BOUDIN, Circuit Judge.

The petitions for review in this case challenge orders of the Federal Energy Regulatory Commission ("FERC") that have the effect of reinstituting an earlier charge — the so-called installed capability ("ICAP") deficiency charge — paid by electric utilities in New England who fail to meet certain reserve capacity requirements. This court stayed implementation of all but a small portion of the charge pending judicial review. The background is complicated but can be summarized as follows.

In New England, as in other regions of the country, electrical power is furnished through a grid of interconnected intercity transmission lines and local distribution lines within each city or town. Power is generated within the region, or purchased from outside ( e.g., from Canada). Some utilities engage in all three functions (generation, intercity transmission, local distribution); but many are local "retailers" who own only the local distribution facilities within a town and purchase all of their needs from generating utilities that have surplus power.

Regulation is divided between FERC and state regulatory commissions, but generally speaking, FERC regulates wholesale transactions ( e.g., between a power-supplying utility with a surplus and a local municipal utility that retails power). 16 U.S.C. § 824(b)(1) (1994); FPC v. Conway Corp., 426 U.S. 271, 276-77, 96 S.Ct. 1999, 48 L.Ed.2d 626 (1976). In the past, power was provided largely on a non-competitive basis with regulated rates based upon costs; but, as in other industries like telecommunications, some power regulators, including FERC, have been moving in recent years toward more reliance upon competition. Town of Norwood v. FERC, 202 F.3d 392, 396 (1st Cir.), cert. denied, 531 U.S. 818, 121 S.Ct. 57, 148 L.Ed.2d 24 (2000).

An abiding concern in regulating electricity supply is the need for adequate reserve capacity. The demand for electricity varies, depending on weather, economic growth, and other factors; supply is constrained by the time needed to build new generating plants and by unexpected breakdowns in generation or transmission facilities; and electricity cannot economically be stored for future use in large quantities. To avoid the extraordinary disruption of blackouts, regulators and utilities calculate reserve requirements based on estimates of how much generating capacity will be needed at the highest point of the peak load.

Our case concerns the mechanism for assuring that utilities will have the reserve power needed to satisfy peak demand. Because the system as a whole must be built to satisfy peak demand, a good deal of generating capacity is destined to lie idle at least some of the time. A prudent utility that retails power to local customers will purchase from a generating utility a certain amount of reserve capacity — in effect, capacity that is reserved for that retailer but may be used only rarely. But it is not certain that under present industry conditions the retailer has adequate economic incentive to purchase all the reserve capacity that it needs. And if reserve capacity is not purchased in sufficient amounts, generating utilities may lack the incentive to build as much capacity as is needed to supply peak demand.

The incentive structure in the industry is immensely complicated and economists take different views on how best ( i.e., cheaply and reliably) to secure adequate reserves. But, for a number of years prior to 1998, FERC and the New England utilities under its supervision used the so-called ICAP deficiency charge as a device to assure that purchasing utilities would buy adequate reserve capacity to cover projected peak load plus a reserve margin. ISO New Eng., Inc., 91 FERC ¶ 61,311, at 62,080, 2000 WL 863242 (2000); see also Municipalities of Groton v. FERC, 587 F.2d 1296, 1302 (D.C. Cir. 1978).

In New England, beginning some time before 1990, wholesale tariffs specified that load serving entities ("LSEs") — crudely, electricity retailers — had to pay a charge if they did not purchase enough reserve capacity. From 1990 onward, the charge was $8.75 per kilowatt-month; thus, if during a particular period the reserves purchased by an LSE fell below its allocated share of all needed reserves, the LSE paid the specified charge times the amount of its deficiency. Through a complicated arrangement, most of the money went as extra compensation to power-supplying utilities, supplementing the amounts they earned by selling electrical power and committing reserve capacity.

The theory underlying the $8.75 figure, which prevailed in New England from 1990 to 1998, was that it approximated the cost (appropriately amortized) to construct a kilowatt-month of new generating capacity available for peak demand conditions, plus an additional "penalty." The penalty's theoretical basis is not crystal clear, but the existence and value of the penalty are not issues in this case. The key issue is whether FERC, after allowing the $8.75 charge to be abandoned in New England in 1998 and 1999, could reinstate it in 2000 in the way that it did.

In 1998, New England utilities obtained FERC's approval to abandon the flat ICAP deficiency charge in favor of a so-called auction market for buying additional reserve capacity, that is, required capacity over and above that acquired through bilateral contracts (which are common and often long-term). However, the auction market appears not to have been successful — among other things, the prices were thought to have been subject to manipulation. In March 2000, ISO New England, Inc. ("ISO-NE") — the "independent system operator" that manages power transactions on behalf of the utilities in the New England Power Pool — sought FERC's consent to an end to the auction regime.

On June 28, 2000, FERC released a lengthy order ("the June 28 order") that largely addressed other ISO-NE issues but which toward its end addressed the ICAP issue in a few paragraphs, ISO New Eng., 91 FERC ¶ 61,311, at 62,080-81. In the order, FERC agreed with ISO-NE that the auction did not work, but, because ISO-NE had not proposed an alternative means to ensure that LSEs would purchase adequate reserve capacity, FERC said that although it would permit an end to the auction as of August 1, 2000, it would "also require the ISO to revert to administratively-determined sanctions for failure [of LSEs] to meet the existing ICAP requirement," i.e., the projected reserve capacity requirement. Id. at 62,081. ISO-NE was directed to file tariffs "proposing an appropriate ICAP deficiency charge within 30 days." Id.

ISO-NE responded with a filing that proposed a charge of $0.17 — a fraction (less than 2 percent) of the charge that had prevailed from 1990 to 1998; this figure was proffered as an average price paid for power during 1999 on the now abandoned auction market. Various utilities who supply surplus power protested bitterly; they not only challenged the proposal as unsupported but said that if adopted this minimal figure would become a de facto cap on prices in the bilateral-contract market in which much of the reserve capacity is purchased, further reducing incentives for generating utilities to construct new plant.

On December 15, 2000, FERC issued a fourteen-paragraph order ("the December 15 order") rejecting the proposed $0.17 charge. FERC brushed aside this figure as "a token payment," saying that use of the auction average was not appropriate. ISO New Eng., Inc., 93 FERC ¶ 61,290, at 61,974-75, 2000 WL 1840335 (2000). FERC then ordered reinstatement of the original $8.75 figure, effective August 1, 2000, until ISO-NE proposed some other adequate regime. FERC said that $8.75 "represents an approximation of the cost to install a peaking unit and represents a reasonable basis for setting a level to incent the construction of new generation." Id. at 61,975.

Petitions for rehearing followed, and FERC deferred the $8.75 charge pending their disposition. On March 6, 2001, FERC issued a decision on reconsideration ("the March 6 order"), rejecting the petitions with one qualification. ISO New Eng., Inc., 94 FERC ¶ 61,237, at 61,846-47, 2001 WL 275399 (2001). After elaborating on its reasons for disallowing the $0.17 charge and requiring a return to the $8.75 level, FERC said that it would make the higher charge effective only from April 1, 2001, onward; because of intervening reliance on the $0.17 figure, it would be used for payments from August 1, 2000, through March 31, 2001. Id.

Various utilities then petitioned for review in this court, and other utilities intervened on both sides. 16 U.S.C. § 8251(b). This court then granted petitioners' motion to stay the $8.75 charge pendente lite and expedited briefing. As the stay order permitted, FERC then ordered the $0.17 charge to continue in effect pending this court's review. Briefing and oral argument in this court followed. In this court, petitioners and others who support them principally attack the adoption of the $8.75 charge and, secondarily, the rejection of ISO-NE's proposed $0.17 charge. The standard of review varies, of course, with the particular claim of error.

On review, petitioners' main assault is on FERC's imposition of the $8.75 charge. They do not contest FERC's statutory power to adopt an ICAP charge pursuant to its authority to set just and reasonable rates for wholesale power, 16 U.S.C. §§ 824(b)(1), 824e(a) (1994); but they say that the $8.75 charge is not supported by substantial evidence. We begin with this claim and consider, in later sections, related claims that FERC failed to consider reasonable alternatives and that the orders are procedurally flawed (above all, by the lack of an evidentiary hearing).

The Basis for the $8.75 Charge. Although petitioners' caption refers to a lack of substantial evidence for adopting the $8.75 charge, their attack is actually a multi-faceted challenge to FERC's reasoning, substantive judgments, alleged misstatements and asserted failures to discuss issues or evidence. There is, indeed, a good deal to criticize in FERC's orders, and we conclude that further proceedings are necessary. However, the criticisms can be understood and assessed only against the background of FERC's basic rationale for adopting the $8.75 charge, and it is useful to set forth at the outset our understanding of what FERC did and why it did it.

As already noted, FERC long ago approved an ICAP charge for the New England area to cope with what it conceives to be a need to assure adequate plant construction and margin of reserves. Such ICAP charges are used in the New York region and in the Pennsylvania-New Jersey-Maryland area; how the matter is handled away from the east coast is less clear. See note 5 below. FERC has been moving federal energy regulation toward more competition, but the transition is far from complete, and no one knows for certain whether special incentives will be needed indefinitely or what kind would work best.

Against this background, FERC has insisted that reserve requirements be maintained but, in 1998 and 1999, it allowed ISO-NE to substitute a short-term auction market as an enforcement device. That scheme, it appears, did not work and petitioners do not contend otherwise. So FERC simply restored the pre-existing $8.75 ICAP charge for the time being while making clear that it was open to other proposals from ISO-NE or others for a better way to assure adequate reserves. In the abstract, there is nothing in this rationale that is startling, let alone irrational; but the devil, as usual, is in the details.

First, petitioners attack FERC's statement that $8.75 "represents an approximation of the cost to install a peaking [generating] unit." In fact, say petitioners, FERC has not cited "even one iota of evidence in support of this fact-intensive assertion" and the highest figure that "even the proponents of the $8.75 charge could muster [in affidavits] . . . was $5." Petitioners may be correct on the numbers, but the problem is with FERC's verbal sloppiness rather than any more fundamental defect.

As petitioners know perfectly well, the $8.75 figure was, even when adopted years ago, an overstatement of the cost of a peaking unit; the record before FERC in these proceedings established that in 1990 the peaking unit cost was estimated at about $5 and the balance, and certainly the balance above $6, see note 2 above, therefore represented a further "penalty" charge with a different rationale, see New Eng. Power Pool Agreement, 56 FPC 1562, 1599-1600, 1976 WL 14771 (1976), petition for rev. denied sub nom. Municipalities of Groton v. FERC, 587 F.2d 1296 (D.C. Cir. 1978). Petitioners do not even explicitly address, let alone challenge, the rationale for the additional "penalty" charge.

Thus, it is apparent that when FERC called the $8.75 charge "the cost" of a peaking unit, it was adopting the original rationale as well as the original figure, but carelessly simplifying the rationale. This is a venial sin except so far as it symbolizes FERC's broader failure to address carefully and fully objections to a rate change — the forced re-adoption of the original $8.75 charge — that might increase local electrical rates in New England by many millions of dollars annually.

Admittedly, the actual impact of the $8.75 charge on retail electric rates in New England is not certain — despite dire predictions from retailing utilities. The ICAP charge is paid only if an LSE fails to buy adequate reserve capacity in advance; the charge, including the penalty, is the spur to do so. So if LSEs comply with their obligations through bilateral contracts, there remains a question as to how much the background ICAP charge will drive up retail prices in New England. In any event, FERC's description of $8.75 as an approximation of generation costs is not a serious problem, unless actual peaking costs are shown to have fallen (an issue to which we return below).

Petitioners' second claim of error is more basic. Petitioners point to FERC's central determination that $8.75 represents "a reasonable basis for setting a level to incent [ i.e., create the incentive for] the construction of new generation." They then say that ISO-NE showed that there was enough planned construction in New England to provide adequate reserves without the charge; that FERC did not respond to affidavit showings by opponents that the charge was unnecessary; that there is nothing to show that a charge above $0.17 is needed; and that merely to point to the prior $8.75 charge is insufficient given the lapse in time and changes in the industry.

These contentions have more force than FERC admits but less than petitioners pretend. As to whether any charge is needed, FERC says on appeal that this issue is not properly presented because it was decided by FERC in its June 28 order and judicial review from that order has not been sought. But the reason review has not been sought is that petitions for rehearing were filed at FERC and FERC has still not acted upon them. We doubt that an agency can use a prior order as a premise for a drastic second step — a possibly multimillion dollar rate increase — but insulate the premise from review by failing to act on rehearing petitions. Cf. Competitive Telecomms. Ass'n v. FCC, 87 F.3d 522, 531 (D.C. Cir. 1996).

In all events, the question whether any charge is needed overlaps with the further questions whether a substantial charge is needed and whether $8.75, $0.17 or some other figure should be used. This last question, at least, was not decided in the June 28 order; indeed, it arose in response to that order. Nor did FERC make in the June 28 order any detailed findings of a kind that might control at the second stage, the setting of an interim charge. In short, FERC's "jurisdictional" objection to our consideration of the claimed error is hollow.

On the other hand, petitioners' own claims on the merits are overstated. That ISO-NE says reserves are adequate without a charge proves little: it appears that most of the utilities served by ISO-NE are net buyers and have a self-interested incentive to object to a charge. And even if present "plans" promise ample new construction, FERC's orders (bolstered by affidavits presented below) apparently reflect the reasonable view that plans are only that and can easily be affected by the prospect that reserve requirements for buyers will or will not have teeth. If a buying utility can purchase below its reserve requirement without penalty, it may find it worthwhile to do so most of the time — while counting on the system to bail it out in a crisis.

The more difficult question concerns FERC's failure to discuss in any detail the extensive affidavit filings by objectors who claim that — given the current state of the industry and conditions in New England — no substantial charge or at least none greater than $0.17 is needed. Some of these filings are by recognized experts and are substantial and detailed; and although there are counter-affidavits from utilities who are presumably net sellers, this is the kind of material that one would normally expect an agency to analyze seriously in its decision before making drastic changes, especially where millions of dollars may ride on the outcome. E.g., NorAm Gas Transmission Co. v. FERC, 148 F.3d 1158, 1162-65 (D.C. Cir. 1998); K N Energy, Inc. v. FERC, 968 F.2d 1295, 1302-04 (D.C. Cir. 1992).

Yet this case is not the typical one in which something drastic is being done for the first time. The reserve requirements have long existed in New England and are not directly challenged by petitioners. And enforcement through a substantial ICAP charge (indeed, from 1990 on with the precise $8.75 figure) had been standard practice in New England for over a decade before the temporary use of an auction in 1998 and 1999 which was either a failure (as FERC found) or at least not successful enough for petitioners to defend in this court. So FERC's best answer is that the real status quo ante is the $8.75 figure and that the burden was really upon petitioners to justify a departure ( i.e., no charge or something less than $8.75).

No one has cited any case law closely in point, but common sense suggests the answer. FERC did not in the first instance have to provide additional justification either for the need for a substantial charge or for the $8.75 figure; but given the detailed arguments in opposition and the possible dollar impact of reinstatement of the $8.75 charge, it owed petitioners (and the public who will likely pay some of any ICAP charge through passed-on retail rate increases) some explanation as to why FERC was not persuaded by petitioners' efforts to discredit the notion of a substantial charge in general or the $8.75 charge in particular.

Undoubtedly FERC was annoyed, perhaps rightly, when its direction to ISO-NE to reinstitute an administrative charge was met by a proposal that FERC regarded as risible: a figure that was a tiny fraction of the prior charge and of the ICAP charges prevailing elsewhere on the Atlantic coast. And given the impact of any widespread power shortage, FERC has every reason to want effective enforcement of reserve requirements, even if enforcement comes at a high price. Still, FERC is not therefore excused from explaining its actions. Addressing contrary arguments is part of establishing public acceptability and, in any event, is part of FERC's own responsibility. See Motor Vehicle Mfrs. Ass'n of U.S., Inc. v. State Farm Mut. Auto. Ins. Co., 463 U.S. 29, 42-44, 48-50, 103 S.Ct. 2856, 77 L.Ed.2d 443 (1983).

What we have said applies as much to the specific choice of the $8.75 figure as to the need for a substantial charge. Absent evidence of significant change in construction costs, FERC was entitled to revert to the pre-existing figure rather than justify it afresh in the first instance, cf. S. La. Area Rate Cases v. FPC, 428 F.2d 407, 433 (5th Cir.), cert. denied, 400 U.S. 950, 91 S.Ct. 241, 27 L.Ed.2d 257 (1970); but it was not entitled to ignore claims that the cost of peaking facilities is less than it was in the past. Such claims have been made in this case; perhaps they can be easily answered; or, if lengthy proceedings would be needed to do so, FERC may be able to justify restoring the $8.75 charge on a provisional basis while making the evaluation. But some response from FERC is required.

An agency's expert judgments are entitled to deference, especially where safety concerns are on one side of the balance, as they are here. E.g., Chem. Waste Mgmt., Inc. v. U.S. EPA, 869 F.2d 1526, 1538-40 (D.C. Cir. 1989); cf. K N Energy, 968 F.2d at 1303. If FERC had provided even a semblance of serious discussion as to why a substantial ICAP charge was still required and why the pre-existing figure was the best solution on short notice, quite probably the charge would be sustained outright. California's recent experience is likely irrelevant as to immediate causes but not as to consequences. But this is all the more reason for FERC to write a decision reflecting a serious look at objections.

Alternatives. In addition to attacking the $8.75 charge as unsupported, petitioners make a series of further arguments, beginning with the claim that FERC "failed to consider reasonable alternatives [to the $8.75 charge], which it was obligated to do under reasoned decision-making principles." Petitioners say that "FERC's only response was that it was inappropriate to consider these alternatives when proposed in a rehearing request," and they contend that this response is inadequate because rehearing was "the first time" in which parties opposed to the $8.75 charge could propose alternatives.

Petitioners have markedly overstated any legitimate grievance. Broadly speaking, agencies are expected to consider reasonable alternatives to proposed actions, Farmers Union Cent. Exch., Inc. v. FERC, 734 F.2d 1486, 1511 (D.C. Cir.), cert. denied, 469 U.S. 1034, 105 S.Ct. 507, 83 L.Ed.2d 398 (1984); but here FERC did not refuse to consider alternatives or rest "only" on the failure to present the alternatives earlier. It said that the record did not provide a basis to sustain the main alternative urged ( i.e., using bilateral contract prices to construct an ICAP charge) and that ISO-NE was welcome to formulate alternatives and support them in a new filing. ISO New Eng., 94 FERC ¶ 61,237, at 61,847.

Furthermore, the sense that agencies should address alternatives before acting is not hard and fast: if prompt action is necessary and delay would be harmful, agencies sometimes do need to take interim action, deferring to further proceedings other facets of the problem or alternative solutions that may take more time to develop. See Competitive Telecomms. Ass'n, 87 F.3d at 531; Nat'l Air Carrier Ass'n v. Civil Aeronautics Bd., 436 F.2d 185, 194-95 (D.C. Cir. 1970). And, for multiple reasons in this case, there is nothing unreasonable in FERC's demand that proponents furnish the proposed alternatives with adequate support.

It is quite true that FERC should have said more to explain why it concluded that "the record in this proceeding does not contain adequate information to support a just and reasonable ICAP deficiency charge based on prices in the bilateral market." ISO New Eng., 94 FERC ¶ 61,237, at 61,847. This is so even if we assume, with FERC, that petitioners could have pointed to such alternatives before the December 15 order, knowing that protests to the earlier compliance filing had urged that the $8.75 charge be re-adopted. But FERC did address the main alternative and found it wanting even if FERC's explanation was too cryptic; the alternative was not rejected simply because it was belatedly offered.

In all events, on remand FERC should explain why, on this record, the bilateral-market prices are not an adequately supported substitute. If other substantial alternatives were presented, FERC should also address them. But we harbor some doubts about the strength of petitioners' claim as to other alternatives; it is hard in their briefs to find any substantial discussion of what the alternatives were and why (absent the $8.75 charge) they might be expected effectively to enforce the reserve obligations that already exist.

The $0.17 Charge. This brings us to petitioners' next objection, which is superficially quite different because it focuses on the rejection of the $0.17 charge rather than the adoption of the $8.75 charge. Although petitioners do not contest FERC's general authority to review the $0.17 charge, 16 U.S.C. §§ 824d, 824e (1994), they say that under established case law, FERC had no power to "reject" ISO-NE's $0.17 proposal without a "hearing" because it was not procedurally deficient or substantively a "nullity," Mun. Light Bds. v. FPC, 450 F.2d 1341, 1345 (D.C. Cir. 1971), cert. denied, 405 U.S. 989, 92 S.Ct. 1251, 31 L.Ed.2d 455 (1972). Petitioners also say that the reasons given by FERC for the rejection were unsound or unsupported.

There is a body of case law, not entirely consistent in rhetoric or substance, that limits the power of federal ratemaking agencies like FERC summarily to reject new tariffs except as to narrow grounds. The case law is a gloss on statutory procedures designed to allow carriers to institute rate changes on their own initiative and, subject both to temporary delay or "suspension" and to refund of later-overturned obligations, to place new rates into effect during agency investigation and hearing. E.g., AT T Co. v. FCC, 487 F.2d 865, 871-72 (2d Cir. 1973). But the $0.17 charge was not just a typical utility-initiated rate charge but was purportedly filed in compliance with FERC's June 28 order.

Here, FERC says that the $0.17 rate was inconsistent with its June 28 order and that that order cannot be attacked in this proceeding. The two propositions are quite different. We have already explained why the premises of the June 28 order cannot be insulated from criticism insofar as those premises are used to support FERC's later actions. However, the fact that the June 28 order is still before FERC on rehearing petitions does not prevent FERC from relying upon it in evaluating a purported compliance filing, which is just what the $0.17 proposal constituted. The June 28 order was effective unless stayed, and no stay was secured either from FERC or from a reviewing court.

According to FERC's December 15 order rejecting the proposal, the $0.17 charge was a nominal rather than a plausible response to the June 28 directive. If FERC was correct, it was entitled to reject the filing as patently inconsistent with an existing, unstayed directive. "[A] utility is not entitled to a hearing before the non-conforming portion of its rate filing is rejected or when it challenges an established policy," Jersey Cent. Power Light Co. v. FERC, 810 F.2d 1168, 1205 (D.C. Cir. 1987) (internal citations omitted); see also ANR Pipeline Co. v. FERC, 931 F.2d 88, 92 (D.C. Cir. 1991). FERC certainly did make a determination of non-compliance, and in this determination it enjoys a favorable standard of review, Pepperell Assocs. v. U.S. EPA, 246 F.3d 15, 22 (1st Cir. 2001). Whether FERC adequately explained the determination is a different question.

In critiquing the determination, petitioners focus on a terse and arguably confusing statement by FERC that the $0.17 was merely an average derived from the (now abandoned) auction market and was not "useful." Yet, fairly read, FERC's June 28 order clearly envisaged a return to a substantial ICAP charge; and the core concept of such a charge has long been in New England, and still is in other Atlantic Coast regions, the cost of adding new peak load generating capacity (not excluding a possible penalty as well). That cost is far above $0.17 and probably lies in the $5.00 range. On this premise, the $0.17 proposal was plainly non-complying, and FERC's rejection of the $0.17 proposal was therefore well justified.

Procedural Matters. Petitioners' final argument on review brings them back to FERC's adoption of the $8.75 charge. In adopting this charge, say petitioners, FERC "failed to provide due process and unreasonably refused to hold a hearing." Two quite different arguments are presented under this heading: first, that FERC was required to hold not just a hearing (which it did) but a specific kind of hearing before adopting the $8.75 charge; and second, that because selling utilities first proposed the $8.75 charge in protesting the $0.17 filing, the buying utilities had no fair opportunity to respond since "[u]nder FERC's rules, answers to protests are not permitted."

The term "hearing" is notoriously malleable, but what petitioners got here was not only a hearing but a species of evidentiary hearing which is now quite common in utility and carrier regulation, e.g., Town of Norwood, 202 F.3d at 404. Very extensive evidentiary submissions were made by both sides in the form of affidavits from experts and others, together with extensive written argument; indeed, it is these submissions by buying utilities that petitioners complain (rightly in some degree) were before FERC but not seriously addressed in its orders.

At least where forward-looking industry-wide regulation is at issue, it is increasingly common for agencies to employ such hearings by affidavit and nothing more. In such situations, courts have approved of this agency practice when any genuine issues of material fact can be "adequately resolved on the written record." E.g., Town of Norwood, 202 F.3d at 404; La. Energy Power Auth. v. FERC, 141 F.3d 364, 371 (D.C. Cir. 1998); see also La. Ass'n of Indep. Producers Royalty Owners v. FERC, 958 F.2d 1101, 1113 (D.C. Cir. 1992) (trial-type evidentiary hearing not necessary to answer "purely technical issue" "whether additional pipeline capacity is needed to meet future demand" (internal quotation marks omitted)).

Of course, an agency can do more: it can afford oral hearings in which cross-examination is conducted before an administrative law judge; in utility regulation, direct testimony is commonly offered (as here) in written form. Whether and when particular procedures are necessary is no longer governed by clear-cut judicial rules. It is often said that an agency's decisions as to procedure are reviewed for abuse of discretion, and, again, the reasons for deference are especially strong where the decision is entangled with the agency's expert judgment regarding forward-looking industry-wide regulation, see, e.g., Fresno Mobile Radio, Inc. v. FCC, 165 F.3d 965, 971 (D.C. Cir. 1999).

Turning to the question whether buying utilities had a fair opportunity to respond to the prospect of an $8.75 charge, there are two additional points. First, in 77 pages of comments filed with FERC on August 28, 2000 (the very same day that sellers filed protests arguing for the $8.75 charge), buying utilities specifically anticipated, and responded extensively to, arguments for the $8.75 charge. Thus, under the particular circumstances of this case, FERC's general rule against answers to protests, 18 C.F.R. § 385.213(a)(2) (2000), did not prevent buying utilities from presenting the principal arguments against the $8.75 charge.

A second, more decisive point is that, whatever the adequacy of the "hearing" provided before December 15, opponents to the $8.75 charge had ample opportunity to contest the $8.75 charge in their requests for rehearing. See Boston Edison Co. v. FERC, 885 F.2d 962, 966 (1st Cir. 1989). As noted above, FERC's March 6 order considered the various arguments against the $8.75 charge and responded to them. The main question remains not whether petitioners' arguments were heard, but what FERC should have done to respond to them adequately.

In all events, on remand petitioners are free to ask FERC for the opportunity to make further submissions and to show FERC why oral cross-examination is necessary. As we are structuring our remand, petitioners will have no incentive for delay and good reason for expedition; they can then consider again whether they think oral cross-examination is vital. And something may depend on how FERC proposes to answer the objections made to its contemplated $8.75 charge and just how much of an interim measure it intends the $8.75 charge to be.

Remand. We conclude that certain specific issues raised by opponents of the $8.75 charge require more reasoned consideration than FERC afforded. The results of FERC's orders may well be defensible, given FERC's view that immediate action is required and that the charge can be reconsidered if opponents provide an adequate alternative regime. However, the immediate impact of those orders is high and FERC's errors and omissions are troubling. Thus, a remand is appropriate for further explanation. NorAm Gas, 148 F.3d at 1165-66.

The principal questions that FERC needs to answer more fully are these: why, despite petitioners' various claims to the contrary, a substantial ICAP charge is still required to enforce reserve obligations; why, in light of petitioners' claims of a lower present cost of peaking capacity, $8.75 is the proper interim figure; and why any alternatives already proffered by opponents are inadequate or are otherwise not properly considered at this time. Answers can be imagined, but it is FERC that must formulate and adopt them in the first instance. SEC v. Chenery Corp., 332 U.S. 194, 196, 67 S.Ct. 1575, 91 L.Ed. 1995 (1947).

A reviewing court that perceives flaws in an agency's explanation is not required automatically to set aside the inadequately explained order. Allied-Signal, Inc. v. U.S. Nuclear Regulatory Comm'n, 988 F.2d 146, 150-51 (D.C. Cir. 1993). Whether to do so rests in the sound discretion of the reviewing court; and it depends inter alia on the severity of the errors, the likelihood that they can be mended without altering the order, and on the balance of equities and public interest considerations. Int'l Union, United Mine Workers of Am. v. Fed. Mine Safety Health Admin., 920 F.2d 960, 966-67 (D.C. Cir. 1990). Here, a preliminary assessment suggests that the errors at issue can probably be mended.

Further, FERC's warrant that the charge is needed now to assure adequate energy supplies in New England carries weight. FERC has plausibly adverted to the need for confidence among power suppliers that reserve requirements will be meaningfully enforced. Cf. Fresno Mobile Radio, 165 F.3d at 971. An on again-off again ICAP charge is not likely to encourage suppliers to maintain marginal ( i.e., high cost) existing plant or to build new facilities for peak demand. At least at this time, we think that the public interest in assuring power is decisive.

In remanding, we leave open to FERC's informed judgment the decision whether to conduct further proceedings (and, if so, what kind) or whether simply to write a further decision on reconsideration. Nor do we preclude FERC from modifying the outcome if it is so advised. No time limit need be imposed at this time. If FERC unduly delays, any party to this case may apply to us for an order fixing a deadline for agency reconsideration. Cf. Int'l Union, 920 F.2d at 967.

We retain jurisdiction over this case to issue all orders necessary to assure compliance with our mandate and to review whatever decision FERC makes on reconsideration in response to our mandate, assuming that the decision remains contested. Cf. BASF Wyandotte Corp. v. Costle, 598 F.2d 637, 663 (1st Cir. 1979); Kennecott Copper Corp. v. EPA, 462 F.2d 846, 851 n. 21 (D.C. Cir. 1972). FERC's counsel is directed to file a status report, served on all other parties, with the Clerk of this court every 45 days from the date of this decision.

Accordingly, FERC's orders of December 15, 2000, and March 6, 2001, are not vacated at this time but the case is remanded for further action consistent with this decision. Parties may petition for rehearing of this court's decision in the ordinary course. However, our prior stay order barring the $8.75 charge is vacated forthwith, and FERC is free to re-impose its $8.75 charge prospectively, either at once or as of some specified future date. All parties shall bear their own costs in this court.

It is so ordered.