concurring in part and dissenting in part:
For procedural and substantive reasons, I respectfully dissent from the majority’s holding on Farmer’s § 3729(a)(2) claim.1 The parties argued this claim primarily (and before this court, exclusively) in terms of whether § 3729(a)(2) implies a presentment requirement. The majority bypasses both the original source and presentment issues and instead affirms summary judgment based on an undeveloped argument that there was no “knowing” misuse of federal funds in this case. Though the majority fails to make this clear, only one of the defendants raised this “scienter” issue below, the district court did not rule on it and neither defendant has argued it to this court. The scienter issue has not been sufficiently preserved, or the record adequately devel*345oped, to justify the majority’s decision on this point. Worse still, the evidence of record strongly suggests that someone treated the federal repair funds as a honey pot rather than a public trust. Yet the majority goes out of its way to minimize palpably suspicious circumstances.
1. The majority’s assertion that we “may affirm the district court’s judgment on any grounds supported by the record” is an oversimplification of the rule. Both the nature of appellate review and this circuit’s precedent require a more cautious approach. As this court explained en banc in United States v. Brace, “we are a court of review, not of original error .... we do not craft new issues or otherwise search for them in the record.” 145 F.3d 247, 255-56 (5th Cir.1998). Rather, “[i]t is for the parties, those who have a stake in the litigation, to decide which issues they want to pursue at trial and on appeal.” Id. at 256.
The defendants in this case decided how best to carry their respective burdens in pursuit of summary judgment and success on appeal, but it turns out those tactical decisions are of little consequence. The majority affirms summary judgment for both defendants on grounds that the City never raised at any stage, and neither defendant pursued before this Court.2 HAUL did raise the issue below, but prevailed on the presentment issue and chose to place all its eggs in the same basket on appeal. To affirm summary judgment for HAUL on grounds that it has abandoned may be justifiable in theory, but to extend that benefit to the City is nothing short of a windfall, a sua sponte summary judgment on appeal.
Generally, this court may affirm summary judgment on grounds not relied on by the district court. But this is not a complete statement of the rule; in most cases we require, at a minimum, that those grounds be raised before the district court. Leverette v. Louisville Ladder Co., 183 F.3d 339, 342 (5th Cir.1999) (“This Court will not consider an issue that a party fails to raise in the district court absent extraordinary circumstances.”).3 As Judge Garwood carefully explained in FDIC v. Laguarta:
This Court has clearly held, however, that it will generally not consider a new ground on appeal raised by an appellant in opposition to summary judgment. The same should apply to new grounds raised by an appellee in defense of summary judgment where the parties were not afforded an opportunity to develop the issue below, and it was not implicit or included in the issues or evidence tendered below, so that the party was not on notice of the need to meet it, and the record appears not to be adequately developed in that respect.
939 F.2d 1231, 1240 (5th Cir.1991) (internal citations omitted). As Judge Garwood acknowledged, there may be exceptions to this rule, “if the issues were implicit or included in those raised below or the evidence in support thereof, or if the record *346appears to be adequately developed in respect thereto.” Id. This recognized exception was properly invoked by the court in FDIC v. Lee, 130 F.3d 1139, 1142 (5th Cir.1997). Lee, after a thorough analysis, affirmed summary judgment on a purely legal ground that was raised in the Complaint, was uncomplicated, and had been completely developed for appellate review. Id. at 1142-43.
This case is not like Lee: first, the “knowingness” issue has not been fully developed and is not amenable to simple resolution on appeal. The parties spent no time arguing scienter to this Court, and as demonstrated infra, the majority’s unassisted review of a cold record is woefully incomplete. The City, having failed to put Farmer on notice that it was attacking the scienter element of her claim, is simply not entitled to summary judgment on the issue. Nor am I persuaded that a judgment in favor of HAUL on reckless disregard exonerates the City with regard to its own responsibilities in administering the EHRP. The City — which should bear the burden on its own motion for summary judgment — chose not to raise the issue. Yet the majority, without explaining what authority it has to do so, simply “cuts the City in” on HAUL’S victory.4
Further, unlike Lee’s purely legal question of statutory construction, the scienter finding in this ease turns on a morass of factual distinctions that — as the majority’s “on-the-one-hand, on-the-other-hand” analysis demonstrates — are far from clear-cut. Perhaps this sprawling factual determination is an exception to the rule that we limit ourselves to the issues the parties have raised on appeal. But the majority’s failure properly to articulate that rule, much less explain how an exception applies, serves only to confuse the law. This decision does not do justice to the governing legal standard or to the parties before this court.
2. In addition to these procedural objections, I dissent from the majority’s substantive resolution of the scienter question. To resolve this fact issue as a matter of law, a court must reach a conclusion that is beyond dispute by all reasonable jurors. The majority falls short of this standard with its protracted and unusual — yet ostensibly inescapable — weighing and reweighing of a laundry list of suspicious transactions that HAUL and the City approved. The majority’s account of what a jury would “expect” to find in the evidence, or what sort of “thinking would drive the jury’s assessment” is not only unpersuasive, but highlights that the majority is performing its own fact-finding. The majority also tips its hand with a lengthy prognostication as to what facts a jury “would put little weight on,” or what a jury would be “less likely to infer” or “less likely to conclude.” I have never seen an appellate opinion that purported to resolve a factual question like this, on a record which raises so many questions, in a patently “juri-fied” way. I cannot agree that all reasonable jurors would have no choice but to follow the majority’s serpentine course through the evidence and reach its conclusion.
My disagreement is sharpened by the majority’s gross understatement of the evidence Farmer presented on summary judgment. HUD did not simply conclude, *347as the majority states, that “many” third party contracts lacked an acceptable cost breakdown. Rather, the HUD inspector chose 62 files at random and found that “none of the contracts contained an acceptable cost breakdown.” R. 1753, 1788 (emphasis added). The majority makes it look as if there were problems in only a handful of cases — all with reasonable, legitimate explanations. To the contrary, the evidence indicates that, program-wide, the EHRP was unsupervised, rife with fraud, and effectively unreviewable due to an obvious lack of fundamental documentation. Based on 62 sample files and 26 on-site visits, HUD made the following observations that the majority chooses to overlook:
• “Across the board,” the work write-ups were “too general and loosely written to allow a reviewer to conclude that equipment and/or materials are acceptable, quantities are accurate, and costs are reasonable.” R. 1788.
• “None of the files reviewed contained a Scope of Work to describe what was to be done and how it would be done .... This information is critical in estimating costs.” R. 1788 (emphasis added).
• “Change orders” — alterations to a bid after a contract was signed — did not show justification for the changes. R. 1788.
• None of the files contained state-required forms for termite inspection and treatment, even though this was listed as a cost item. R. 1788.
• “All files” contained too many standalone and lump sum estimates without a reasonable explanation. “None of the contracts contained an acceptable cost breakdown .... it appears that contractors were just assigning prices to accommodate the amount of the grant.” R. 1788 (emphasis added).
• “Across the board,” on-site visits revealed “huge differences” between the estimates of materials needed and the actual materials applied, as well as “obvious evidence of poor on-site supervision and inspections.” R. 1789.
• On-site visits verified that “on numerous occasions contractors collected payments for units of materials that were not applied.” R. 1789.
• There was “no evidence that the contractors made any attempt to adjust costs to correct excessive estimates.” R. 1789 (emphasis added).
• In light of these conspicuous overcharges, the HUD inspector observed that “there is no evidence that the subrecipients managing the program or the City of Houston questioned the contractors’ actions and overpay-ments — they failed to exercise any quality control.” R. 1789 (emphasis added).
This is only a sample. Farmer submitted over 500 pages of exhibits including reports, documents, and photographs that depict blatant misrepresentations and obviously substandard work. On two occasions it appears EHRP funds were used to make down payments on newly-constructed homes — an admittedly impermissible expenditure. The City denies this took place (R. 2303, 2306), but the record contains a check to Habitat for Humanity in the amount of $20,000 (R. 1941); a $20,000 invoice from Habitat for Humanity reading “downpayment assistance for a home” (R. 1943); a request for payment detailing another $15,000 check to Habitat for Humanity (R. 2017); and a note from a City inspector on this second property that reads “Upon review of the file it was noticed that the homeowner received $15,000 in down payment .... ” R. 2276. Farmer also presented evidence that $6,700 in program funds were spent to fill in a swimming pool — another facially impermissible *348expenditure that the City denies. (R. 2304). Yet the record contains a request for payment to “ABC Drainage” in this amount (R. 2110), and the City’s re-inspection report (the very basis for its denial) details this expense as “swimming pool fill-in.” R. 2279.
In light of the evidence in the record, the majority’s analysis is not only unwieldy, it is untenable. The record belies the majority’s assertion that it would have been “relatively difficult” for the defendants to recognize these overcharges. Rather, the evidence suggests the defendants may have turned a blind eye to patently false submissions, and, at best, “failed to exercise any quality control at all.” As more than one court has noted, the FCA reckless disregard standard “is intended to reach the ‘ostrich-with-its-head-in-the-sand’ problem where government contractors hide behind the fact that they were not personally aware that such overcharges may have occurred.”5
If the majority is holding out for a “smoking gun” admission of fraudulent intent, this would vitiate the statute’s provision that “no proof of specific intent to defraud is required.” 31 U.S.C. § 3729(b). At a minimum, the FCA requires reckless disregard of truth or falsity. As noted in one opinion cited by the majority, “failure to make a minimal examination of records constitutes deliberate ignorance or reckless disregard.” UMC Elecs. Co. v. United States, 43 Fed.Cl. 776, 794 (1999) (“It is apparent that this reckless disregard standard prevents defendants from simply pointing to confusion over invoices and billing records as a complete defense.”). There is enough in this record to allow a jury to determine whether something worse than gross negligence led the City and HAUL to approve so many obviously improper expenditures. There is certainly much more in the record than the majority has accounted for. Given the evidence of clear and pervasive fraud in this case, the majority’s finding that the defendants cannot be found culpable of anything worse than “mismanagement” is hardly credible, much less beyond reasonable debate.
I respectfully dissent.
. I concur in the majority’s judgment as to Farmer’s conspiracy claim under § 3729(a)(3). Farmer has not produced sufficient evidence showing that both defendants agreed on a plan to defraud the government.
. The majority’s representation that "defendants — primarily HAUL — argued” the lack of scienter below is simply incorrect. One defendant, HAUL, raised this issue below. The City never did.
. Even in opinions that recite the majority's abbreviated version of this rule, there is the implication that the district court at least had the opportunity to pass judgment on the issue. See, e.g., Mangaroo v. Nelson, 864 F.2d 1202, 1204 n. 2 (5th Cir.1989) (Smith, J.) (stating that the court may affirm for any grounds in the record, "even if those grounds were rejected by the trial court”); Hoyt R. Matise Co. v. Zurn, 754 F.2d 560, 566 (5th Cir.1985) (stating that the court may affirm on any ground, "including one rejected or ignored in the lower court”).
. This is even more unexpected in light of the City’s argument in its motion to dismiss that Farmer was not an “original source” because HUD was already "squarely on the 'trail of fraud' against [HAUL].” The City never argued this wrongdoing would have been difficult to detect; rather, it submitted evidence showing the "very same types of fraud” Farmer identified were readily apparent to the HUD inspectors who were looking into the EHRP.
. United States v. Krizek, 111 F.3d 934, 942 (D.C.Cir.1997) (quoting 132 Cong. Rec. H9382-03 (daily ed. Oct. 7, 1986) (statement of Rep. Berman)); see also UMC Elecs. Co. v. United States, 43 Fed.Cl. 776, 794 (1999) (similar).