MICHELIN v. WARDEN MOSHANNON VALLEY CORRECTIONAL CENTER 24 2990 (2026)

SUR PETITION FOR REHEARING

The petitions for rehearing filed by the Appellants, Warden Moshannon Valley Correctional Center, et al., in the above-entitled cases having been submitted to the judges who participated in the decision of this Court and to all the other available circuit judges of the circuit in regular active service, and no judge who concurred in the decision having asked for rehearing, and a majority of the judges of the circuit in regular service not having voted for rehearing, the petitions for rehearing by panel and the Court en banc, are denied.¹

The panel's decision in these cases requires taxpayers to help fund aliens' efforts to prolong their stay in this Country despite orders of removal. The panel held that the sovereign-immunity waiver in the Equal Access to Justice Act (EAJA) permitted attorney-fee awards to Petitioners because the term “any civil action,” 28 U.S.C. § 2412(d)(1)(A), unambiguously included their habeas petitions. See Michelin v. Warden Moshannon Valley Corr. Ctr., ––– F.4th ––––, ––––, 2026 WL 263483, at *3 (3d Cir. 2026).¹ Although the differing interpretations of the EAJA that have surfaced around the Country may ultimately require a nationwide solution, we should have reheard these cases en banc because the question is one of exceptional importance and required our Court to choose sides in a deep Circuit split.

Congress must use clear and unambiguous language to waive the government's sovereign immunity. My disagreement with the outcome in these cases has roots in a simple question from the panel's thorough and thoughtful opinion: “In isolation, almost anything can be ambiguous. ‘Mary had a little lamb’ seems clear. But did Mary own a baby sheep or eat a piece of one for dinner?” Michelin, ––– F.4th at –––– n.7, 2026 WL 263483, at *7 n.7. The question had not previously crossed my mind, but I knew the answer immediately. Context from the rest of the poem makes the answer clear. See Sarah Josepha Hale, Poems For Our Children 6-7 (Marsh, Capen & Lyon ed. 1830). I do not know if Mary was a vegetarian, but I know she had at least one pet. In my mind, there is no ambiguity worth mentioning there. The government's interpretation of the EAJA in these cases does not rest on anything that even approaches the type of manufactured hyper-textualist ambiguity suggested by the panel's question. Instead, there is a real question, based on text, context, case law, and history, regarding whether the statutory language covers these habeas petitions.

In fact, there is significant tension between the panel's holding that “any civil action” unambiguously includes habeas and our en banc decision in United States v. Bendolph, which flatly rejected the “premise that habeas cases and ordinary civil cases are indistinguishable.” 409 F.3d 155, 166 (3d Cir. 2005). There are several cases from the Supreme Court and our Court with similar reasoning. There are also cases, cited by the panel, that characterize habeas as civil. The juxtaposition of these authorities demonstrates that § 2412(d)(1)(A) does not meet the “demanding” standard for a waiver because “there is a plausible interpretation of the statute that preserves sovereign immunity.” Lac du Flambeau Band of Lake Superior Chippewa Indians v. Coughlin, 599 U.S. 382, 388, 143 S.Ct. 1689, 216 L.Ed.2d 342 (2023).

One alternative interpretation, which I favor, is that “civil action” in the EAJA means the same thing as “civil action” in the Federal Rules of Civil Procedure. The phrase is a term of art assigned a fixed meaning by Congress in the Rules Enabling Act, and then defined in the Civil Rules. Congress transplanted the phrase into Title 28 in connection with the positive-law codification process—more than 30 times—accompanied by Reviser's Notes citing the definition in the Civil Rules. Habeas proceedings are civil in nature, but they are not “civil actions” under the Civil Rules or the EAJA. Therefore, as discussed in Part I, the government did not waive sovereign immunity in connection with Petitioners' applications for attorneys' fees.

The fee awards in these cases were inappropriate for an additional reason. The government's positions regarding Petitioners' detention were “substantially justified” and “special circumstances” made the fee awards “unjust.” 28 U.S.C. § 2412(d)(1)(A). The habeas petitions at issue relied on case law establishing a due process balancing test based on a non-exhaustive list of factors. The main cases creating that test involved lawful permanent residents who had developed connections to this Country—and, perhaps, corresponding due process protections—that Petitioners lack. When we create this kind of amorphous standard, it is to be expected that there will be times when a judge sees that balancing differently than the government. Then, the government loses. What is atypical, and what I do not think the record supported here, is for a court to find that the government's balancing is so lacking in justification that a fee award is warranted.

Take, for example, the case of Petitioner Adolph Michelin. Michelin overstayed a tourist visa issued to him in 2010, and he was later convicted of a marijuana offense as well as resisting an officer in a separate incident. While remaining in the United States without authorization, he was arrested several other times for offenses relating to narcotics, aggravated assault with a firearm, and aggravated battery on a pregnant woman. After DHS entered a final order of removal in 2018, but bailed Michelin while he challenged the order, he absconded. Following another narcotics arrest in 2021, and facing removal on March 31, 2022, the BIA granted Michelin's March 25, 2022 motion to reopen but decided to keep him detained during that process. A magistrate judge granted habeas relief, holding that this detention violated Michelin's due process rights because it was unreasonable for DHS to presume that he presented a flight risk and danger to the community. I do not see it that way for an alien with Michelin's history, lack of ties to the United States, rap sheet, and a documented instance of flight, but that was not my call to make. The magistrate judge later directed the government to pay Michelin's attorneys' fees relating to that litigation. I believe that was an abuse of discretion that led to an outcome Congress could not have intended.

The government pressed this issue on appeal with respect to the second Petitioner, Adewumi Abioye. Abioye also overstayed his visa, appears to have attempted to obtain citizenship through marriage fraud, and perpetrated a massive financial fraud in this Country that caused $4.6 million in losses to more than 10 victims. He admitted that $1.5 million of the fraud proceeds went to accounts he controlled. Abioye had not completed the supervised-release component of his sentence at the time he argued in habeas proceedings that his continued detention was unreasonable, and his sentence also included orders to pay millions of dollars in forfeiture and restitution. In effect, the fee award at issue directed the government to pay him instead. In Part II, I explain why that ruling was an abuse of discretion.

I.

We are now on the wrong side of a Circuit split regarding the government's sovereign immunity from EAJA fee awards in habeas proceedings initiated by aliens. The Fourth and Fifth Circuits have held, correctly in my view, that a § 2241 petition by an alien is not a “civil action” under the EAJA. See Barco v. Witte, 65 F.4th 782, 785 (5th Cir. 2023); Obando-Segura v. Garland, 999 F.3d 190, 195 (4th Cir. 2021).² Those cases emphasize authorities regarding the “hybrid” nature of habeas. 65 F.4th at 785; 999 F.3d at 195. I outline these and other supporting authorities in Part A.

On the other side of the split, the panel joined the Second, Ninth, and Tenth Circuits in holding that an alien's habeas petition is a “civil action.” See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *10; Daley v. Ceja, 158 F.4th 1152, 1164 (10th Cir. 2025); Vacchio v. Ashcroft, 404 F.3d 663, 668-69 (2d Cir. 2005); In re Hill, 775 F.2d 1037, 1040-41 (9th Cir. 1985). The panel rightly distanced itself from the Second and Ninth Circuits. See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *10. The Second Circuit conceded “ambiguity” with respect the question presented, 404 F.3d at 669, which is correct but means the government is entitled to sovereign immunity. See United States v. Williams, 514 U.S. 527, 531, 115 S.Ct. 1611, 131 L.Ed.2d 608 (1995). Both the Second and Ninth Circuits relied on legislative history to find a sovereign-immunity waiver not clearly evident from the statutory text, which is impermissible for reasons the panel explained. See 404 F.3d at 669-70; 775 F.2d at 1040-41; Michelin, ––– F.4th at –––– n.3, 2026 WL 263483, at *3 n.3. The panel instead aligned with the Tenth Circuit based in large part on an “evaluation of the historic civil status of habeas actions.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *10 (citing Daley, 158 F.4th at 1164). My disagreement with that historical evaluation, including the application of the “old-soil” principle and discussion of other pre-and post-enactment history, is set forth in Parts B through E.

A.

“The United States enjoys sovereign immunity and cannot be sued without its consent.” United States Postal Service v. Konan, 607 U.S. ––––, ––––, ––– S.Ct. ––––, ––– L.Ed.2d ––––, 2026 WL 501765, at *2 (2026). Sovereign immunity “ensures that elected officials, not judges, choose when to open the public purse.” Doe 1 v. United States, 37 F.4th 84, 88 (3d Cir. 2022). There is no dispute that Congress waived the federal government's sovereign immunity in “any civil action” when it passed the EAJA in 1980. 28 U.S.C. § 2412(d)(1)(A); see also Ardestani v. INS, 502 U.S. 129, 138, 112 S.Ct. 515, 116 L.Ed.2d 496 (1991). These cases are about the scope of that waiver. The question is whether the government has waived sovereign immunity with respect to Petitioners' § 2241 applications.

“[W]e must construe any ambiguities in the scope of a waiver in favor of the sovereign.” United States v. Miller, 604 U.S. 518, 532, 145 S.Ct. 839, 221 L.Ed.2d 373 (2025). “Importantly, even when there is a statutory waiver of immunity, we should not take it upon ourselves to extend the waiver beyond that which Congress intended.” Giovanni v. United States Dep't of Navy, 906 F.3d 94, 118 (3d Cir. 2018). “It is the province of Congress, not the courts, to rewrite the statute to include proceedings that are not clearly within its scope.” Clarke v. INS, 904 F.2d 172, 178 (3d Cir. 1990). Thus, courts usually give the government a wide berth in this space. See, e.g., Libr. of Cong. v. Shaw, 478 U.S. 310, 318, 106 S.Ct. 2957, 92 L.Ed.2d 250 (1986) (holding that sovereign immunity barred recovery of interest despite statute “making the United States liable ‘the same as a private person’ for ‘costs,’ including ‘a reasonable attorney's fee’ ”); Cudjoe ex rel. Cudjoe v. Dep't of Veterans Affs., 426 F.3d 241, 244, 247-48 (3d Cir. 2005) (holding that sovereign immunity barred private lawsuit despite statute subjecting government to “substantive and procedural requirements” including “all civil and administrative penalties and fines”).

The panel conceded as much and noted, correctly, that “if the phrase ‘any civil action’ is ambiguous between an interpretation that reaches habeas actions and one that does not, then we must construe the scope of the waiver to exclude them.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *3. The panel concluded, however, that the EAJA “cannot plausibly be read” to exclude habeas from the term “any civil action.” Id. Suggesting otherwise is the fact that 17 Circuit judges, including a member of the panel, previously reached the opposite conclusion in a variety of settings. See Barco, 65 F.4th at 785; Obando-Segura, 999 F.3d at 195; Daley v. BOP, 199 F. App'x 119, 121 (3d Cir. 2006); O'Brien, 395 F.3d at 507; Ewing, 826 F.2d at 971; Boudin, 732 F.2d at 1111-15. While the weight of this authority is not dispositive, the fact that there are opinions dating back over 30 years supporting the government's sovereign immunity in this context is, at minimum, suggestive of a lack of clarity about what “any civil action” means in § 2412(d)(1)(A).

To evade those cases, the panel treated as roughly equivalent terms like “civil in nature,” “civil proceedings,” “civil matter,” and “civil status.” See, e.g., Michelin, ––– F.4th at –––– – ––––, 2026 WL 263483, at *7-10. Too much of a stretch for me. Sovereign immunity is appropriate where “Congress could have more clearly authorized recovery.” In re Lansaw, 853 F.3d 657, 667 n.8 (3d Cir. 2017). That is true of the provision at issue, which is apparent upon examination of broader phrasing in different sections within the same Title of the U.S. Code. See Babcock v. Kijakazi, 595 U.S. 77, 83, 142 S.Ct. 641, 211 L.Ed.2d 424 (2022) (“This statute's plain meaning becomes even more apparent when viewed in the broader statutory context.”). In contrast to “any civil action,” 28 U.S.C. § 2412(d)(1)(A), Congress has used broader phrases like “an action,” id. § 1292(d)(4)(A)-(B); “any case,” id. § 1292(c)(1); “any action or proceeding,” id. § 1355(a); “any action brought in a court of the United States or of a State,” id. § 1610(d); “any suit, action or proceeding,” id. § 1915(a)(1); “any action, suit or proceeding,” id. §§ 254, 1404(b), 2403(a); and “an action, suit or proceeding,” id. § 2107(a). Outside of Title 28, Congress used the term “suits of a civil nature.” 12 U.S.C. §§ 1789(a)(2), 1819(b)(2)(A), 2277a-7(4)(B); 47 U.S.C. § 33. These phrases are not synonymous. Many of them were on the books when the EAJA was passed, which suggests that Congress was aware of its options at that time. “We usually presume differences in language like this convey differences in meaning.” Wis. Cent. Ltd. v. United States, 585 U.S. 274, 279, 138 S.Ct. 2067, 201 L.Ed.2d 490 (2018). “Any civil action” is narrower than the words Congress chose in other areas that would have more easily reached habeas.

Modern § 2241 adjudications can be civil “in nature” with civil “characteristics,” as the panel demonstrated, without being EAJA “civil actions.” Using a civil “label” to characterize adjudication of a habeas petition is “gross and inexact.” Harris v. Nelson, 394 U.S. 286, 293-94, 89 S.Ct. 1082, 22 L.Ed.2d 281 (1969); see also Brown v. Vasquez, 952 F.2d 1164, 1169 (9th Cir. 1991) (“While it is true that a habeas corpus proceeding is civil in nature, it is equally true that such a proceeding is dramatically different from any other type of civil action.”). Habeas proceedings are “elastic” and “fluid[ ].” Price v. Johnston, 334 U.S. 266, 283, 68 S.Ct. 1049, 92 L.Ed. 1356 (1948). This makes habeas “unique.” Harris, 394 U.S. at 294, 89 S.Ct. 1082. Thus, as the panel observed, courts have repeatedly referred to habeas as “hybrid.” See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *6; see also, e.g., Parrott v. Gov't of V.I., 230 F.3d 615, 620 n.7 (3d Cir. 2000); Hassine v. Zimmerman, 160 F.3d 941, 954 (3d Cir. 1998).

The panel sought to distinguish some of these cases by arguing that they were “hybrids of civil actions and criminal ones.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *5. This is a distinction without a difference. “There are not ‘criminal habeas writs’ and ‘non-criminal habeas writs’: there are just writs for habeas corpus.” Obando-Segura, 999 F.3d at 194. Several of the “hybrid” cases cited by the panel involved challenges to sentencing computations or conditions of confinement. See O'Brien, 395 F.3d at 501; Sloan, 351 F.3d at 1321; Ewing, 826 F.2d at 975; Boudin, 732 F.2d at 1109-10. Those petitioners were essentially challenging agency action by the responsible custodian, such as the Bureau of Prisons, which is an objective that lines up well with the relief Petitioners sought here. And the fact that immigration proceedings are civil in nature does not mean that we are dealing with a situation where a “hybrid of a civil action and a civil action is a civil action” under the EAJA. Michelin, ––– F.4th at ––––, 2026 WL 263483, at *5. Immigration law has its own “specialized administrative procedure,” and immigration proceedings are not covered by the EAJA. Marcello v. Bonds, 349 U.S. 302, 308, 75 S.Ct. 757, 99 L.Ed. 1107 (1955); see also Ardestani, 502 U.S. at 138-39, 112 S.Ct. 515. A hybrid between two specialized proceedings—immigration and habeas—does not transform the latter into enough of a “civil action” to require a sovereign-immunity waiver.

The panel also drew false equivalence between habeas and “civil actions” by contending that the Federal Rules of Civil Procedure “generally presumed habeas actions were civil actions, even if they were also distinctive ones.” See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *4. That is not correct. See, e.g., Bracy v. Gramley, 520 U.S. 899, 904, 117 S.Ct. 1793, 138 L.Ed.2d 97 (1997) (distinguishing between a “habeas petitioner” and “the usual civil litigant in federal court” for purposes of discovery); Hilton v. Braunskill, 481 U.S. 770, 776 n.5, 107 S.Ct. 2113, 95 L.Ed.2d 724 (1987) (acknowledging “differences between general civil litigation and habeas corpus proceedings”); Preiser v. Rodriguez, 411 U.S. 475, 495-96, 93 S.Ct. 1827, 36 L.Ed.2d 439 (1973) (distinguishing between “swift, flexible, and summary determination[s]” in habeas proceedings and “an original plenary civil action [under § 1983], governed by the full panoply of the Federal Rules of Civil Procedure”).

A habeas proceeding is not a “civil action” under the Civil Rules, and I do not see any presumption to that effect. To illustrate the point, “[l]et's count some of the ways in which habeas proceedings deviate from the Civil Rules.” Banister v. Davis, 590 U.S. 504, 528, 140 S.Ct. 1698, 207 L.Ed.2d 58 (2020) (Alito, J., dissenting). “There shall be one form of action to be known as ‘civil action.’ ” Fed. R. Civ. P. 2 (1980). “A civil action is commenced by filing a complaint with the court,” and “[n]o other pleading shall be allowed.” Fed. R. Civ. P. 3, 7(a) (1980); see also In re Teter, 90 F.4th 493, 499 (6th Cir. 2024) (“As neither [debtor] nor the [Bankruptcy] Trustee filed a complaint, it is difficult to believe that this proceeding is the kind to which the EAJA applies.”). Habeas proceedings are commenced by an “application,” 28 U.S.C. § 2241(b), which is also referred to as a “petition” under Habeas Rule 2.

The required contents of a petition under Habeas Rule 2 are different than the pleading standards for a complaint under Civil Rule 8. The Civil Rules do not provide for a screening procedure like the one in Habeas Rule 4, and the statutory version of that screening process applies to a narrower class of civil plaintiffs. See 28 U.S.C. § 1915(e)(2). There is no habeas analog to dismissal motions under Civil Rule 12 and, in the absence of defense motion practice, responsive pleadings are generally required as a default under the Civil Rules. That is not the case in habeas proceedings. See Browder v. Dir., Dep't of Corr. of Ill., 434 U.S. 257, 269 n.14, 98 S.Ct. 556, 54 L.Ed.2d 521 (1978); see also Habeas Rule 5. Under Habeas Rule 6, discovery procedures under the Civil Rules are only available upon a showing of “good cause,” and the party seeking discovery must “provide reasons” for each request. Habeas Rule 6(a)-(b). Discovery in a “civil action” is not so restrictive. The fact-finding procedures at the summary judgment and trial stages of a civil action are nowhere to be found in the Habeas Rules.

Despite these and other differences, the panel relied on the 1980 version of Rule 81(a) to assert that, “at the time of the EAJA's enactment, the Federal Rules of Civil Procedure treated habeas proceedings as civil actions.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *4. That is not quite right. See AARP v. Trump, 605 U.S. 91, 107, 145 S.Ct. 1364, 221 L.Ed.2d 765 (2025) (Alito, J., dissenting) (“[I]n accordance with Rule 81, we have acknowledged that some Federal Rules are inapplicable in habeas.”); Schlanger v. Seamans, 401 U.S. 487, 490 n.4, 91 S.Ct. 995, 28 L.Ed.2d 251 (1971) (“Though habeas corpus is technically ‘civil,’ it is not automatically subject to all the rules governing ordinary civil actions.”). Rule 81 referred to habeas “proceedings” not “actions.” Fed. R. Civ. P. 81(a)(2) (1980). If a habeas “proceeding” was a “civil action” for purposes of the Civil Rules, Rule 81(a) would be superfluous—that is, we would not need the reference to habeas in Rule 81(a) if it is as clear as the panel suggested that habeas is a “civil action.”

In 1980, Rule 81(a) stated that the Civil Rules applied only “to the extent that the practice in such proceedings [1] is not set forth in statutes of the United States and [2] has heretofore conformed to the practice in civil actions.” See Fed. R. Civ. P. 81(a)(2) (1980). Those were big caveats. Habeas practice had been governed by “statutes of the United States,” id., since the Judiciary Act of 1789. See Pub. L. No. 1-20, § 14, 1 Stat. 73, 81-82 (1789). Distinct habeas procedures that differed from the Civil Rules were set forth in the Habeas Rules, which were approved by Congress prior to the enactment of the EAJA, as well as existing statutes. See Pub. L. No. 94-426, § 1, 90 Stat. 1334, 1334 (1976); see also 8 U.S.C. §§ 2241-2254 (setting forth additional habeas procedures). Because “[t]he problems presented by these [habeas] proceedings are materially different from those dealt with in the Federal Rules of Civil Procedure,” “reliance upon usage and the opaque language of Civil Rule 81(a)(2) is transparently inadequate.” Harris, 394 U.S. at 300 n.7, 89 S.Ct. 1082.

All of this context helps to explain why seven judges of this Court refused to treat habeas and “ordinary civil cases” as “indistinguishable” in Bendolph. See 409 F.3d at 166. While we have described habeas as “civil in nature,” that does not mean the § 2241 petitions at issue were “civil actions” under the EAJA. See Long v. Wilson, 393 F.3d 390, 402 (3d Cir. 2004) (“While civil in nature, habeas corpus cases are different from ordinary civil cases where only the interests of the parties are involved.”); Santana v. United States, 98 F.3d 752, 754 (3d Cir. 1996) (“[H]abeas corpus cases are, in effect, hybrid actions whose nature is not adequately captured by the phrase ‘civil action.’ ”); Callwood v. Enos, 230 F.3d 627, 632 (3d Cir. 2000) (“A suit seeking a writ of habeas corpus, although admittedly somewhat of a hybrid, is considered civil in nature.”); Lee v. Johnson, 799 F.2d 31, 40 (3d Cir. 1986) (reasoning that the Second Circuit's holding in Boudin “certainly lends some support to the government's proposed restrictive interpretation of the ‘civil action’ language in the [EAJA]”).

The panel described Santana as an instance where this Court “acknowledged that the ‘plain meaning’ and ‘literal scope’ of the phrase ‘civil action’ include habeas actions.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *4 (quoting Santana, 98 F.3d at 754). Those words appear in Santana, but they are qualified in the opinion by the phrase “[a]t first blush.” 98 F.3d at 754. The qualifier was necessary because Santana held that the term “civil action” in 28 U.S.C. § 1915(b), from the 1996 Prison Litigation Reform Act, “lacks a plain meaning” and does not include “habeas proceedings.” Id. at 755. We later extended that reasoning to the neighboring three-strikes provision, § 1915(g). See Garrett v. Murphy, 17 F.4th 419, 431 (3d Cir. 2021). Thus, Santana and Garrett provide strong precedential support for the government's position here, as do Bendolph, Long, Callwood, and Lee.

Seven other Circuits have interpreted “civil action” in § 1915, as in Santana and Garrett, to exclude habeas proceedings.³ Similarly, the Supreme Court has held that “civil action” in 28 U.S.C. § 1391(e) does not include habeas. See Schlanger, 401 U.S. at 490 n.4, 91 S.Ct. 995. The panel cherry-picked from the discussion of Schlanger in Stafford v. Briggs, without noting the Stafford Court's conclusion that “[t]he clear [import] of our statement in Schlanger is that Congress did not intend the phrase ‘civil action’ to be given the sweeping definition argued for it in that case.” 444 U.S. 527, 543, 100 S.Ct. 774, 63 L.Ed.2d 1 (1980).

The panel sought to distinguish Schlanger, Harris, and Santana by arguing that those opinions “reached that conclusion by doing something we may not: overriding the plain meaning of statutory text with legislative history.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *6. True, “[l]egislative history cannot supply a waiver that is not clearly evident from the language of the statute.” FAA v. Cooper, 566 U.S. 284, 290, 132 S.Ct. 1441, 182 L.Ed.2d 497 (2012). I certainly share the panel's view that, ideally, “[w]e do not read statutes this way anymore” because of the limited value that legislative history offers to the project of determining a statute's single, best meaning. Michelin, ––– F.4th at ––––, 2026 WL 263483, at *6; see also Conroy v. Aniskoff, 507 U.S. 511, 519, 113 S.Ct. 1562, 123 L.Ed.2d 229 (1993) (Scalia, J., concurring in the judgment) (“We are governed by laws, not by the intentions of legislators.”). But Schlanger, Harris, and Santana are compelling because of the manner in which they explained that habeas is sui generis, not because of their legislative history analyses. There is no basis for casting aside the reasoning in those cases, which helps to explain why “any civil action” is ambiguous in the EAJA.

More broadly, in light of the case law construing § 1391(e), § 1915(b), and § 1915(g) to exclude habeas, I respectfully disagree that “[e]very time Congress has used ‘any civil action’ or a cognate, the phrase encompasses habeas proceedings, save where Congress expressly excluded them.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *9. The word “any” in “any civil action” cannot bear the weight placed upon it by the panel when making that assertion. Id. at –––– – ––––, 2026 WL 263483 at *7-8. “Any” must be interpreted “narrowly” where “a broad reading ․ implicate[s] sovereignty concerns.” Ali v. BOP, 552 U.S. 214, 220 n.4, 128 S.Ct. 831, 169 L.Ed.2d 680 (2008). The Ali Court cited Raygor v. Regents of University of Minnesota, where the Supreme Court relied on sovereign immunity principles to conclude that “any civil action” in 28 U.S.C. § 1367(a) did not include claims against States, “even though nothing in the statute expressly excludes such claims.” 534 U.S. 533, 541, 122 S.Ct. 999, 152 L.Ed.2d 27 (2002). Raygor, in turn, cited Blatchford v. Native Village of Noatak, where the Court held that “all civil actions” in 28 U.S.C. § 1362 did not include claims by Indian tribes against nonconsenting States. 501 U.S. 775, 786-87, 111 S.Ct. 2578, 115 L.Ed.2d 686 (1991) (finding no sovereign-immunity waiver in the absence of “unmistakably clear” language from Congress); see also Atascadero State Hosp. v. Scanlon, 473 U.S. 234, 245-46, 105 S.Ct. 3142, 87 L.Ed.2d 171 (1985) (holding that sovereign immunity barred recovery against State despite statute authorizing claims against “any recipient of Federal assistance”); Coll. Sav. Bank v. Fla. Prepaid Postsecondary Educ. Expense Bd., 527 U.S. 666, 682, 119 S.Ct. 2219, 144 L.Ed.2d 605 (1999) (explaining that federal sovereign immunity is “obviously the closest analogy” to state sovereign immunity). Thus, Ali, Raygor, Blatchford, and Scanlon are additional powerful authority supporting the government's interpretation of the EAJA.

B.

The panel started with quotes from Blackstone and placed significant emphasis on the old-soil principle in concluding that “any civil action” unambiguously includes habeas. The reasoning does not hold up to close inspection. There is no indication that Congress looked to English common law or the nineteenth century when crafting the EAJA in 1980, and the panel erred by focusing on the common law treatment of habeas rather than the relevant statutory phrase.

1.

The old-soil principle permits an inference that when Congress chooses a term that was “obviously transplanted from another legal source,” Congress intended the term to have the meaning attributed to it in that other source. Taggart v. Lorenzen, 587 U.S. 554, 560, 139 S.Ct. 1795, 204 L.Ed.2d 129 (2019). The panel jumped to the conclusion that “civil action” is “one such term.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *4. The Tenth Circuit opinion that the panel found “illuminating” adopted a similarly flawed approach with slightly more explanation. Id. at ––––, 2026 WL 263483 at *10 (citing Daley, 158 F.4th 1152). The Daley panel contended that “ ‘[c]ivil action’ was a common law ‘term of art in which was accumulated the legal tradition and meaning of centuries of practice.’ ” 158 F.4th at 1157 (quoting Sekhar v. United States, 570 U.S. 729, 733, 133 S.Ct. 2720, 186 L.Ed.2d 794 (2013)). The single case cited in Daley for that purported tradition, Sekhar, was not about civil actions or habeas petitions. There, the Supreme Court addressed extortion, “one of the oldest crimes in our legal tradition.” 570 U.S. at 733, 133 S.Ct. 2720.

The term “civil action” lacks a similar provenance and, thus, the panel's old-soil logic faltered out of the gate. I am not suggesting that the phrase was unknown in the Colonies. But any soil around that term was rocky at best—barren, in my view—and insufficient to support a waiver of sovereign immunity. See Kousisis v. United States, 605 U.S. 114, 129, 145 S.Ct. 1382, 221 L.Ed.2d 781 (2025) (“The old-soil principle does not apply in the absence of a well-settled rule.”); Kemp v. United States, 596 U.S. 528, 539, 142 S.Ct. 1856, 213 L.Ed.2d 90 (2022) (reasoning that the old-soil “principle applies only when a term's meaning was well-settled before the transplantation”).

In pre-Founding England, there was not a single form of civil action. Far from it. See Ndungu v. AG, 126 F.4th 150, 159 (3d Cir. 2025) (explaining that the Court has not “look[ed] to the common-law meaning” of “crime involving moral turpitude” “for good reason: it is not clear that the term had an established meaning prior to its inclusion in the immigration statute”). There were common law and equity courts with varied complex proceedings that bore little resemblance to the federal civil actions that existed when the EAJA was passed. See, e.g., 3 W. Blackstone, Commentaries on the Laws of England 23-24 (1768) (describing “several species of courts of justice” and “what injuries are cognizable, and how redressed, in each respective species of courts”). This is noted in one of the dictionaries the panel cited, which included in the definition of “civil action” the fact that “former distinctions between actions at law and suits in equity, and the separate forms of those actions and suits, have been abolished” under the Civil Rules. Civil Action, Black's Law Dictionary (5th ed. 1979); see also Michelin, ––– F.4th at ––––, 2026 WL 263483, at *4; In re Teter, 90 F.4th at 499 (“[W]ith respect to a federal civil action in particular, the manner of ‘action’ at issue here, Black's understanding parrots that of Federal Rule of Civil Procedure 3 ”).

Founding-era authorities further undercut any suggestion of a single form of “civil action” sufficient to justify recourse to the old-soil principle. The Seventh Amendment refers to “Suits at common law,” not “civil actions,” in order to distinguish from “admiralty, vice admiralty, and chancery courts” that the British had abused to “evad[e] American juries.” SEC v. Jarkesy, 603 U.S. 109, 121, 144 S.Ct. 2117, 219 L.Ed.2d 650 (2024). The Revised Statutes codified U.S. enactments prior to 1874, which referenced a wide variety of proceedings: “suits at common law,” “suits in equity,” “causes of action arising under the postal laws of the United States,” “civil causes of admiralty and maritime jurisdiction,” and “the writ of quo warranto.” Revised Statutes § 563 (2d ed. 1878). Appellate jurisdiction extended to several different types of “personal action[s],” “cause[s],” and “suits,” including “suits of a civil nature at common law or in equity.” Id. §§ 629, 637, 640-641, 644. Perhaps some of these proceedings could be characterized as “civil in nature,” but they were not called “civil actions” at the Founding.

During this period, the admitted States' procedures loomed large, as “nearly all federal court business consisted of state-law diversity claims, while nearly all federal claims began and ended in state court.” Stanford ex rel. Phillips v. Brandon Nursing & Rehab. Ctr., LLC, 160 F.4th 118, 125 (5th Cir. 2025) (Oldham, J., dissenting). There was hardly any federal procedure to speak of. The Process Act of 1789 directed federal courts to follow established common-law procedures of the States in which they were located. See Pub. L. No. 1-21, § 2, 1 Stat. 93, 93 (1789); see also Conformity Act of 1828, Pub. L. No. 20-68, § 1, 4 Stat. 278, 278-80 (1828); Conformity Act of 1842, Pub. L. No. 27-109, 5 Stat. 499, 499 (1842). There was also a separate Process Act for equity matters. See Process Act of 1792, Pub. L. No. 2-36, § 2, 1 Stat. 275, 275-76 (1792). Some believed federal courts' application of state-law procedures led to “a mass of worthless pleadings and exceptions” that “serve[d] only to perplex the court, and impede the due administration of justice.” Randon v. Toby, 52 U.S. (11 How.) 493, 517, 13 L.Ed. 784 (1850); see also McFaul v. Ramsey, 61 U.S. (20 How.) 523, 525, 15 L.Ed. 1010 (1857) (explaining that the common-law system was “ruthlessly abolished in many of our States, who have rashly substituted in its place the suggestions of sciolists, who invent new codes and systems of pleading to order”).

In many respects, a deep dive in this area raises more questions than answers about the particulars of how trial courts functioned at common law and in the nineteenth century. In the relevant respect, however, it is clear that neither the panel's British authorities nor the chaotic onset of private litigation in the United States supports the type of unambiguous interpretation of “civil action” that is necessary to find a waiver of sovereign immunity.

2.

The panel's old-soil reasoning focused on the historical treatment of habeas, which is essentially backwards. The principle does not work by looking at what history tells us about whether a term that was not included in the statute (“habeas”) is within the scope of the term that made it into the enacted law (“any civil action”). See Ysleta Del Sur Pueblo v. Texas, 596 U.S. 685, 718, 142 S.Ct. 1929, 213 L.Ed.2d 221 (2022) (Roberts, C.J., dissenting) (reasoning that the old-soil principle does not apply where statutory text “bears little resemblance” to historical comparator). We do not know if Congress thought about habeas at all when passing the EAJA, much less whether those elected officials adopted concepts of the writ from Blackstone or the nineteenth century. Because “habeas” did not make it through bicameralism and presentment in 1980, it is not appropriate to rely on the common law habeas soil to draw such strong inferences about the meaning of a different term that Congress actually chose.

In addition to this methodological flaw, the panel's historical habeas authorities do not demonstrate that § 2241 petitions raising due process arguments like Petitioners' were unambiguously “civil actions” at that time. There are “shortcomings in the historical record” regarding habeas because of serious gaps in common-law opinion reporting. Boumediene v. Bush, 553 U.S. 723, 752, 128 S.Ct. 2229, 171 L.Ed.2d 41 (2008) (citing Paul D. Halliday, et ano., The Suspension Clause: English Text, Imperial Contexts, and American Implications, 94 Va. L. Rev. 575, 588-90 (2008)). And the panel's contrary suggestion “excessively simplifies a complex history.” Harris, 394 U.S. at 294 n.4, 89 S.Ct. 1082.

We do know, however, that “the writ played only a procedural role,” and that a common law habeas proceeding was treated more like a summary matter than full-blown civil litigation. DHS v. Thuraissigiam, 591 U.S. 103, 144 n.2, 140 S.Ct. 1959, 207 L.Ed.2d 427 (2020) (Thomas, J., concurring); see also Maxwell Cohen, Some Considerations on the Origins of Habeas Corpus, 16 Can. B. Rev. 92, 110 (1938) (arguing that in the twelfth and thirteenth centuries “[e]very branch of the judicial process of the day knew of it as a ready device to compel the persons or parties to pleas, civil and criminal, to appear if the initial processes failed to achieve the purpose”). As in 1980 and the present, there were significant differences between habeas proceedings and private litigation in the common law system. See, e.g., McCleskey v. Zant, 499 U.S. 467, 479, 111 S.Ct. 1454, 113 L.Ed.2d 517 (1991) (“At common law, res judicata did not attach to a court's denial of habeas relief.”); Sanders v. United States, 373 U.S. 1, 8, 83 S.Ct. 1068, 10 L.Ed.2d 148 (1963) (“[A]t common law habeas corpus judgments were not appealable.”).

Much of this was elided by the panel in favor of binary distinctions between rights and wrongs, private and public wrongs, and civil and criminal proceedings. See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *1. The truth is, it was not that simple. See Browning-Ferris Indus. of Vt., Inc. v. Kelco Disposal, Inc., 492 U.S. 257, 271, 109 S.Ct. 2909, 106 L.Ed.2d 219 (1989) (“Petitioners ultimately rely on little more than the fact that the distinction between civil and criminal law was cloudy (and perhaps nonexistent) at the time of Magna Carta.”). At common law, a violation of personal liberty had private and criminal implications, and habeas relief was separate from remedies to be pursued through lawsuits and prosecutions.

When Blackstone inventoried private wrongs, he drew a distinction between “injuries with and without force.” 3 W. Blackstone, Commentaries on the Laws of England 119 (1768). A violation of personal liberty caused an injury with force. See id. Injuries with force were “something of the criminal kind, being always attended with some violation of the peace; for which in strictness of law a fine ought to be paid to the king, as well as private satisfaction to the party injured.” Id. at 118-19. Thus, when discussing the liberty interest that the panel focused on, Blackstone used language that is entirely consistent with courts' subsequent descriptions of habeas as being a hybrid proceeding rather than a standalone civil action.

As an injury with force, a violation of the right of personal liberty gave rise to private remedies “of two sorts; the one removing the injury, the other making satisfaction for it.” 3 W. Blackstone, Commentaries on the Laws of England 128 (1768); see also id. at 127 (explaining that a violation of personal liberty caused “the injury of false imprisonment,” which was a “heinous public crime” and also compensable by “removing the actual confinement for the present” and “private reparation”). With respect to the first type of remedy, Blackstone described four writs that functioned as “[t]he means of removing the actual injury of false imprisonment,” including the form of habeas writ relevant here: habeas corpus ad faciendum, subjiciendum, et recipiendum. Id. at 128-36. This remedy was accomplished, and the habeas process was “complete[d],” by “removing the injury of unjust and illegal confinement.” Id. at 137.

The panel used shorthand to describe these four categories of writs as “civil actions available for ‘the redress of private wrongs.’ ” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *1. But the shorthand does not work. According to Blackstone, the “civil action” was distinct and to be pursued “after [confinement] is over, by subjecting the wrongdoer to a civil action, on account of the damage sustained by the loss of time and liberty.” 3 W. Blackstone, Commentaries on the Laws of England 127 (1768). The “satisfactory remedy” for the injury was “an action of trespass” or false imprisonment, which was “almost unavoidably ․ accompanied with a charge of assault and battery” as well as a fine payable “to the king for the violation of the public peace.” Id. at 138. Therefore, while “the writ of habeas corpus has deep roots in that rich soil,” that soil does not support the panel's conclusion. Michelin, ––– F.4th at ––––, 2026 WL 263483, at *4. Habeas was a procedural tool for putting an end to the intrusion on liberty. Civil actions for damages, and sometimes criminal prosecutions, followed separately.

Moving forward on the timeline, “the scope of habeas has been tightly regulated by statute” since the Founding. Thuraissigiam, 591 U.S. at 144 n.2, 140 S.Ct. 1959 (Thomas, J., concurring). In the Judiciary Act of 1789, and continuing through the Revised Statutes, habeas was addressed in laws that were separate from those defining the civil and criminal jurisdiction of the federal courts. See Judiciary Act §§ 9-13, 1 Stat. at 76-81 (civil and criminal jurisdiction), id. § 14, 1 Stat. at 81-82 (habeas); Revised Statutes chs. 3, 7, 11 (civil and criminal jurisdiction), id. ch. 13 (habeas procedure). As at common law, habeas was treated as “a special mode, and confined to a limited class of cases.” Ex parte Parks, 93 U.S. 18, 21, 23 L.Ed. 787 (1876); see also Wight v. Nicholson, 134 U.S. 136, 148, 10 S.Ct. 487, 33 L.Ed. 865 (1890) (“[T]he proposition is so clear that in a writ of habeas corpus nothing can be inquired into but the jurisdiction of the court.”); Ex parte Belt, 159 U.S. 95, 100, 15 S.Ct. 987, 40 L.Ed. 88 (1895) (“The general rule is that the writ of habeas corpus will not issue unless the court under whose warrant the petitioner is held is without jurisdiction, and that it cannot be used to correct errors.”); Harlan v. McGourin, 218 U.S. 442, 448, 31 S.Ct. 44, 54 L.Ed. 1101 (1910) (“Upon habeas corpus the court examines only the power and authority of the court to act, not the correctness of its conclusions.”).

Regardless of whether habeas had civil characteristics, this statutory treatment led to habeas proceedings being handled expeditiously, summarily, and quite differently than other proceedings we now think of as “civil actions.” See, e.g., Salinger v. Loisel, 265 U.S. 224, 231, 44 S.Ct. 519, 68 L.Ed. 989 (1924) (explaining that Revised Statutes § 761, governing habeas procedure, “has been construed as meaning that each application is to be disposed of in the exercise of a sound judicial discretion guided and controlled by a consideration of whatever has a rational bearing on the propriety of the discharge sought”); Nishimura Ekiu v. United States, 142 U.S. 651, 662, 12 S.Ct. 336, 35 L.Ed. 1146 (1892) (“A writ of habeas corpus is not like an action to recover damages for an unlawful arrest or commitment ”).

The panel's citation to Holmes v. Jennison, 39 U.S. (14 Pet.) 540, 10 L.Ed. 579 (1840) only underscores the lack of a consensus on habeas minutia in the early days of the Republic. See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *1. The Holmes Court was “so divided that no opinion [could] be delivered as the opinion of the Court.” 39 U.S. at 561. Chief Justice Taney and three other Justices believed it “too plain for argument” that a Vermont habeas proceeding was a “suit” that the U.S. Supreme Court had jurisdiction to review “upon a writ of error” under § 25 of the Judiciary Act, 1 Stat. at 85-87. 39 U.S. at 545, 565-67. The Justices said nothing about federal habeas proceedings under the existing relevant provision, § 14 of the Judiciary Act.

I do not mean to make too much of the citation to Holmes, as the panel cited other cases from that period that added force to their historical position about the civil characteristics of habeas proceedings. Not enough, though, to invoke the old-soil principle as a means of eliminating ambiguity in the scope of “any civil action.”

3.

Even if the panel selected the appropriate historical frame of reference, and even if habeas was the right focus for the old-soil inquiry—neither of which is true—the panel addressed habeas at a level of generality that is fatally untethered from the circumstances of these cases.

While habeas generally was available to review executive detention, Blackstone recognized that the King had exclusive authority to exclude aliens. See Qatanani v. AG, 144 F.4th 485, 507 (3d Cir. 2025) (Matey, J., dissenting). “The writ of habeas corpus known to the Framers was quite different from that which exists today.” Felker v. Turpin, 518 U.S. 651, 663, 116 S.Ct. 2333, 135 L.Ed.2d 827 (1996); cf. Ex parte Wilson, 10 U.S. (6 Cranch) 52, 52-53, 3 L.Ed. 149 (1810) (“MARSHALL, Ch. J. after consultation with the other judges, stated that the court was not satisfied that a habeas corpus is the proper remedy, in a case of arrest under a civil process.”).

Immigration enforcement looked quite different too. “The right of a nation to expel or deport” aliens was as “absolute and unqualified[ ] as the right to prohibit and prevent their entrance into the country.” Fong Yue Ting v. United States, 149 U.S. 698, 707, 13 S.Ct. 1016, 37 L.Ed. 905 (1893); see also Ping v. United States, 130 U.S. 581, 603-04, 9 S.Ct. 623, 32 L.Ed. 1068 (1889) (“That the government of the United States, through the action of the legislative department, can exclude aliens from its territory is a proposition which we do not think open to controversy. It is a part of its independence. If it could not exclude aliens it would be to that extent subject to the control of another power.”). In the nineteenth century, it was believed that “[p]roceedings to exclude or expel would be vain if those accused could not be held in custody pending the inquiry into their true character, and while arrangements were being made for their deportation.” Wong Wing v. United States, 163 U.S. 228, 235, 16 S.Ct. 977, 41 L.Ed. 140 (1896).

The due process rights of aliens who entered the Country without being formally admitted were limited to “those conferred by statute.” Thuraissigiam, 591 U.S. at 131, 140 S.Ct. 1959.

It [was] not within the province of the judiciary to order that foreigners who have never been naturalized, nor acquired any domicile or residence within the United States, nor even been admitted into the country pursuant to law, shall be permitted to enter, in opposition to the constitutional and lawful measures of the legislative and executive branches of the national government. As to such persons, the decisions of executive or administrative officers, acting within powers expressly conferred by congress, are due process of law.

Nishimura Ekiu, 142 U.S. at 660, 12 S.Ct. 336; see also Pino-Porras v. AG, 2025 WL 1752491, at *5 n.10 (3d Cir. 2025) (Matey, J., dissenting). Due process was backstopped by the protection principle; “grounded in tradition, the long-accepted rule that a sovereign's laws and the privilege of the sovereign's protections extend only to ‘persons and things within its own territory according to its own sovereign will and public policy.’ ” Robles v. AG, 2025 WL 2924867, at *4 (3d Cir. 2025) (Matey, J., dissenting) (quoting Joseph Story, Commentaries on the Conflict of Laws, Foreign and Domestic § 22 (Boston, Hilliard, Gray, and Co. 1834)).

So while there were habeas proceedings involving aliens in the nineteenth century, it is not at all clear that they looked anything like what Petitioners did in these cases. See Johnson v. Arteaga-Martinez, 596 U.S. 573, 585, 142 S.Ct. 1827, 213 L.Ed.2d 125 (2022) (Thomas, J., concurring) (“[T]here is considerable historical evidence that the Due Process Clause does not apply to laws governing the removal of aliens.”); Sessions v. Dimaya, 584 U.S. 148, 211, 138 S.Ct. 1204, 200 L.Ed.2d 549 (2018) (Thomas, J., dissenting) (“[T]he notion that the Due Process Clause governed the removal of aliens was not announced until the 20th century.”). In fact, “the equivalent of the habeas relief Justice Story ordered in a case while riding circuit” in 1813 would be to release Petitioners “in the cabin of a plane bound for” their countries of origin. Thuraissigiam, 591 U.S. at 119, 140 S.Ct. 1959 (citing Ex parte D'Olivera, 7 F. Cas. 853, 854 (C.C.D. Mass. 1813)). For all of these reasons, the old-soil principle does not support the waiver of sovereign immunity declared by the panel.

C.

For purposes of the question at hand, the relevant soil is more recent: the Rules Enabling Act (1934), the resulting Federal Rules of Civil Procedure (1938), and the positive-law codification of Title 28 (1948). See United States v. Castleman, 572 U.S. 157, 176-77, 134 S.Ct. 1405, 188 L.Ed.2d 426 (2014) (Scalia, J., concurring in part and concurring in the judgment) (reasoning that a “more accurate formulation” of the old-soil principle “is that when a word is obviously transplanted from another legal source, whether the common law or other legislation, it brings the old soil with it”). In my view, those enactments were “obviously” where Congress was looking when placing “any civil action” in the EAJA. Id.

Congress established the modern concept of a single “civil action” by passing the Rules Enabling Act in 1934. See Pub. L. No. 73-415, § 2, 48 Stat. 1064, 1064 (1934). The Act authorized the Supreme Court to “secure one form of civil action and procedure.” Id. (emphasis added). The Supreme Court appointed a committee to propose procedural rules and adopted the committee's proposals, with changes, in 1937. See 4 Wright & Miller's Federal Practice & Procedure § 1004 (4th ed. 2025). The Attorney General submitted the committee's report to Congress on behalf of the Court in 1938, and the Civil Rules became effective in September of that year. See id.

The first version of Rule 2 was similar to the 1980 and current versions: “There shall be one form of action to be known as ‘civil action.’ ” Fed. R. Civ. P. 2 (1938). The Civil Rules applied to habeas proceedings “on appeal” but “not ․ otherwise ․ except to the extent that the practice in such proceedings is not set forth in statutes of the United States and has heretofore conformed to the practice in actions at law or suits in equity.” Fed. R. Civ. P. 81(a)(2) (1938). At the time, the “procedure to be followed” in habeas proceedings was codified at Chapter 13 of the Revised Statutes. Holiday v. Johnston, 313 U.S. 342, 350, 61 S.Ct. 1015, 85 L.Ed. 1392 (1941). Relying on the existing version of Rule 81, the Holiday Court rejected the use of a special master under Civil Rule 53 by reasoning that “the practice in habeas corpus is set forth in plain terms in the Revised Statutes.” Id. at 353. Holiday is one of several pre-EAJA signals from the Supreme Court that, although some Civil Rules may apply in habeas proceedings, not all of them do because habeas procedure is codified elsewhere.

That continued to be the case following the 1948 codification of Title 28. See Act of June 25, 1948, Pub. L. No. 80-773, 62 Stat. 869 (1948). During that process, Congress modified the existing habeas statutes. See id. §§ 2241-2255; see also H.R. Rep. No. 80-308, at 7 (1st Sess. 1947) (“The habeas corpus chapter has been rewritten to conform with legislation pending in Congress and approved by the Judicial Conference of the United States.”). Some of the revisions were made to conform to “actual” and “existing practice” as well as to “clarif[y] existing law” and “practice.” H.R. Rep. No. 80-308, at A178 (citing 28 U.S.C. §§ 2242-2243); id. at A179 (citing 28 U.S.C. §§ 2245-2246). Thus, as the Holiday Court had held with respect to the Revised Statutes, most habeas procedures continued to exist outside the Civil Rules after 1948. The “practice in such proceedings” under Rule 81(a) was largely set forth in separate statutes.

In the 1948 codification, Congress also adopted 31 proposals to insert the term “civil action” into Title 28, which the Reviser's Notes explained was based on “Rule 2 of the Federal Rules of Procedure.” See, e.g., H.R. Rep. No. 80-308, at A105; see also Am. Cyanamid Co. v. Hammond Lead Prods., Inc., 495 F.2d 1183, 1186 (3d Cir. 1974) (explaining that “the legislators and drafters of the Judicial Code [in Title 28] emphasized that the Reviser's Notes explained all changes in the law”).⁴ One example was § 1404, which codified forum non conveniens in “any civil action.” The panel cited a case noting the “unmistakable” scope of “any civil action” in § 1404. See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *8 (citing Ex parte Collett, 337 U.S. 55, 58, 69 S.Ct. 944, 93 L.Ed. 1207 (1949)). A footnote in Ex parte Collett confirmed, however, that the term had been imported from the Civil Rules. See 337 U.S. at 58 n.6, 69 S.Ct. 944.

Rather than English common law, the Rules Enabling Act, Civil Rules, and Title 28 are the old soil bearing on the meaning of the phrase “any civil action” in the EAJA. Habeas is not a “civil action” under the Civil Rules, and Congress chose that phrase rather than other broader options for the EAJA. This soil inures to the benefit of the government in this interpretive exercise.

D.

In the decades after Congress copy-pasted “civil action” from the Civil Rules to more than 30 sections of Title 28, there were several developments that bear on the public meaning of that phrase in 1980 and caution against finding a waiver of sovereign immunity.

Prior to the enactment of the EAJA, habeas proceedings relating to criminal convictions exploded in a way that makes it hard to believe Congress was trying to incentivize more of them. Res judicata still did not foreclose successive petitions, and courts were cautious in applying the “abuse of the writ” concept that Congress had introduced for federal petitions. See McCleskey, 499 U.S. at 477, 111 S.Ct. 1454. In 1953, the Supreme Court, in essence, “decided that federal courts could grant a writ of habeas corpus simply because they disagreed with a state court's judgment.” Edwards v. Vannoy, 593 U.S. 255, 278, 141 S.Ct. 1547, 209 L.Ed.2d 651 (2021) (Thomas, J., concurring) (citing Brown v. Allen, 344 U.S. 443, 463, 73 S.Ct. 397, 97 L.Ed. 469 (1953)). The “proliferation of new federal procedural rights, combined with a federal post-conviction mechanism that functioned like an ordinary appeal, soon yielded a giant haystack of habeas petitions.” Id. at 288-89, 141 S.Ct. 1547 (Gorsuch, J., concurring). The Bureau of Justice Statistics reported that the number of habeas petitions by state and federal prisoners increased by over 347% between 1961 and 1980—when prisoners filed about 8,444 petitions.⁵

During that increase, the Supreme Court issued three decisions suggesting that a habeas proceeding was not a “civil action” under the Civil Rules. In Harris, the Court held that discovery under Civil Rule 33 was not available in habeas. See 394 U.S. at 292-98, 89 S.Ct. 1082. The Harris Court noted that “[s]uch specific evidence as there is with respect to the intent of the draftsmen of the rules indicates nothing more than a general and nonspecific understanding that the rules would have very limited application to habeas corpus proceedings.” Id. at 295, 89 S.Ct. 1082. In Schlanger, as noted above, the Court reasoned that the term “civil action” in 28 U.S.C. § 1391(e) did not include habeas. 401 U.S. at 490 n.4, 91 S.Ct. 995. The Schlanger Court explained that “[t]he procedure governing issuance of the writ is provided by statute.” Id. at 489, 91 S.Ct. 995. In 1978, despite characterizing habeas as a “civil proceeding”—as opposed to a “civil action”—the Supreme Court noted “differences between general civil litigation and habeas corpus proceedings” and that Rule 81(a)(2) “recognizes the supremacy of the statutory procedures over the Federal Rules.” Browder, 434 U.S. at 268-69, 98 S.Ct. 556. The Court held that Rules 52 and 59 govern post-judgment relief in habeas proceedings based in large part on the “settled conformity of habeas corpus and other civil proceedings” with respect to that issue. Id. at 271, 98 S.Ct. 556. Nevertheless, the upshot of Harris, Schlanger, and Browder was that a habeas proceeding was not a normal “civil action.”

While habeas review of criminal convictions skyrocketed, habeas review relating to immigration detention was infrequent. At least in part because there was not much detention going on. In contrast to the 8,444 petitions filed by convicted prisoners in 1980 alone, there were only 318 petitions filed by aliens adjudicated between 1976 and 1980.⁶ INS's approach prior to 1981 “was to release aliens on parole unless they were security risks or thought likely to abscond.” Amanullah v. Nelson, 811 F.2d 1, 7 (1st Cir. 1987). Congress later rejected that approach because it “imposed more than a monetary cost on the Nation.” Demore v. Kim, 538 U.S. 510, 518, 123 S.Ct. 1708, 155 L.Ed.2d 724 (2003). Following the passage of the Illegal Immigration Reform and Immigrant Responsibility Act of 1996, the Supreme Court noted that Congress had adopted the mandatory detention provision relied upon by the government in one of these cases, 8 U.S.C. § 1226(c), “against a backdrop of wholesale failure by the INS to deal with increasing rates of criminal activity by aliens.” Demore, 538 U.S. at 518, 123 S.Ct. 1708. That backdrop is an additional reason to doubt Congress was trying to create incentives in the EAJA for aliens to pursue release through habeas.

Also keep in mind that there was already a way for habeas petitioners to seek fees prior to the EAJA. In 1970, Congress authorized courts to appoint counsel for habeas petitioners in the “interests of justice.” Act of Oct. 14, 1970, Pub. L. No. 91-447, § g, 84 Stat. 916, 919 (1970) (amending 18 U.S.C. § 3006A). Habeas Rule 8(c), enacted in 1976, also provided for appointments of counsel. See § 2(5)-(6), 90 Stat. at 1335. The availability of these appointments was “a marked difference from general civil practice” in civil actions. Ewing, 826 F.2d at 971. “The volume of habeas petitions processed in the federal courts together with the opportunity for court-appointed representation suggests that there [was] little, if any, economic deterrent in seeking review of the legality of criminal confinement.” Id. This is consistent with other textual and contextual indications that Congress did not have habeas in mind when it used the phrase “any civil action” in the EAJA.

Finally, when the EAJA was enacted, two provisions of the Immigration and Nationality Act stated that aliens could obtain counsel “at no expense to the government” in immigration proceedings. 8 U.S.C. §§ 1252(b)(2), 1362 (1976). In that setting, Congress provided that the government would not pay for aliens' lawyers. Two courts that improperly relied on legislative history in sovereign-immunity analyses have contended that §§ 1252 and 1362 led to a gap in representation in those immigration proceedings, which are distinct from habeas, that Congress wanted to fill through the EAJA. See Vacchio, 404 F.3d at 670 (citing In re Hill, 775 F.2d at 1041). It is highly unlikely that Congress would have sought to limit the reach of the specific language from §§ 1252 and 1362 prohibiting attorneys' fees in immigration proceedings by using an ambiguous reference to “civil actions” to permit fee awards in habeas challenges to those proceedings.

The Eleventh Circuit had the better read on the relevance of §§ 1252 and 1362. In a holding that the Supreme Court did not reach in Ardestani, see 502 U.S. at 139, 112 S.Ct. 515, that court reasoned that “[i]t would be unjust to allow such an [EAJA] award against the government since Congress specifically has determined that fees against the government are not available” in immigration proceedings. Ardestani v. INS, 904 F.2d 1505, 1513 (11th Cir. 1990). The Eleventh Circuit did not apply that reasoning to habeas proceedings, but extending the EAJA to immigration-related habeas would have created friction with §§ 1252 and 1362.

These developments plainly do not override clear statutory text, but they are part of the reason that I am not as sanguine as the panel that the phrase “any civil action” provides an unambiguous sovereign-immunity waiver in these cases.

E.

“[B]roader congressional context” reinforces my interpretation of the EAJA. Learning Res., Inc. v. Trump, 607 U.S. ––––, ––––, ––– S.Ct. ––––, ––– L.Ed.2d ––––, 2026 WL 477534, at *63 (2026) (Kavanaugh, J., dissenting).

Congress passed the EAJA in 1980, reenacted the statute with amendments in 1985, and amended it once more in 1992. Despite actively responding to other judicial and regulatory activity during that process, Congress did not address contrary authority regarding habeas. While I would not typically place much, if any, weight on congressional silence when interpreting a statute, the Supreme Court has already done so with respect to the EAJA. See Ardestani, 502 U.S. at 138, 112 S.Ct. 515 (reasoning that “Congress has twice expanded the EAJA's definition of ‘adversary adjudications’ ” without addressing “administrative deportation proceedings”). We have too. See Clarke, 904 F.2d at 178; see also Hashim v. INS, 936 F.2d 711, 715 (2d Cir. 1991) (“Congress has amended the EAJA explicitly to include certain previously uncovered administrative actions but has failed to include deportation proceedings.”). While this sort of reasoning may not be persuasive in de novo interpretation, I do not think we are free to reject it out of hand when considering the government's arguments regarding the lack of an unambiguous sovereign-immunity waiver.

The 1980 version of the EAJA expired on October 1, 1984. See Act of Oct. 21, 1980, Pub. L. No. 96-481, § 204(c), 94 Stat. 2321, 2329 (1980). In April of 1984, the Second Circuit held that a habeas proceeding relating to a felon's conditions of confinement was not a “civil action” under § 2412(d)(1)(A). See Boudin, 732 F.2d at 1111-15. When Congress renewed the EAJA in 1985, it did not address Boudin's “civil action” holding or habeas. See Act of Aug. 5, 1985, Pub. L. No. 99-80, §§ 6-7, 99 Stat. 183, 186-87 (1985).

Some might say that the silence takes on significance when viewed in context. In the 1985 reenactment, after hearings and input from interested parties as well as a veto by President Reagan of Congress's first reenactment attempt, Congress responded to other developments that reflect careful consideration of litigation developments relating to the EAJA. See generally H.R. Rep. No. 99-120 (1985); Sen. Rep. No. 98-586 (1984). For example, Congress clarified in Pub. L. No. 99-80 that the EAJA covered:

• “[J]udicial review of agency action.” § 2(a)(2), 99 Stat. at 184;

• Appeals in federal contracting disputes under Title 41. Compare Fid. Constr. Co. v. United States, 700 F.2d 1379, 1387 (Fed. Cir. 1983), with § 1(c)(2), 99 Stat. at 184;

• Cases transitioned to the Court of Federal Claims (then known as the Claims Court) following the Federal Courts Improvement Act of 1982. See § 2(c)(2), 99 Stat. at 185; Ellis v. United States, 711 F.2d 1571, 1573-74 (Fed. Cir. 1983);

• Condemnation actions involving eminent domain. Compare United States v. 329.73 Acres of Land, 704 F.2d 800, 804 (5th Cir. 1983), with § 2(c)(2), 99 Stat. at 185; and

• Social security claims. Compare 48 Fed. Reg. 45251 (1983) (HHS regulations), with § 3(2), 99 Stat. at 186.

Despite all of this, including a change to the statutory definition of “civil action,” Congress offered no legislative response to DOJ regulations asserting that the EAJA did not apply to INS proceedings. See 46 Fed. Reg. 48921, 48922 (1981). A House Report referenced Boudin's holding about the meaning of the phrase “position of the United States.” See H.R. Rep. No. 99-120, at 12 n.21 (1985). But Congress was silent about the availability of fees in habeas proceedings. See, e.g., Learning Resources, 2026 WL 477534, at *56 n.11, ––– U.S. at –––– n.11, ––– S.Ct. –––– (Kavanaugh, J., dissenting) (relying on congressional silence “not for determining the meaning of” a statute, “but rather to help show as an historical and factual matter that Members of Congress were aware of ․ the appeals court decision”).

Congress was also silent when it amended the EAJA in 1992 to clarify coverage for cases involving the Court of Veterans Appeals. See Jones v. Principi, 985 F.2d 582 (Fed. Cir. 1992) (unpublished). Before that amendment, the Tenth Circuit had followed Boudin and the Ninth Circuit came out the other way. Compare Ewing, 826 F.2d at 971, with In re Hill, 775 F.2d at 1040-41. Congress did nothing to address the 2-1 Circuit split regarding habeas in the 1992 amendments.

The 1992 amendments to the EAJA undermine the panel's conclusion in another respect. The panel pointed to “veterans' benefits actions” as a type of case with “unique rules of practice or procedure” for which fees were available under the EAJA. Michelin, ––– F.4th at ––––, 2026 WL 263483, at *5. But that was only clear after Congress explicitly expanded the statute's coverage. The same is true for some of the 1985 expansions noted above. If “civil action” in the EAJA was broad enough to reach any matter with civil characteristics, there would have been no need for Congress to add “proceedings for judicial review of agency action” to § 2412. See § 2(a)(2), 99 Stat. at 184. The language seems to contemplate fee awards in cases filed directly in federal appellate courts pursuant to procedures that are not a perfect fit with the Civil Rules. See 28 U.S.C. § 2342. Likewise for the 1985 expansion to reach appeals in Title 41 contracting disputes, as those appeals do not commence with the filing of a complaint. The fact that Congress extended the express reach of the statute to categories of cases that are not “civil actions” under the Civil Rules, only in limited circumstances and while remaining silent regarding habeas, supports the government's position.

* * *

In sum, the sovereign-immunity question presented in these cases is complex and exceptionally important. See Fed. R. App. P. 40(b)(2)(D). There is a deep Circuit split, and the outcome has significant implications for taxpayer dollars. See Fed. R. App. P. 40(b)(2)(C). I respect the panel's careful approach to the question, though I come out the other way. Text, context, case law, and history demonstrate that there is a square-peg-round-hole problem with trying to jam immigration-related habeas petitions into the EAJA in order to require the Executive Branch to pay for aliens' lawyers.

II.

There was an alternative basis for rejecting Petitioners' fee applications, which the government only pressed in the appeal relating to Abioye. In light of the uniquely egregious facts underlying his time in this Country, I am left with the “definite and firm conviction” that the magistrate judge “committed a clear error of judgment in the conclusion it reached upon a weighing of the relevant factors.” Morgan v. Perry, 142 F.3d 670, 683 (3d Cir. 1998). Translation: The magistrate abused his discretion by finding that the government's refusal to give Abioye a bond hearing was not “substantially justified.” 28 U.S.C. § 2412(d)(1)(A). Furthermore, the equitable “special circumstances” exception plainly “make[s] an award unjust.” Id.

A.

The magistrate judges who addressed these issues in the District Court relied on abbreviated summaries of the facts relevant to Abioye, which skipped over some of the information that is relevant to this fact-intensive inquiry.

Abioye is a Nigerian national and Nigeria-barred attorney. He entered the United States in 2018 and overstayed his six-month tourist visa. Between January 2019 and January 2020, he participated in a wire fraud conspiracy that defrauded more than 10 victims of over $4.6 million. Abioye personally obtained $1.5 million of the fraud proceeds. During the course of the conspiracy, Abioye entered into what appears to have been a fraudulent marriage with a U.S. citizen. Abioye lived separately with a paramour, but his wife submitted an unsuccessful application for immigration benefits on his behalf in August 2019.

In January 2020, federal authorities arrested Abioye based on charges relating to the financial fraud. He consented to pretrial detention and remained detained following his July 2020 guilty plea. Abioye's plea agreement stated that he agreed to forfeit at least $1 million and to pay at least $3.3 million in restitution. In May 2022, a district judge sentenced Abioye to 27 months' imprisonment and 24 months of supervised release. The court also ordered Abioye to pay $1,335,923.77 in restitution to victims and an additional $1 million in forfeiture.

Abioye was released that month based on credit for time served, and DHS took him into custody because he had no status in the United States. DHS held Abioye principally at the Moshannon Valley Processing Center. DHS charged Abioye with removability based on his aggravated felony conviction and visa overstay. See 8 U.S.C. §§ 1227(a)(2)(A)(iii), 1227(a)(1)(B). After Abioye's counsel obtained a six-week continuance, Abioye conceded removability on both charges and sought deferral of removal under the Convention Against Torture. An IJ denied Abioye's CAT application in November 2022. The IJ found that aspects of his testimony “strained credulity” but declined to make an adverse credibility finding. A139. Abioye's counsel obtained a three-week extension of the briefing schedule for his BIA appeal. In May 2023, the BIA dismissed the appeal. Abioye petitioned the Fourth Circuit for review of the BIA's decision. At Abioye's request, the Fourth Circuit entered an administrative stay followed by a stay of removal. Those proceedings remain stayed.

In August 2023, DHS notified Abioye that it had reviewed his custodial status and determined not to release him. See 8 C.F.R. § 241.4. DHS explained that Abioye posed a danger to the community in light of his fraud conviction, and noted that he was expected to be removed “in the reasonably foreseeable future,” pending resolution of the Petition For Review in the Fourth Circuit. DHS's notice indicated that Abioye would be subject to another review if the Fourth Circuit's stay of removal had not been lifted within a year.

Abioye did not wait to find out. In October 2023, a “Pro Bono Counsel” filed a “Petition For A Writ Of Habeas Corpus” on Abioye's behalf in the Western District of Pennsylvania pursuant to 28 U.S.C. § 2241. A93, 115. In November 2023, a magistrate judge granted the petition and ordered that an IJ conduct a bond hearing within 30 days. In December 2023, an IJ decided that Abioye did not pose a danger to the community—notwithstanding his multi-million-dollar fraud scheme—or a flight risk—notwithstanding his lack of status and failure to abide by the terms of his original entry.

Abioye moved for attorneys' fees under the EAJA in February 2024. A magistrate judge granted the motion in September 2024, and ordered the government to pay Abioye $18,224.58 in “attorneys' fees and expenses” under the EAJA.

B.

After DHS reviewed Abioye's detention status in August 2023 and concluded that he was still a flight risk, the government was substantially justified in opposing Abioye's October 2023 habeas petition on the basis that his continued detention did not violate due process.

1.

This issue required a magistrate judge to assess whether the government's opposition to a constitutional challenge to § 1226(c), based on a balancing test involving a non-exhaustive list of judge-made factors bearing on Abioye's due process rights, was substantially justified. In light of the practical reality of such a test, the instances in which the answer to that question is “no” should be rare indeed. This case was not one of them.

“Substantial justification” under the EAJA means a “reasonable basis in both law and fact.” Hanover Potato Prods., Inc. v. Shalala, 989 F.2d 123, 128 (3d Cir. 1993). “[A] court cannot assume that the government's position was not substantially justified simply because the government lost on the merits.” Williams v. Astrue, 600 F.3d 299, 302 (3d Cir. 2009). “[A] legal position is substantially justified if it relates to an unsettled or close question of law, but not if it clearly offends established precedent.” Russell v. Heckler, 814 F.2d 148, 153 (3d Cir. 1987).

We look to the positions of the agency and the government litigators, but here they were the same. The government's position was that Abioye was properly detained pursuant to 8 U.S.C. § 1226(c). Getting the government off to a good start, that statute says that the Attorney General “shall” detain an alien subject to removal for an offense like Abioye's financial fraud. Id. § 1226(c)(1)(B). This mandatory detention provision has survived a facial constitutional challenge. See Demore, 538 U.S. at 531, 123 S.Ct. 1708. Section 1226(c) “does not on its face limit the length of the detention it authorizes.” Jennings v. Rodriguez, 583 U.S. 281, 303, 138 S.Ct. 830, 200 L.Ed.2d 122 (2018); see also id. at 305-06, 138 S.Ct. 830 (“§ 1226(c) mandates detention of any alien falling within its scope and that detention may end prior to the conclusion of removal proceedings ‘only if’ the alien is released for witness-protection purposes.”). For these reasons, at least as a statutory matter, the government was on solid footing in law and fact because § 1226(c) is explicitly mandatory and Abioye has a qualifying conviction.

The closer question relates to Abioye's as-applied constitutional challenge to § 1226(c). He argued that continued detention violated due process, as applied to him, under German Santos v. Warden Pike County Correctional Facility. 965 F.3d 203, 210 (3d Cir. 2020); see also Black v. Decker, 103 F.4th 133, 150-51 (2d Cir. 2024). Though binding in this Circuit, German Santos is not without detractors. See Banyee v. Garland, 115 F.4th 928, 932 & n.3 (8th Cir. 2024); see also Black v. Almodovar, 156 F.4th 171, 181-91 (2d Cir. 2025) (Menashi, J., dissenting from the denial of rehearing en banc).

German Santos described “a nonexhaustive list of four factors to consider in assessing whether an alien's detention has grown unreasonable.” 965 F.3d at 211. This EAJA litigation was not the place to reconsider that precedent in light of Banyee and the Black dissentals, and the government did not ask this Court to do that. At the same time, the complexity of the issues presented, arising in part from the ongoing judicial debate, presented novel and close questions relating to a statute that the government was “duty-bound to defend.” Kiareldeen v. Ashcroft, 273 F.3d 542, 549 (3d Cir. 2001); see also Grace v. Burger, 763 F.2d 457, 458 n.5 (D.C. Cir. 1985) (“[S]ituations in which the government's defense of the constitutionality of a federal statute fails the ‘substantially justified’ test should be exceptional.”); Vacchio, 404 F.3d at 675. We should not take lightly the suggestion that the government lacks a substantial justification in carrying out that duty.

Courts adjudicating EAJA fee battles also must be sensitive to the fact that this type of litigation under German Santos takes place in the context of a “grand balancing test in which unweighted factors mysteriously are weighed,” resulting in risks that “ ‘equality of treatment is ․ impossible to achieve; predictability is destroyed; judicial arbitrariness is facilitated; judicial courage is impaired.’ ” June Med. Servs. LLC v. Russo, 591 U.S. 299, 348, 140 S.Ct. 2103, 207 L.Ed.2d 566 (2020) (Roberts, C.J., concurring in the judgment) (quoting Antonin Scalia, The Rule of Law as a Law of Rules, 56 U. Chi. L. Rev. 1175, 1182 (1989)); Qatanani, 144 F.4th at 512 n.20 (Matey, J., dissenting) (urging that discussion of “that notoriously vague phrase [due process] requires precision”). The risks presented by such a standardless standard are apparent in these cases, and even the magistrate judge who issued Abioye's fee award felt that “the Third Circuit has not laid out a specific standard for district courts to follow.” A16.

It is difficult to understand how the government's position in such a setting could be unreasonable. We presume the government acts in good faith, and I am aware of no suggestion to the contrary. Therefore, while precedent requires application of law to fact, the German Santos factor test should not be a fertile field for fee awards benefitting aliens.

2.

The question presented by Abioye's habeas petition, under German Santos, was whether the government denied him due process by continuing to “presume that detention is needed to prevent flight or danger to the community.” 965 F.3d at 209. The due process issue required a “highly fact-specific inquiry.” Id. at 210. The question presented by Abioye's fee application under the EAJA was whether the government was substantially justified in arguing that there was no due process violation within the context of that highly fact-specific inquiry. For that issue, characteristics and factual considerations that distinguished Abioye's case from German Santos were highly relevant to determine whether the government's position was colorable. Thus, before addressing the four factors identified in German Santos, it is necessary to consider extenuating circumstances that weighed strongly in the government's favor.

Unlike in German Santos, Abioye is not a lawful permanent resident. See 965 F.3d at 206; Chavez–Alvarez v. Warden York Cnty. Prison, 783 F.3d 469, 471 (3d Cir. 2015). Abioye overstayed his visa without authorization in violation of U.S. law. See 8 U.S.C. §§ 1182(a)(9)(B), 1202(g). There are strong indications in the record that he attempted to perpetrate a marriage fraud in order to obtain status and prolong his illegal stay. See 8 U.S.C. § 1325(c); 18 U.S.C. 1546.

Within a year of getting here, Abioye commenced a multi-million-dollar fraud. He victimized U.S. nationals during the course of that crime. In the criminal case, Abioye presented indicia of flight risk and danger to the community that were so strong that he consented to detention for approximately 27 months. See 18 U.S.C. §§ 3142-3143. He had not even completed the supervised release component of his sentence when he filed his habeas petition.

Collectively, these circumstances strongly suggest that Abioye's due process rights secured him nothing more than § 1226(c) offered, which is what “Congress has provided by statute.” Thuraissigiam, 591 U.S. at 140, 140 S.Ct. 1959. He never made a valid entry, and he failed to demonstrate the types of allegiance to the sovereign that merit additional protection under our Constitution. See Robles, 2025 WL 2924867, at *4 (Matey, J., dissenting); Qatanani, 144 F.4th at 517-20 (Matey, J., dissenting). Instead, he affirmatively caused a great deal of harm here. Based on these material distinctions between Abioye and the petitioner in German Santos, I disagree with the panel that “[n]othing in the facts here gave the Government a reasonable basis for arguing against a bond hearing.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *11. Rather, the facts make German Santos and related authorities distinguishable in relevant respects that justified the government's position.

3.

The government also had reasonable arguments under the four German Santos factors:

Length Of Detention. Abioye's 16 months of administrative detention compared favorably with our relevant decisions. See German Santos, 965 F.3d at 212 (30 months); Chavez–Alvarez, 783 F.3d at 472 n.4 (3d Cir. 2015) (20 months); Leslie v. AG, 678 F.3d 265, 266 (3d Cir. 2012) (48 months); Diop v. DHS, 656 F.3d 221, 234 (3d Cir. 2011) (35 months).

The magistrate judge who granted Abioye's habeas petition stated that this first German Santos factor was “concededly a close call.” A29. Being on the wrong side of a close call should not result in an EAJA fee award. The magistrate judge who awarded the fees understood that the “exact time” allowable for administrative detention “var[ies] with the facts of the case,” but failed to explain why he did not feel the issue was debatable. A15. Whereas the panel characterized “close call” as an “offhand remark,” it looks to me like a finding. Michelin, ––– F.4th at ––––, 2026 WL 263483, at *12. Regardless, characterizing the magistrate judge's statement as an offhand remark did not provide license to excuse the second judge's failure to explain why he departed from it. See Williams v. Runyon, 130 F.3d 568, 573 (3d Cir. 1997) (“[T]he law of the case doctrine does not limit the power of trial judges to reconsider their prior decisions,” but “the court must explain on the record the reasoning behind its decision to reconsider the prior ruling”); Fagan v. City of Vineland, 22 F.3d 1283, 1290 (3d Cir. 1994) (“Although [law-of-the-case doctrine] does not limit the power of trial judges from reconsidering issues previously decided by a predecessor judge from the same court or from a court of coordinate jurisdiction, it does recognize that as a matter of comity a successor judge should not lightly overturn decisions of his predecessors in a given case.”).

In concluding that the length of detention was indefensible, the panel cited German Santos and Chavez Alvarez for the proposition that “we have held detention became unreasonable sometime between six months and one year after it began.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *11. The panel's reasoning established something very close to the type of “bright-line threshold” we previously disavowed. German Santos, 965 F.3d at 211; Chavez–Alvarez, 783 F.3d at 474 n.7. We seem to have a different threshold for the constitutional speedy-trial rights of presumed-innocent criminal defendants. See, e.g., Conroy v. Leone, 316 F. App'x 140, 144-45 (3d. Cir. 2009) (480 months' pretrial detention); United States v. Taylor, 469 F.2d 284, 285 & n.2 (3d Cir. 1972) (21 months' pretrial detention); United States v. Sodano, 592 F. App'x 114, 115-16 (3d Cir. 2014) (20 months' pretrial detention). In any event, the portion of German Santos cited by the panel expressly acknowledged the lawful permanent resident status of the petitioner in Chavez Alvarez. See 965 F.3d at 211. That petitioner completed his criminal sentence more than 10 years prior to the start of his 20 months of administrative detention. See 783 F.3d at 471. Thus, the timeframes discussed in German Santos and Chavez Alvarez did not foreclose the government's position regarding Abioye, who lacked status, had committed serious crimes much more recently, and was still serving his sentence.

We have instructed District Courts that the goal here is to identify the “tipping point” when it is no longer reasonable for the government to “presume that detention is needed to prevent flight or danger to the community.” German Santos, 965 F.3d at 209. If that is so, I find it hard to quibble with the government's assertion that, for Abioye, his lack of immigration status, criminal conduct, and incomplete criminal sentence warranted a longer presumption of flight risk and danger than was appropriate for the petitioners in German Santos and Chavez Alvarez.

Likelihood of Continued Detention. The government filed its opposition to Abioye's habeas petition on November 14, 2023. Abioye completed briefing in support of his Petition For Review in the Fourth Circuit prior to that filing, on November 9. Argument in the Fourth Circuit had been scheduled for December 12, 2024. Thus, I do not agree that, when the government took the contested legal position by opposing the habeas petition, “detention was likely to continue much longer because of his appeal to the Fourth Circuit.” Michelin, ––– F.4th at ––––, 2026 WL 263483, at *11.

Reasons For The Delay. The government was not unreasonable in taking the position that Abioye bears more of the responsibility for delaying his removal.

This factor did not weigh against the alien in Chavez Alvarez, where “the legal questions were complex and unusual” because the parties were forced to grapple with the application of immigration laws to violations of the Uniform Code of Military Justice. Not true here. The issues are straightforward, and the proceedings following the completion of briefing on the habeas petition are telling. The magistrate judge granted the petition on November 29, 2023. Two weeks before his oral argument in the Fourth Circuit, Abioye convinced the government and the court to hold the appeal in abeyance so that he can search for “what he contends is material new evidence to the BIA via a motion to reopen.” Gov. Br. 6 n.2. Over a year later, that search appears to be ongoing. Abioye's delays appear to be strategic, and they should not be weighed in favor of the government cutting him a check.

This Court has recognized that “[a]n argument could be made that aliens who are merely gaming the system to delay their removal should not be rewarded with a bond hearing that they would not otherwise get under the statute.” Chavez–Alvarez, 783 F.3d at 476; see Black, 156 F.4th at 186 (Menashi, J., dissenting from the denial of rehearing en banc) (“The fact is that ‘an alien detained under § 1226(c) has the keys in his pocket and can end his detention immediately by ‘withdrawing his defense and returning to his native land.’ ” (quoting Banyee, 115 F.4th at 933)). “[T]he legal system ․ is replete with situations requiring the making of difficult judgments as to which course to follow, and, even in the criminal context, there is no constitutional prohibition against requiring parties to make such choices.” Demore, 538 U.S. at 530 n.14, 123 S.Ct. 1708. These delay tactics are Abioye's choices to make, but he should shoulder any resulting costs instead of the government.

Conditions Of Confinement. The magistrate who resolved Abioye's habeas petition found that “this factor weighs in favor of” the government because his conditions of confinement were significantly different than those faced by the petitioner in German Santos. A32. The panel agreed that the government's position regarding this factor was substantially justified, but they felt it was not dispositive of the due process claim. See Michelin, ––– F.4th at ––––, 2026 WL 263483, at *12.

I do not think the conditions issue is dispositive either, but it is quite clear that Abioye faced circumstances that were very different than the petitioner in German Santos. By my count, each of the four German Santos factors at least arguably favored the government. It necessarily follows that the government's position was substantially justified.

C.

Although the government failed to press the point on appeal, Abioye was not entitled to attorneys' fees for the additional reason that “special circumstances make an award unjust.” 28 U.S.C. § 2412(d)(1)(A). “Courts which have found special circumstances have done so on the basis of [1] novel legal issues, or on [2] the equitable basis of a prevailing party's unclean hands.” Brinker v. Guiffrida, 798 F.2d 661, 667 (3d Cir. 1986). This case presents both circumstances.

Many of the same factors bearing on whether the government's position was substantially justified are relevant to the “novel legal issues” prong of the “special circumstances” exception. In addition, “all the circumstances” are relevant to “whether under the facts of this case the equitable considerations dictate an award should not be made.” Taylor v. United States, 815 F.2d 249, 253 (3d Cir. 1987). Abioye's visa overstay, attempted marriage fraud, and actual wire fraud are exactly the type of special circumstances that should bar recovery.

Abioye “would not have been incarcerated in the first place but for his notorious and repeated violations of United States immigration law.” Oguachuba v. INS, 706 F.2d 93, 99 (2d Cir. 1983). In his plea agreement, Abioye agreed to forfeit “at least $1 million” and pay “at least $3,300,000” in restitution to his victims. A269-70. But the fee award at issue requires, in substance, the government to pay Abioye. See Astrue v. Ratliff, 560 U.S. 586, 591-93, 130 S.Ct. 2521, 177 L.Ed.2d 91 (2010). “It is plainly inequitable to allow [petitioner] to flout American law in this fashion and then to require the public fisc to support his legal bills to terminate his detention ․” Oguachuba, 706 F.2d at 99. The EAJA should have functioned as “a ‘safety valve’ directed at protecting the good faith advancement of novel legal theories.” Brinker, 798 F.2d at 668; see also Kiareldeen, 273 F.3d at 549-50.

* * *

The need for courts to review and potentially disagree with the government's German Santos balancing is a natural consequence of the test that our precedent has created. The habeas relief in these cases is unchallenged, even if not unquestionable. Yet it is still true that “proper respect for the political branches' plenary power over immigration has repeatedly moved the courts against second guessing their judgment.” Qatanani, 144 F.4th at 520 n.40 (Matey, J., dissenting). Invasive second-guessing of the government's litigation positions regarding balancing tests, in the context of EAJA applications, is not a path courts should find themselves on very often—if ever. In addition to sovereign immunity, this is another issue that merited en banc review by our Court.

The panel opinion here essentially determines that two individuals charged with a cumulative total of at least eight crimes between them, and each subject to a final removal order, should receive attorneys' fees from the government for the due process claims they filed against continued detention while appealing their removal. See Michelin v. Warden Moshannon Valley Corr. Ctr., ––– F. 4th ––––, ––––, 2026 WL 263483, at *3 (3d Cir. 2026). See also A19, A37, A248-49, A267, A273-75, A452-460. Respectfully, I dissent from the denial of en banc rehearing in these cases.

In addition to a number of grounds thoughtfully raised by colleague Judge Bove and joined by Judge Matey and Judge Phipps, the en banc court should rehear these cases for three principal reasons. First, as Judge Bove's opinion underscores, U.S. Supreme Court precedent instructs that we are to interpret statutes to waive federal sovereign immunity only if such a waiver is clear from the statutory text. See Lac du Flambeau Band of Lake Superior Chippewa Indians v. Coughlin, 599 U.S. 382, 387-88, 143 S.Ct. 1689, 216 L.Ed.2d 342 (2023) (“This clear-statement rule is a demanding standard. If ‘there is a plausible interpretation of the statute’ that preserves sovereign immunity, Congress has not unambiguously expressed the requisite intent.”). The deep circuit split over the question whether a habeas challenge to immigrant detention is a “civil action” meriting government attorney's fees under 28 U.S.C. § 2412(d)(1)(A) belies such clarity. Compare Barco v. Witte, 65 F. 4th 782, 785 (5th Cir. 2023), Obando-Segura v. Garland, 999 F.3d 190, 195 (4th Cir. 2021), with Daley v. Ceja, 158 F. 4th 1152, 1164 (10th Cir. 2025), Vacchio v. Ashcroft, 404 F.3d 663, 668-69 (2d Cir. 2005), and In re Hill, 775 F.2d 1037, 1040-41 (9th Cir. 1985).

Second, the implications of the panel opinion here, and the awarding of attorney's fees in such detention challenges, are weighty and significant. The potential of receiving thousands of dollars in attorney's fees creates a powerful incentive for detained, illegally present individuals to perpetuate and further extend their stay in the United States through the filing of habeas claims. In the four weeks since the panel opinion issued, already at least two district court judges within this circuit have affirmatively encouraged detainees to consider filing for such fees. See, e.g., Cova v. Rose, No. 3:26-CV-101, 2026 WL 376921, at *2 (W.D. Pa. Feb. 11, 2026); Aguirre-Guevara v. Oddo, No. 3:26-CV-70, 2026 WL 376846, at *1 (W.D. Pa. Feb. 11, 2026).

Third, the panel opinion is a backdoor, inadvertent extension of circuit precedent, expanding the category of cases in which courts within this circuit may now conclude that the government's position in immigration detention due process habeas challenges is not “substantially justified,” see 28 U.S.C. § 2412(d)(1)(A). The key past precedent analyzed by the panel in evaluating whether the government's detention position in these cases was “substantially justified” involved a due process challenge where a lawful permanent resident had been detained. See Michelin, ––– F.4th at –––– – ––––, 2026 WL 263483, at **11-13 (evaluating the German Santos factors contributing to whether the government's position is “substantially justified” and therefore not meriting the penalty of attorney's fees). See also German Santos, v. Warden Pike County Correctional Facility, 965 F.3d 203, 206, 210 (3d Cir. 2010). Such individuals may have due process expectations that illegally present individuals lack. See Opinion of Bove, J., supra at 3, 47-52. The individuals awarded attorneys' fees in these instant cases, in direct contrast, had received the opposite adjudication of lawful status and were subject to final orders requiring removal, making their claims for constitutional rights to additional process on U.S. soil more attenuated. See id. at 3. And yet, after the panel opinion for which this court has denied en banc review, the government's continued detention of such individuals while they file additional challenges, must be affirmatively “substantially justified” to avoid an attorney's fees burden. This extension should be reviewed now, before being cemented as circuit precedent. These cases readily merit en banc consideration.

FOOTNOTES

1.   Judges Matey, Porter, Phipps, Bove, and Mascott would grant the petitions for rehearing by the en banc court. Judge Bove, joined by Judges Matey, Porter, and Phipps, files the attached dissent sur denial of rehearing. Judge Mascott files the attached separate dissent sur denied rehearing.

1.   Unless otherwise indicated, case quotations omit all internal citations, quotation marks, footnotes, alterations, and subsequent history. Citations to a “Rule” or “Civil Rule” are to the Federal Rules of Civil Procedure. Citations to “Habeas Rules” are to the Rules Governing Section 2254 Cases in the United States District Courts and Rules Governing Section 2255 Cases in the United States District Courts, which are substantially similar with respect to the propositions for which they are both cited in this opinion.

2.   Consistent with Barco and Obando-Segura, the Second, Fourth, and Tenth Circuits have held that habeas petitions by criminal prisoners are not “civil actions” under the EAJA. See O'Brien v. Moore, 395 F.3d 499, 508 (4th Cir. 2005); Sloan v. Pugh, 351 F.3d 1319, 1322-23 (10th Cir. 2003); Ewing v. Rodgers, 826 F.2d 967, 971 (10th Cir. 1987); Boudin v. Thomas, 732 F.2d 1107, 1111-15 (2d Cir. 1984).

3.   See Paige v. Bacarisse, 80 F. App'x 299, 300 (5th Cir. 2003); Jennings v. Natrona Cnty. Det. Ctr. Med. Facility, 175 F.3d 775, 780 (10th Cir. 1999); Blair-Bey v. Quick, 151 F.3d 1036, 1040 (D.C. Cir. 1998); Anderson v. Singletary, 111 F.3d 801, 805 (11th Cir. 1997); Naddi v. Hill, 106 F.3d 275, 277 (9th Cir. 1997); Reyes v. Keane, 90 F.3d 676, 678 (2d Cir. 1996); Martin v. United States, 96 F.3d 853, 855 (7th Cir. 1996).

4.   Accord H.R. Rep. No. 80-308, at A105, A110-11, A114-15, A118-21, A124-26, A128-32, A133-34, A145 (citing 28 U.S.C. §§ 1252, 1292, 1331, 1332, 1335, 1336, 1337, 1338, 1340, 1341, 1342, 1343, 1344, 1346, 1347, 1348, 1349, 1350, 1351, 1353, 1357, 1392, 1394, 1397, 1399, 1400, 1401, 1402, 1404, 1441, 1652 (1948)).

5.   DOJ, Bureau of Justice Statistics, Federal Review of State Prisoner Petitions Habeas Corpus 2 (Mar. 1984), https://perma.cc/CN7Z-5XTU.

6.   INS, Statistical Yearbook 129 (1980), https://perma.cc/T6M9-F2WN.

THOMAS L. AMBRO Circuit Judge