It is axiomatic that the judiciary’s mandate is to “say what the law is.”^1×1. Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803). And the axiom is no less true when an agency has already weighed in on the question.^2×2. See, e.g., Matthew C. Stephenson & Adrian Vermeule, Essay, Chevron Has Only One Step, 95 Va. L. Rev. 597, 601–02 (2009) (explaining that the role of the courts in conducting a Chevron analysis still requires identifying a range of permissible statutory interpretations). Recently, in Bowers v. United States Parole Commission,^3×3. 775 F. App’x 504 (11th Cir. 2019). the Eleventh Circuit shut down a federal prisoner’s challenge to a parole denial that was based on the United States Parole Commission’s (Commission) reading of its statutory authority.^4×4. See id. at 517–18. The plaintiff in Bowers challenged the Commission’s restrictive interpretation of § 4206(d) of the Parole Commission and Reorganization Act^5×5. Parole Commission and Reorganization Act, Pub. L. No. 94-233, § 2, 90 Stat. 219, 224 (1976) (codified at 18 U.S.C. § 4206(d) (1982)) (repealed 1984). (Parole Act), which is the Act’s mandatory parole provision.^6×6. See Bowers, 775 F. App’x at 517. Section 4206(d) outlines the requirements for mandatory parole, as well as the select instances in which the Commission is obligated to deny parole. See id. at 508. But rather than address the interpretive dispute presented by Bowers, the court exploited ambiguity in the Commission’s decision to avoid identifying the contested meaning of the statute.^7×7. See id. at 517–19. Specifically, the court declined to establish exactly how the Commission had interpreted the mandatory parole provision before applying it to Bowers’s case and then perfunctorily labeled the Commission’s decision a mere “application” of the statute.^8×8. See id. at 519. In so doing, the Bowers court effectively abdicated its responsibility to clarify what the Parole Act means. The court’s reasoning is problematic, not only for its distortion of the judiciary’s relationship to the administrative state, but also for its potential to hamper individuals’ ability to challenge agency interpretations of statutes.

In 1974, Veronza Bowers was sentenced to life imprisonment for the murder of a park ranger.^9×9. See Bowers v. Keller, 651 F.3d 1277, 1283 (11th Cir. 2011). Five years into his sentence, he attempted (and failed) to escape from prison.^10×10. Id. Bowers and another inmate climbed an interior fence but were halted by tower gunfire, which created a risk of injury to other inmates in the area. Id. But since then, Bowers has maintained an exemplary disciplinary record.^11×11. See Bowers, 775 F. App’x at 508 (noting that Bowers “had not been the subject of a disciplinary report since 1988”). He earned an associate’s degree and was described by a chaplain at the prison as having “the most positive attitude that could be imagined.” Id. On April 7, 2004, after serving thirty years, he became eligible for mandatory parole under § 4206(d) of the since-repealed Parole Act.^12×12. See id. The Parole Act was repealed with the enactment of the Federal Sentencing Guidelines, but its requirements still apply to anyone sentenced prior to the Guidelines’ effective date. See Walden v. U.S. Parole Comm’n, 114 F.3d 1136, 1138 (11th Cir. 1997). In May 2005, the Commission granted Bowers parole,^13×13. See Bowers, 651 F.3d at 1286. only to reopen the case in June and reverse its position.^14×14. See id. at 1287–88. The case was reopened at the request of the Attorney General; notably, this was the first time in the Commission’s history that an Attorney General had sought review of a Commission determination. See id. at 1287. It decided that Bowers had “seriously violated” prison rules with his escape attempt twenty-five years prior, triggering an exception to mandatory parole.^15×15. Id. at 1288; see id. at 1285–88. Section 4206(d) sets a timeline for mandatory parole, provided that “the Commission shall not release such prisoner if it determines that he has seriously or frequently violated institution rules and regulations.” 18 U.S.C. § 4206(d) (1982) (repealed 1984) (emphasis added).

Bowers brought a series of habeas petitions to secure his original parole grant.^16×16. See Bowers, 775 F. App’x at 508–15. In response to the first, an Eleventh Circuit panel vacated the 2005 parole denial and remanded the case to the Commission.^17×17. See Bowers, 651 F.3d at 1296. The Eleventh Circuit held that the June 2005 reopening of Bowers’s case had been unlawful because a Commissioner indicated clear bias against Bowers, violating the Commission’s mandate of neutrality. See id. at 1292–96. On remand, the Commission reopened the case but once again voted to deny parole.^18×18. See Bowers v. Keller, No. 08-cv-02095, 2012 WL 12964326, at *3 (N.D. Ga. Sept. 24, 2012). Another Eleventh Circuit panel granted Bowers discovery and leave to amend, after which he filed the amended petition at issue here, making two principal arguments.^19×19. See Bowers v. U.S. Parole Comm’n, 760 F.3d 1177, 1185 (11th Cir. 2014) (granting Bowers leave to amend and discovery on the impact that political pressure may have had on the 2011 denial). First, he alleged the Commission violated his due process rights by acting with bias.^20×20. Bowers v. Drew, No. 08-CV-2095, 2016 WL 9107434, at *5 (N.D. Ga. June 28, 2016). Bowers alleged that the Commission, instead of maintaining its statutorily mandated political independence, had been influenced by the Department of Justice in its decisionmaking. See id. at *7–8. Second, he argued the Commission violated both the Parole Act and its rules and regulations by denying him parole, and he specifically challenged the Commission’s interpretation of a “serious[]” infraction under § 4206(d) as erroneously precluding a consideration of how much time had passed since his last rule violation.^21×21. See id. at *10–13. Bowers argued that his offense necessarily became less serious because it was followed by a period of good behavior. Id. at *10.

The United States District Court for the Northern District of Georgia dismissed the petition on both grounds.^22×22. Id. at *15. It held that Bowers failed to prove specific bias because he provided no evidence of undue influence on the Commission from the Department of Justice (DOJ) or Attorney General.^23×23. See id. at *7–9. The court dismissed the latter challenge on the grounds that the Commission’s interpretation of “seriously” in § 4206(d) was not a “flagrant” abuse of the Commission’s discretion.^24×24. Id. at *12.

The Eleventh Circuit affirmed.^25×25. Bowers, 775 F. App’x at 522. Writing for the panel, Judge Julie Carnes^26×26. Judge Carnes was joined by Judges Jordan and Schlesinger. Judge Schlesinger was sitting by designation from the Middle District of Florida. held that the Commission did not apply an incorrect interpretation of “seriously” in § 4206(d) and thus did not violate the Parole Act.^27×27. See Bowers, 775 F. App’x at 515. The court began with the standard of review. If the Commission’s parole denial involved a disputed interpretation of § 4206(d), that would require a Chevron^28×28. Chevron U.S.A. Inc. v. Nat. Res. Def. Council, Inc., 467 U.S. 837 (1984). or Skidmore^29×29. Skidmore v. Swift & Co., 323 U.S. 134 (1944). deference analysis; however, if the denial comprised a mere application of the statute, the Commission’s decision would be reviewed for abuse of discretion.^30×30. See Bowers, 775 F. App’x at 515–17. Bowers alleged his case fit the former mold.^31×31. Id. at 517. He argued that the Commission interpreted “seriously” too restrictively because it read the statute as precluding consideration of the time that had passed since the rule violation.^32×32. See id. at 518. Bowers believed the Commission was considering only whether the violation was serious when it was committed, rather than evaluating it in the context of the prisoner’s record. Id. But in his view, a violation is necessarily less “serious” if followed by a period of good behavior. Id. The Commission sought abuse of discretion review, but it argued that even under Chevron, its interpretation of “seriously” — which required examining only the gravity of the offense in isolation from elapsed time — would be perfectly reasonable.^33×33. See Brief of Appellees at 20, 34, Bowers, 775 F. App’x 504 (No. 16-15737); see also Bowers, 775 F. App’x at 517–18.

Judge Carnes first held that neither Chevron nor Skidmore deference was applicable because the parties did not disagree about the statute’s meaning.^34×34. Bowers, 775 F. App’x at 518 (“[T]here is no disputed interpretation of the statute to referee.”). Specifically, she explained, there was no evidence that the Commission had actually adopted an interpretation of § 4206(d) that precluded a consideration of time.^35×35. See id. For support, the court pointed to a line in the Commission’s Denial Letter: “The passage of time does not diminish the gravity of this rule violation.”^36×36. Id. The letter contained notice of denial and an explanation for Bowers. Id. at 517. Judge Carnes acknowledged that both the Commission’s General Counsel and the agency in litigation articulated a different, stricter view: that “seriously” did not allow a consideration of time.^37×37. See id. at 519; Brief of Appellees, supra note 33, at 17, 22–33 (making various arguments in support of the Commission’s construction of the statute, including that the statute does not mandate that the Commission “discount the seriousness of an incident based on the passage of time from the date of the violation to the date of the parole review,” id. at 26). But the government’s litigating position did not bind the court. The court ultimately looked to the relevant line in the Denial Letter to conclude that the agency agreed with Bowers on the meaning of § 4206(d).^38×38. Bowers, 775 F. App’x at 518 (explaining that both parties seemed to agree that “one serious rule violation” was sufficient to deny parole, and that the Commission agreed with Bowers that “the passage of time could play a role” in characterizing a violation as “serious”). Having found no dispute as to the law itself, the court turned to whether the Commission “abused its discretion” by denying parole.^39×39. See id. Bowers had the burden of showing that the Commission’s application of § 4206(d) involved a “flagrant, unwarranted, or unauthorized action.”^40×40. Id. (quoting Glumb v. Honsted, 891 F.2d 872, 873 (11th Cir. 1990) (per curiam)). The court found that Bowers fell short under that standard.^41×41. Id. It noted the standard’s deferential nature and highlighted Bowers’s concession that his attempted escape created significant danger.^42×42. See id. at 518–19; see also supra note 10.

Judge Carnes then assessed whether the Commission violated Bowers’s due process rights by failing to review his case in an unbiased manner.^43×43. See Bowers, 775 F. App’x at 519. To succeed on a due process claim based on a lack of impartiality, Bowers had to show that “the probability of actual bias . . . [was] too high to be constitutionally tolerable.” Id. at 520 (quoting Withrow v. Larkin, 421 U.S. 35, 47 (1975)). The court concluded that it had not — Bowers had not produced evidence of any specific bias, nor had he shown that DOJ’s prior positions on his parole tainted the Commission’s eventual review.^44×44. Id. at 522. Yes, DOJ had invested significant time and energy into Bowers’s parole, but because the Commission is an independent agency housed within DOJ, some level of interaction between the two bodies was unavoidable and so not dispositive on the bias question.^45×45. See id. Judge Jordan concurred. Id. at 522–23 (Jordan, J., concurring). Section 4207 of the Parole Act, he wrote, states that the Commission can consider any reasonably available information in making parole decisions, which supports the conclusion that the Commission did consider all of Bowers’s prison history. See id. (citing 18 U.S.C. § 4207 (1982) (repealed 1984)).

The Bowers court’s equivocation about whether and how the Commission interpreted § 4206(d) left open a central question in the case. The court failed to clarify how the Commission weighs the passage of time in determining the seriousness of a rule violation, if it does so at all. This is troubling for individuals who are approaching eligibility for mandatory parole, as they will have little clarity about how the Commission does or should make decisions. But Bowers is also concerning for its cursory — and ultimately inadequate — engagement with a quintessentially legal issue: agency statutory interpretation. In jumping to an abuse of discretion standard of review when faced with an interpretive dispute, the circuit abdicated its principal responsibility to say what the law is. That move calls into question courts’ effectiveness at checking unreasonable administrative action, and it may well heighten the burden for individuals challenging agency decisions.

The court should have addressed the key dispute on the meaning of “seriously.” Instead, it held that scrutiny of the agency’s interpretation was unnecessary,^46×46. See id. at 517–18 (majority opinion). but neither of its two justifications for doing so was convincing. First, the court was wrong to imply that an examination of the agency’s interpretation was dispensable because there was no interpretive disagreement between the parties for it to “referee.”^47×47. Id. at 518 (“Given . . . the statement in the Denial Letter indicating . . . that the passage of time could play a role in a decision to characterize a particular violation as ‘serious,’ we conclude that there is no disputed interpretation of the statute to referee.” (emphasis added)). Second, the court should not have found that the Commission merely applied, and did not interpret, the statute.^48×48. See id. at 519.

The court erred in finding that the agency adopted Bowers’s interpretation of the statute, as the Commission’s arguments during litigation and its internal communications suggest a disagreement. In the court’s view, the agency agreed with Bowers that, when evaluating if a prisoner “seriously” violated prison rules, it had to account for the time passed since the violation in question.^49×49. See id. at 518 (pointing out that “the Commission saw fit to add . . . a statement suggesting that for another instance of putative, serious misconduct, the passage of time might well result in a conclusion that the violation” was not serious). But only limited evidence suggests that the agency used this interpretation; at best, the record was inconclusive as to how the Commission interpreted § 4206(d). First, throughout litigation, the agency actively contested Bowers’s interpretation. Bowers argued that assessing the seriousness of a violation requires the Commission to consider the prisoner’s record, because a violation can become less “serious” if followed by good behavior.^50×50. Appellant’s Brief at 31, Bowers, 775 F. App’x 504 (No. 16-15737) (“Conduct deemed serious at the time it was committed may be viewed much differently after decades of . . . unblemished behavior.”). The agency pushed back, arguing that seriousness turns on the gravity of the offense at the time it was committed.^51×51. Brief of Appellees, supra note 33, at 26 (“Nothing in the statute itself mandates that the Parole Commission discount the seriousness of an incident based on the passage of time . . . . [U]nder the plain language of § 4206(d) . . . the Commission should consider whether an inmate has any history of institutional rule violations . . . .” (emphasis added)). Second, the agency’s counsel had written a memo to Commissioners stating that “seriously” precludes a consideration of time since a violation.^52×52. The memo stated that “the word ‘seriously’ in § 4206(d) does not allow the Commission to consider the antiquity of a particular rule violation.” Bowers, 775 F. App’x at 519 (emphasis added). Lastly, at least one Commissioner who voted to deny parole testified to having adopted that interpretation.^53×53. See id. at 519 n.10. Another Commissioner suggested she did not weigh time in this case. Id.

To avoid the dispute reflected in the evidence, the court relied almost exclusively on the Denial Letter, but that letter contained enough ambiguity that the court should have at least remanded for the Commission to clarify its interpretive posture. The circuit specifically relied on one sentence: “The passage of time does not diminish the gravity of this rule violation.”^54×54. Id. at 518. But that line’s meaning is not facially clear. Does it mean that time is relevant generally but was not enough in this case? Or that time is always irrelevant and so had no mitigating effect in Bowers’s case, like all others? The court read the line as evidence of the former: that the Commission considered Bowers’s record and determined that the passage of time, while relevant, did not ultimately outweigh the gravity of his particular rule violation.^55×55. See id. (pointing to this line to conclude that the Commission believed that the passage of time could “play a role” in finding a particular violation “serious”). But it is equally plausible the line means the opposite: perhaps the Commission was emphasizing that Bowers’s record was not relevant to the “seriousness” of his violation, because time is never a consideration.^56×56. Prior to Bowers, the Commission had never issued a rule or regulation elaborating on its interpretation of the word “seriously.” See id. at 519. As such, there was no “official position” to which the court could point in refusing to accept the agency’s litigating position. This distinguishes Bowers from Bowen v. Georgetown University Hospital, 488 U.S. 204 (1988), which the court referenced. Bowers, 775 F. App’x at 517 (citing Bowen, 488 U.S. at 212). In Bowen, the Supreme Court refused to defer to the litigating position of the Department of Health and Human Services on its interpretation of the Medicare Act because that position was “wholly unsupported by regulations, rulings, or administrative practice” and contrary to the position taken by the agency itself in prior cases. Bowen, 488 U.S. at 212. Notably, Bowen is the only case that the Eleventh Circuit cited in support of its proposition that it was not bound to accept the Parole Commission’s litigating position. See Bowers, 775 F. App’x at 517. It is unclear which of these is the best reading. But in light of the other available evidence, it seems more plausible that the Commission interpreted “seriously” consistent with its litigating position: that the passage of time is irrelevant. If that was the case, the court should have stepped in to decide whether that reading was the best, or at least a “reasonable,” construction of the statute. At the very least, the court should have acknowledged the Denial Letter’s ambiguity and remanded to ask the Commission to clarify its interpretive posture.

The court’s second justification for avoiding review of the agency’s interpretation of the Act — that the agency merely applied § 4206(d)^57×57. See Bowers, 775 F. App’x at 519 (“[W]e need not consider whether [the General Counsel’s] interpretation is accurate or merits deference. Instead, we look to the statements made by the Commission in its Denial Letter and . . . conclude that the Commission’s factual findings and application of the Parole Act did not constitute an abuse of discretion.” (emphasis added)).  — was question-begging. Of course the Commission applied the statute when it made a determination about Bowers’s parole — every parole adjudication necessitates applying a statute to individualized facts.^58×58. 18 U.S.C. § 4203(b) (1982) (repealed 1984) (establishing that the role of the Commission is to grant or deny parole to individual prisoners “pursuant to the procedures set out” in the Parole Act). But how did the Commission interpret the word “seriously” in that statute before applying it? Both sides litigated that central question,^59×59. Appellant’s Brief, supra note 50, at 25–44; Brief of Appellees, supra note 33, at 22–36. yet the court sidestepped it. The opinion did emphasize that the agency never announced it would be interpreting the statute.^60×60. See Bowers, 775 F. App’x at 519 (explaining that the Commission did not promulgate a rule adopting its General Counsel’s interpretation). That is hardly dispositive, though; agencies frequently propagate interpretations in the course of adjudication.^61×61. See, e.g., Clark-Cowlitz Joint Operating Agency v. Fed. Energy Regulatory Comm’n, 826 F.2d 1074, 1081 (D.C. Cir. 1987) (“The general principle is that when as an incident of its adjudicatory function an agency interprets a statute, it may apply that new interpretation in the proceeding before it.” (emphasis added)); see also Owen M. Fiss, Objectivity and Interpretation, 34 Stan. L. Rev. 739, 739 (1982) (arguing that “[a]djudication is interpretation”); cf. NLRB v. Wyman-Gordon Co., 394 U.S. 759, 765 (1969) (“Adjudicated cases may and do, of course, serve as vehicles for the formulation of agency policies, which are applied and announced therein.”); SEC v. Chenery Corp., 332 U.S. 194, 202 (1947) (“Not every principle essential to the effective administration of a statute can or should be cast immediately into the mold of a general rule.”). NLRB v. Hearst Publications,^62×62. 322 U.S. 111 (1944). which addressed whether newsboys were “employees” under the National Labor Relations Act,^63×63. 29 U.S.C. §§ 151–169 (2012). offers a high-profile example.^64×64. Hearst Publications, 322 U.S. at 120. There, the agency applied a broad statutory term to specific facts. The Court deferred to the agency’s decision, but only after analyzing whether the interpretation of “employee” applied by the NLRB had “ample basis in the law.”^65×65. Id. at 132; see also, e.g., Inv. Co. Inst. v. Conover, 790 F.2d 925, 926–27, 932 (D.C. Cir. 1986) (analyzing as an issue of statutory interpretation the Comptroller of the Currency’s finding that units of interest in a bank trust did not constitute “securities” for the purposes of a statute). In short, interpretations need not be declared to be treated as such, and the Bowers court never explained why it could assume that the Commission did not, in applying the statute to Bowers’s case, incidentally interpret it, too.^66×66. The court came closest to an explanation by pointing to the line in the Denial Letter stating that “[t]he passage of time does not diminish the gravity of this rule violation.” Bowers, 775 F. App’x at 518. The court pointed to the use of the word “this” as evidence that the Commission had made a factual determination that time in this case was irrelevant. See id. This may well be correct, but it still skirts the question of how the agency interpreted “seriously.”

Notwithstanding the path of reasoning, the result was clear: the Bowers court avoided the difficult question of how the Commission interpreted § 4206(d). In failing to nail down and then apply judicial scrutiny to the agency’s interpretation of the statute, the court abdicated its primary responsibility to clarify the meaning of statutory mandates for administrative agencies. Courts have long recognized that if they have any job, it is to “say what the law is.”^67×67. E.g., King v. Burwell, 135 S. Ct. 2480, 2496 (2015) (quoting Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803)); see also Brett M. Kavanaugh, Fixing Statutory Interpretation, 129 Harv. L. Rev. 2118, 2154 (2016) (book review) (arguing that even when deference is required, “courts should determine whether the agency’s interpretation is the best reading of the statutory text”); cf. Philip Hamburger, Chevron Bias, 84 Geo. Wash. L. Rev. 1187, 1205–10 (2016) (calling into question the constitutionality of Chevron deference for causing judges to abandon their mandate to exercise independent judgment). The Bowers court, however, not only declined to clarify what the law of mandatory parole is when it comes to “serious” rule violations and the passage of time, but also failed to even conduct a deferential Chevron or Skidmore review to determine whether the agency had applied a “reasonable construction” of the statute.^68×68. Indeed, after a lengthy exposition of both doctrines, the court did not once use either one. Bowers, 775 F. App’x at 515–17. The proper standard of review is beyond the scope of this comment, but it is worth mentioning that it is not entirely clear that the Commission’s interpretation would have merited any deference at all, had the court chosen to review it, as the agency did not go through any formal rulemaking or adjudication process prior to making its decision. Instead, it avoided an analysis of the Commission’s interpretation altogether.^69×69. See id. at 519 (determining that because the agency never promulgated a rule or regulation endorsing its counsel’s interpretation, the court “need not consider whether that interpretation is accurate or merits deference”). Notably, the court failed to complete even the necessary predicate step to analyzing the reasonableness of an agency’s interpretation: clarifying and establishing what that interpretation actually was. The Bowers decision thus quietly cast a shadow over a critical aspect of the judiciary’s relationship to the administrative state: the court’s implied mandate to check agency action by identifying the statutory space within which agencies may act.^70×70. See, e.g., Stephenson & Vermeule, supra note 2, at 602 (“Chevron supposes that interpretation is an exercise in identifying the statute’s range of reasonable interpretations, a range that opens up a ‘policy space’ within which agencies may make reasoned choices . . . .” (footnote omitted)). Given that the Parole Commission is an independent and politically unaccountable body, this facet of the Bowers opinion should elicit particular pause.^71×71. See Randolph J. May, Defining Deference Down: Independent Agencies and Chevron Deference, 58 Admin. L. Rev. 429, 449–50, 453 (2006) (arguing that deference to independent agencies has relatively weak legal grounding, at least compared to more politically accountable agencies); see also Elena Kagan, Presidential Administration, 114 Harv. L. Rev. 2245, 2376–77 (2001) (same). The Commission is an independent agency housed within DOJ. See 18 U.S.C. § 4202 (1982) (repealed 1984).

The Bowers court’s approach is also troubling because it risks limiting recourse for individuals seeking to challenge an agency’s statutory interpretation. The circuit’s reasoning leads to the following perilous scenario: where an agency does not explicitly announce that it has interpreted a statute while applying it, the court will treat the action as if it raises no interpretive dispute — even if the parties agree that they disagree on the meaning, as they did in Bowers.^72×72. Bowers, 775 F. App’x at 519. But this makes it nearly impossible to challenge any parole-like decision where interpretation of a statute and application of its terms occur simultaneously. For litigants, the impact of such a result is magnified where the alternative to judicial review of an agency interpretation is a standard as deferential as “abuse of discretion.”

The stakes of interpretation in this case — namely, Bowers’s freedom — were exceedingly high. Yet, the court failed to complete its role of adjudicating not just the most obvious disputes, but all issues of law presented before it, including difficult interpretive questions. The Parole Act’s repeal notwithstanding, the Bowers court’s reasoning thus raises salient issues. It highlights the importance of judicial review of statutory interpretation that occurs incidental to agency actions. It cautions against the ease with which a court might avoid such review. And it offers a glimpse of the consequences such avoidance can have.