this dispute over the characterization of the FCC’s actions for reasons discussed at length, supra section II-A. Nor is resolution of this procedural issue necessary to addressing the Commission’s duty to consider significant alternatives. That duty inheres in the agency's broader responsibility for exercising its expertise in a reasoned manner. See Motor Vehicle Mfrs. Ass'rt v. State Farm Mut. Auto. Inc. Co., 463 U.S. 29, 48-49, 103 S.Ct. 2856, 2869, 77 L.Ed.2d 443 (1983). Indeed, the Commission does not dispute its obligation to entertain significant alternatives. Instead, it argues that petitioners’ proposals were not significant.
. See Joint Report by USITA Settlements & Separation Committee & Bell System Representatives (Apr. 5, 1971), J.A. at 63, 65; see also FCC Brief at 44 n. 56; Petitioners’ Brief at 8 n. 15.
. Indeed, petitioners have implicitly conceded this point in their reply brief. See Reply Brief at 23.
. See NECA Modifications, app. A, at 13, J.A. at 56.
. See NECA Data Submission, J.A. at 945-47.
. See NECA Reply Comments at 20, J.A. at 281.
. Although the exact numbers are in dispute, it appears that a majority of the 732 average schedule companies receives increased settlements under NECA’s revisions. See NECA Modification at 20, app. A, at 10, J.A. at 40, 53; cf. Reply Brief at 25 n. 55.