with whom Judges SLOVITER, MCKEE, AMBRO and FUENTES join, concurring in part and dissenting in part.
I concur with the majority that we must reverse the dismissal of Otero’s ineffective assistance of counsel claim and remand for the District Court to decide whether Otero is indigent and qualifies for appointed counsel. I concur with the majority that, subject to notice and an opportunity to be heard, a district court may dismiss a habe-as petition sua sponte as part of its preliminary consideration pursuant to Rule 4 of the rules governing both section 2254 and section 2255 cases. I also concur with the majority that the one year statute of limitations contained in the Antiterrorism and Effective Death Penalty Act, 28 U.S.C. §§ 2244(d)(1), 2255, is subject to waiver as a defense by the government. See Robinson v. Johnson, 313 F.3d 128, 134 (3d Cir.2002). But thereafter the majority and I part company. I dissent from the majority’s holding that a district court has the authority to raise AEDPA’s statute of limitations sua sponte after the government has failed to raise that affirmative defense in its answer to the petition.
I.
As an initial matter, I would remand Bendolph’s petition for a factual finding on the equitable tolling claim he raised on appeal. As explained by the majority, the Clerk of the Supreme Court mistakenly filed Bendolph’s untimely petition for cer-tiorari because the date on the copy of our judgment entry had been altered, making the petition appear timely. Because the Clerk did not reject his certiorari petition as untimely, Bendolph believed, in error, that AEDPA’s statute of limitations did not begin to run on his habeas petition until one year from the date the Supreme Court denied certiorari. Bendolph, therefore, did not file his habeas petition within one year of the date his conviction actually became final as AEDPA requires. See 28 U.S.C. § 2255. On appeal before the panel, he argued that the mistaken filing of his certiorari petition by the Clerk of the Supreme Court is grounds for equitable tolling of AEDPA’s statute of limitations. Because it is unclear who altered the record, we cannot know whether equitable tolling is appropriate and we should remand for further factual findings.
AEDPA’s statute of limitations is subject to equitable tolling. Fahy v. Horn, 240 F.3d 239, 244 (3d Cir.2001). Out of a sense of equity, we will toll the statute of limitations when a petitioner demonstrates reasonable diligence in bringing a claim, but “has in some extraordinary way been prevented from asserting his or her rights.” Miller v. New Jersey State Dep’t of Corr., 145 F.3d 616, 618. In non-capital cases, “attorney error, miscalculation, inadequate research, or other mistakes have not been found [by this Court] to rise to the ‘extraordinary’ circumstances required for equitable tolling.” Fahy, 240 F.3d at 244. We have also held, however, that there are “narrow circumstances in which misbehavior of an attorney may merit such equitable relief.” Seitzinger v. Reading Hosp. & Med. Cent., 165 F.3d 236, 239 (3d Cir.1999). For instance, in Nara v. Frank, 264 F.3d 310, 320 (3d Cir.2001), a habeas petitioner accused his attorney of leading *171him to believe that the attorney would file a habeas petition on his behalf, and of telling him that there were no time constraints on habeas petitions. Id. We found that, if true, the allegations “may constitute extraordinary circumstances to justify equitable tolling.” Id. Nevertheless, because the Court was unable to determine the validity of the accusations, it remanded for a factual finding. Id.
Like in Nora, a remand is presently appropriate. Although it is unclear who altered the date on this Court’s judgment entry, the alteration misled the Clerk of the Supreme Court to believe that Ben-dolph’s petition for certiorari was timely. It is entirely possible that the alteration is the result of clerical error. Indeed, there is no evidence in the record that Ben-dolph’s federal public defender altered the document herself, innocently or otherwise. Nor is there evidence that she purposefully or knowingly submitted an altered document to the Clerk of the Supreme Court. If the federal public defender altered the document, her actions would be sufficiently serious to constitute an “extraordinary” prevention of Bendolph’s ability to assert his rights that would merit equitable tolling. I believe it patently unfair of the majority to acknowledge the altered date and then simply affirm the District Court without permitting Bendolph the opportunity to show that he was misled into believing he had time to file his habeas petition. Because the record does not permit us to determine who is responsible for the alteration, we should remand for the District Court to make factual findings, and for it to determine whether the statute of limitations should be tolled. Under these circumstances, there is a possibility that Bendolph’s petition was indeed timely.
II.
In the event the petitions were untimely, Rule 5(b) of the Rules Governing Section 2254 Cases in the United States District Courts requires that an answer to a habe-as petition “must state whether any claim in the petition is barred by a failure to exhaust state remedies, a procedural bar, non-retroactivity, or a statute of limitations.” Section 2254 Rule 5(b) (emphasis added).24 The phrase “statute of limitations” was added to Rule 5(b) in December 2004 as part of a “stylistic” change meant to reflect the existing state of the law. See Section 2254 Rule 5, cmt.
At the time of the amendment, the Courts of Appeals for the Sixth, Ninth, and Eleventh Circuits had addressed the effect of the government’s failure to raise the AEDPA statute of limitations in its answer on a district court’s authority to subsequently raise the issue sua sponte. In Scott v. Collins, 286 F.3d 923, 927 (6th Cir.2002), the Sixth Circuit held that because AEDPA’s statute of limitations is an affirmative defense, “Rule 8(c) of the Federal Rules of Civil Procedure requires that [the government] raise it in the first responsive pleading to avoid waiving it.”25 *172The Court held that the government’s failure to raise the statute of limitations in its answer divests the district court of its authority to raise the defense sua sponte. Id. at 930. It explained that a contrary rule would amount to “an impermissible curing of [the government’s] waiver.” Id. The Ninth Circuit applied similar reasoning in Nardi v. Stewart, 354 F.3d 1134 (9th Cir.2004). It held that a “district court lacks the authority to sua sponte dismiss a habeas petition as time-barred after the state files an answer which fails to raise the statute of limitations defense.” Id. at 1141. By contrast, the Eleventh Circuit, in Jackson v. Secretary for the Department of Corrections, 292 F.3d 1347, 1349 (11th Cir.2002), held without any analysis that the government’s failure to raise the statute of limitations in its answer has no effect on a district court’s ability to raise the issue sua sponte. Thus, at the time of this Court’s decision in Long v. Wilson, 393 F.3d 390 (3d Cir.2004), there existed a circuit split. In Long, we sided with the Eleventh Circuit, id. at 403, and today the majority reaffirms that choice, albeit in the context of a section 2255 petition rather than a section 2254 petition. We are on the wrong side of the split between the circuits.
Underlying the Scott and Nardi decisions is the rule that generally it is not appropriate for a court to sua sponte raise non-jurisdictional defenses not raised by the parties. See Acosta v. Artuz, 221 F.3d 117, 122 (2d Cir.2000) (“Generally, courts should not raise sua sponte nonjurisdic-tional defenses not raised by the parties.”); cf. Zelson v. Thomforde, 412 F.2d 56, 58 (3d Cir.1969) (holding that a court may not raise the defense of lack of personal jurisdiction — a non-jurisdictional defense because it does not concern the power of the court to entertain the suit — once the defendant has waived the issue by appearing). This rule exists because ours is an adversarial system, which relies on advocacy by trained counsel. Cf. United States v. Burke, 504 U.S. 229, 246, 112 S.Ct. 1867, 119 L.Ed.2d 34 (1992) (Scalia, J., concurring) (“The rule that points of law not argued will not be considered is more than just a prudential rule of convenience; its observance, at least in the vast majority of cases, distinguishes our adversary system of justice from the inquisitorial one.”). In an adversarial system, it is not for the courts to bring to light the best arguments for either side; that responsibility is left to the parties themselves. McNeil v. Wisconsin, 501 U.S. 171, 181 n. 2, 111 S.Ct. 2204, 115 L.Ed.2d 158 (1991) (“What makes a system adversarial rather than inquisitorial is ... the presence of a judge who does not (as an inquisitor does) conduct the factual and legal investigation himself, but instead decides on the basis of facts and arguments pro and con adduced by the parties.”) (emphasis added). As the Supreme Court has explained, “[t]he determination of what may be useful to the defense can properly and effectively be made only by an advocate.” Dennis v. United States, 384 U.S. 855, 875, 86 S.Ct. 1840, 16 L.Ed.2d 973 (1966).
We should decline to raise non-jurisdictional defenses for another reason as well: fairness to the other party. Typically, it is not fair for courts to act as surrogate counsel for one side but not the other. See United States v. Pryce, 938 F.2d 1343, 1352 (D.C.Cir.1991) (Silberman, J., dissent*173ing in part) (“We thus ordinarily have no right to consider issues not raised by a party in either briefing or argument ... because of the unfairness of such a practice to the other party.... ”); see infra, Part III.A.
Of course I acknowledge that there are exceptions to this rule of restraint. The Supreme Court has held that, in the interests of comity and federalism, courts may raise a habeas petitioner’s failure to exhaust state remedies sua sponte, even if the government does not. Granberry v. Greer, 481 U.S. 129, 131, 107 S.Ct. 1671, 95 L.Ed.2d 119 (1987); cf. Szuchon v. Lehman, 273 F.3d 299, 321 n. 13 (3d Cir.2001) (holding that in the interests of comity and federalism a court may raise procedural default sua sponte). And in Acosta v. Artuz, a case relied upon heavily by the Court in Long, the Second Circuit held that courts may raise non-jurisdictional defenses sua sponte where “the defenses implicate values beyond the interests of the parties.” Acosta, 221 F.3d at 121. The Acosta Court held that a district court may raise AEDPA’s statute of limitations sua sponte because it believed that defense fosters finality of convictions, promotes judicial efficiency and conservation of resources, and ensures the accuracy of judgments by resolving challenges to convictions while the record is fresh. Id. at 123. However, when a court examines a habeas petition for facial deficiencies pursuant to Rule 4, and finding none orders the government to answer, only to have the government fail to raise timeliness as a defense, each of those values become substantially less significant. When measured against the purpose and history of habeas corpus in our jurisprudence, those values shrink even further.
■ Initially, the Long Court noted that in considering the values beyond the interests of the parties served by AEDPA’s statute of limitations, comity and federalism are the “most important[ ].” Long, 393 F.3d at 403. The logic behind this statement renders it irrelevant. Unlike the defenses of exhaustion and procedural default, statutes of limitation do not implicate the interests of comity and federalism. Those defenses are designed to ensure that state courts have “an initial opportunity to pass upon and correct alleged violations of prisoners’ federal rights.” Duckworth v. Serrano, 454 U.S. 1, 3, 102 S.Ct. 18, 70 L.Ed.2d 1 (1981). The same cannot be said for a statute of limitations. When a state prisoner’s habeas petition is dismissed as untimely, the prisoner is not permitted to file his action in state court. Instead, he simply loses the ability to test the legality of his confinement. In the case of a section 2255 petition filed by a federal prisoner — as we have here — state courts are not involved in the trial and sentencing process whatsoever, and thus federalism and comity are simply irrelevant.
Similarly, judicial efficiency and economy are not implicated by a post-answer sua sponte dismissal pursuant to AEDPA’s statute of limitations. If a district court reviews a petition for deficiencies pursuant to Rule 4 and does not dismiss it, and the government then fails to raise the statute of limitations, the parties and the court begin to expend time and effort on the merits of the petition. At this point, judicial efficiency and economy are already lost. The very purpose of affirmative defenses, such as the statute of limitations, is to conserve judicial resources by requiring the parties to raise them early in the proceedings. Robinson, 313 F.3d at 134. Rescuing the government from its folly or inadvertence by permitting a district court to raise the statute of limitations after the parties have begun to address .the merits of the petition subverts that purpose, and may even have the opposite effect. If the *174statute of limitations is raised late in the proceedings, the parties must then brief yet an additional issue, which consumes the resources of all involved. I recognize that by permitting a district court to dismiss a habeas petition as untimely, some resources may be saved. But the same can be said for dismissing any lawsuit at any point. Yet it is not the general practice of this Court to raise the statute of limitations or other affirmative defenses either in civil or criminal cases in order to conserve judicial resources. Courts exist to resolve disputes. The resources of the courts should be used to that end. And what better use of judicial resources (or higher calling for the federal judiciary for that matter) could there be than ensuring that no one is incarcerated in violation of the law?
The majority makes a great deal of the interests of finality AEDPA was enacted to serve. There is no doubt that habeas corpus “presents a tension between the desire for finality, an end to the proceedings, and a desire to provide the chance for revisions, especially to correct errors.” Erwin Chemerinsky, Thinking About Habeas Corpus, 37 CASE W. RES. L. REV. 748, 789 (1987). As Professor Chemerin-sky has pointed out, “[tjhere obviously must be finality at some point; a time when the defendant has had sufficient opportunities for review and the chances of finding an error are too remote to justify further expenditure of resources.” Id. But on the other hand, conviction of the innocent or conviction in violation of the Constitution should not be tolerated. See id. Even those who have strongly emphasized a preference for curbing the scope of habe-as corpus have acknowledged that “conventional notions of finality” have diminished significance in the context of habeas. See Henry J. Friendly, Is Innocence Irrelevant? Collateral Attack on Criminal Judgments, 38 U. CHI. L. REV. 142, 150 (1970) (quotation omitted).
At the earliest, AEDPA’s statute of limitations begins to run when a prisoner’s conviction becomes final, upon the conclusion of direct review. See 28 U.S.C. §§ 2244, 2255. Indeed, it may well not begin for decades if, for instance, the petitioner relies on a retroactively applicable rule of constitutional law as recognized by the Supreme Court. 28 U.S.C. §§ 2244(d)(1)(C), 2255. Moreover, AED-PA’s statute of limitations is subject to equitable tolling. Fahy, 240 F.3d at 244. Many timely habeas petitions, therefore, might not be heard until years after conviction. In such a situation, the finality of the conviction is put in jeopardy — which, after all, is the very purpose of habeas review — as much as when a habeas petition is untimely filed and the government fails to raise a statute of limitations defense. Thus, concerns of finality are no more implicated after the government answers without raising timeliness than by many, if not most, habeas proceedings where timeliness is not an issue.
“In our adversary system, it is enough for judges to judge.” Dennis, 384 U.S. at 875, 86 S.Ct. 1840. Advocacy is best left to the parties. Thus, as a general rule, courts should not raise non-jurisdictional defenses on behalf of parties who do not raise the defenses for themselves. Acosta, 221 F.3d at 122. The Supreme Court has permitted departures from that general rule only when values beyond the interests of the parties are implicated. See Granberry, 481 U.S. at 131, 107 S.Ct. 1671; see also Szuchon, 273 F.3d at 321 n. 13. No such values are sufficiently implicated by a sua sponte dismissal for lack of timeliness where, as here, the government has failed to raise the statute of limitations in its answer.
*175III.
A.
Even if a sua sponte dismissal of the petitions in this case would serve values beyond the interests of the parties, I would still dissent because of the damage dismissal would work upon two countervailing values the majority fails to consider: the related notions of waiver and fundamental fairness.
The majority’s decision renders the concept of waiver a nullity. We held in Robinson, 313 F.3d at 134, that AEDPA’s statute of limitations is subject to waiver. In both Bendolph and Otero the government explicitly waived that defense.26 And while the majority reaffirms our holding in Robinson, at the same time it effectively destroys that portion of Robinson by holding that the government’s waiver is irrelevant. If a court may raise a non-jurisdictional defense once it has been waived, then waiver has no practical effect; it exists in theory alone. See Haskell v. Washington Township, 864 F.2d 1266, 1273 (6th Cir.1988) (explaining that if a district court may raise the statute of limitations sua sponte once it has been waived, “the waiver aspect of Rule 8(c) would have little meaning”). Thus, as a result of the majority opinion, it appears that AEDPA’s statute of limitations is no longer subject to waiver.27 I doubt very much whether courts will be so forgiving of a procedural error by a habeas petitioner.28
As for the issue of fairness, Pliler v. Ford, 542 U.S. 225, 124 S.Ct. 2441, 159 L.Ed.2d 338 (2004), is instructive. In Pli-ler, the Supreme Court reversed a decision by the Ninth Circuit, which required district courts to give pro se habeas petitioners a warning that if they dismiss their petitions in order to first exhaust state remedies, AEDPA’s statute of limitations might bar them from refiling a future petition in federal court. According to the Ninth Circuit, the failure to provide the pro se habeas petitioner in that case with such a warning deprived him of the opportunity to make a “meaningful” choice concerning his petition. Id. at 2445 (citing Ford v. Hubbard, 330 F.3d 1086, 1102 (9th Cir.2003)). The Supreme Court disagreed. It held thát district courts need not warn pro se litigants that AEDPA’s statute of limitations might preclude them from filing any future petitions if they withdraw a timely petition. Id. at 2446. According to the Court, explaining habeas procedure and calculating the statute of limitations are tasks normally reserved for trained *176counsel, and “[Requiring district courts to advise a pro se litigant in such a manner would undermine the district judges’ role as impartial decisionmakers.” Id.
While one might argue that it is the pro se habeas petitioners — who are without the assistance of trained counsel — who need to be warned by district courts of AEDPA’s labyrinthine pitfalls most of all, the Court’s decision in Pliler is clear: district courts may not act as de facto counsel in habeas proceedings. Fairness dictates that we apply this rule equally to both sides, but the majority does not. The majority permits a district court to act as de facto counsel for the government, working together toward the common goal of dismissal of the petition. That is unfair. A habeas petitioner gets no help .from the courts, and the government needs none either. If, as Justice Thomas explained in Pliler, calculating the statute of limitations is a job ordinarily reserved for trained counsel, then that job should be left in the able hands of the government’s attorneys. Any other rule contravenes the fundamental notion that “the judiciary is on no side.... We judges must be strictly neutral with respect to all cases that come before us.” Pryce, 938 F.2d at 1352 (Silberman, J., dissenting in part).
The Court today permits and perpetuates a double standard. Cf. Latorre v. United States, 193 F.3d 1035, 1042 n. 3 (8th Cir.1999) (Lay, J., concurring in result only) (“It is ironic ... that in a § 2255 case a petitioner may not raise an issue on appeal when the issue has been procedurally defaulted in the district court, yet the court expressly ignores this principle when dealing with the United States government.”). It also sends a disturbing message: We will aid the government in a habeas proceeding, but not the petitioner. The mistakes of the government may be excused and ignored, but any error by the petitioner is fatal — perhaps figuratively, but too often literally. The favored-party status the majority affords to the government in habeas cases undermines both the appearance and fact of judicial neutrality, and I will have no part in it.29
B.
Habeas corpus is not a luxury or an extravagance to be tolerated only when convenience permits. It is a fundamental protection of liberty “against arbitrary and wrongful imprisonment” that predates these United States. Chemerinsky, at 749. The Framers viewed it as “the highest safeguard of liberty,” Smith v. Bennett, 365 U.S. 708, 712, 81 S.Ct. 895, 6 L.Ed.2d 39 (1961) — a protection against arbitrary punishment and convictions to be “provided for in the most ample manner,” THE FEDERALIST NO. 83 (Alexander Hamilton). Since the 1970s, however, the road to habeas relief has “become a narrow! ], more tortuous track among concealed snake-pits and anti-personnel mines calculated to daze cartographers and daunt a modern Gilgamesh.” Anthony G. Amsterdam, Foreward to JAMES S. LIEBMAN & RANDY HERTZ, FEDERAL HABE-AS CORPUS PRACTICE AND PROCEDURE, at v (2d ed.1994). We continue to pile on “petty procedural barriers,” result*177ing in a “Byzantine morass of arbitrary, unnecessary, and unjustifiable impediments to the vindication of federal rights.” Coleman v. Thompson, 501 U.S. 722, 758-59, 111 S.Ct. 2546, 115 L.Ed.2d 640 (1991) (Blackmun, J., dissenting). This, in my view, is a great tragedy.
I constantly counsel myself and my-law clerks that somewhere in the mass of usually convoluted, often marginally-comprehensible pro se habeas petitions, there is another Clarence Earl Gideon, or one of the other faceless names for whom we do issue the Great Writ. Searching for those meritorious petitions is not only our duty, it is one of our most important. I am not alone in believing that the writ of habeas corpus may be the single most significant protection of individual rights in the Constitution. See Chemerinsky, at 749. I am simply adding my voice to the chorus. While technically the writ exists only as a procedural device, “its history is inextricably intertwined with the growth of fundamental rights of personal liberty.” Fay v. Noia, 372 U.S. 391, 399-400, 83 S.Ct. 822, 9 L.Ed.2d 837 (1963), abrogated on other grounds by Coleman, 501 U.S. at 722, 111 S.Ct. 2546. A threat to the writ’s vitality is a threat to those very fundamental rights. “It must never be forgotten that the writ of habeas corpus is the precious safeguard of personal liberty and there is no higher duty than to maintain it unimpaired.” Bowen v. Johnston, 306 U.S. 19, 26, 59 S.Ct. 442, 83 L.Ed. 455 (1939). In the name of reducing the docket of the district courts, and seduced by the lure of a one-line order, the majority evades its sacred duty. I dissent.
. Although Rule 5(b) of the Section 2255 Rules does not require the government to state whether any statute of limitations applies, as the majority points out, "we have followed the practice, whenever we decide an AEDPA issue that arises under § 2254 and the same holding would analytically be required in a case arising under § 2255, or vice versa, of so informing the district courts.” Miller, 145 F.3d at 619 n. 1. I agree with the majority that on the question of a district court's continuing authority (or lack thereof) to raise the statute of limitations sua sponte after the government's answer, we should treat section 2254 and section 2255 cases the same. I reach a different conclusion as to that question, of course.
. The Federal Rules of Civil Procedure may be made applicable to habeas cases "to the extent they are not inconsistent with any statutory provisions or [the Habeas] rules.” Sec*172tion 2254 Rule 11; see Section 2255 Rule 12. Applying Civil Procedure Rule 8(c), which mandates that parties set forth any affirmative defenses they might have — including the statute of limitations — in a responsive pleading, is consistent with the mandate of Rule 5(b) of the Section 2254 Rules that the government shall state in its answer whether the petition is barred by a statute of limitations.
. In United States v. Otero, 02-2624, counsel for the United States stated during oral argument before the panel: “We waived the statute of limitations. There’s no question under this Court's Robinson decision that we waived the statute of limitations.” (Oral Argument Tr. at 36). Similarly, in United States v. Bendolph, 01-2468, counsel for the United States explained, "we waived the statute of limitations.” (Oral Argument Tr. at 56).
. Incidentally, had Congress wanted to protect AEDPA’s statute of limitations from waiver, it could have done so by statute as it. did for the defense of failure to exhaust state remedies. Pursuant to 28 U.S.C. § 2254(b)(3), "[a] State shall not be deemed to have waived the exhaustion requirement ... unless the State, through counsel, expressly waives the requirement.” Congress has afforded the defense of AEDPA's statute of limitations no such protection, and even if it had, counsel for the United States explicitly waived the defense in each of these consolidated cases. The majority simply excuses that waiver, instead vesting in the district courts the discretion whether to raise the statute of limitations sua sponte in any given case. That grant of authority is likely to create vast disparities concerning the treatment of government waiver between judges and from one case to the next. Those disparities will in turn be visited upon petitioners.
.Indeed, courts routinely dismiss petitions, standing upon mere punctilios in the law.
. I do not believe my colleagues in the majority intend to be less principled or less fair than I. I do believe, however, that the majority sacrifices a little of both principle and fairness at the altar of expedience. Courts have admittedly crowded dockets. Those dockets include many habeas petitions, and I fear that the petitions with merit are outnumbered by those without. Any rale that would permit district courts to more easily dispose of habeas petitions might therefore seem desirable. But when individual liberty is at stake — as is the case with any habeas petition — expedience and self-interest ought not inform our decisions.