dissenting:
Although the issue in this case — what precisely constitutes a judgment within the meaning of Rule 58 — seems technical, perhaps even formalistic, the stakes are actually quite high. For litigants, knowing with certainty whether a district court has entered a Rule 58 judgment is critically important. Not only does entry of a Rule 58 judgment start the clock for filing a notice of appeal, Fed. R.App. P. 4(a)(1), but a timely notice of appeal is jurisdictional; parties who fail to file a timely notice of appeal because they are unaware that Rule 58 judgments have been entered lose their right to appeal. See, e.g., Moore v. South Carolina Labor Bd., 100 F.3d 162, 163 (D.C.Cir.1996).
“Prior to 1963, there was considerable uncertainty over what actions of the District Court would constitute an entry of judgment, and occasional grief to the litigants as a result of this uncertainty.” United States v. Indrelunas, 411 U.S. 216, 220, 93 S.Ct. 1562, 36 L.Ed.2d 202 (1973) (per curiam). “To eliminate these uncertainties, which spawned protracted litigation over a technical procedural matter,” id., Rule 58 was amended in 1963 to read: “Every judgment shall be set forth on a separate document. A judgment is effec*42tive only when so set forth.... ” Fed. R.CrvP. 58. Known as the “separate document rule,” amended Rule 58 makes “clear that a party need not file a notice of appeal until a separate judgment has been filed and entered.” Bankers Trust Co. v. Mollis, 435 U.S. 381, 385, 98 S.Ct. 1117, 55 L.Ed.2d 357 (1978) (per curiam). As the Supreme Court has recognized, the rule change “would be subject to criticism for its formalism were it not for the fact that something like this was needed to make certain when a judgment becomes effective, which has a most important bearing ... on the time for appeal.” Indrelunas, 411 U.S. at 220, 93 S.Ct. 1562 (internal quotation marks omitted). Accordingly, the Court has held, the Rule must be “mechanically applied in order to avoid new uncertainties.” Id. at 222, 93 S.Ct. 1562.
The Supreme Court has recognized only one exception to the rule’s “mechanical” application. In Bankers Trust, the Court confronted the issue whether the absence of a conforming Rule 58 judgment rendered a notice of appeal premature, depriving the appeals court of jurisdiction even though the district court had clearly intended to enter final judgment. Because “[t]he rule should be interpreted to prevent loss of the right of appeal, not to facilitate loss,” the Court held that “[t]he need for certainty as to the timeliness of an appeal ... should not prevent the parties from waiving the separate judgment requirement where one has accidentally not been entered.” Bankers Trust, 435 U.S. at 386, 98 S.Ct. 1117 (emphasis added) (internal quotation marks omitted).
Heeding the Supreme Court’s directive to apply Rule 58 strictly to protect appeal rights, we have held that “a combined decision and order” cannot serve as a “judgment” within the meaning of Rule 58. See Diamond v. McKenzie, 770 F.2d 225, 230 n. 10 (D.C.Cir.1985) (distinguishing between “decisions” of the district court and Rule 58 final judgments). In Pack v. Burns International Security Service, 130 F.3d 1071, 1071-72 (D.C.Cir.1997) (per cu-riam), we found that a document “stating that the motion to dismiss would be treated as conceded and granted” and giving several reasons for dismissal failed to comply with Rule 58. We reached the same conclusion in Diamond, 770 F.2d at 229-30, with respect to a document captioned “order” that gave the “basis, albeit briefly, of the court’s reasoning, along with citations to legal authorities.” The caption and the presence of the court’s reasoning (including its citation to authority), we noted, meant that the order differed in “at least two significant respects” from Model Forms 31 and 32, which were passed contemporaneously with the amendment to Rule 58. Id. at 229 n. 9.
It is precisely this kind of uncertainty about whether the District Court intended to enter a final order that warrants the mechanical application of Rule 58. By mechanically applying this rule, as the Supreme Court has taught us to do, a court may avoid speculation as to whether an appellant should or should not have known that the time for appeal had begun to run.
Id. at 230 (footnote omitted). ,We concluded: “While we do not mean to suggest that the model form of judgment is the only means of complying with Rule 58, adherence to [that] format ... would be of considerable assistance in eliminating uncertainty as to the nature of the District Court’s action.” Id. at 229 n. 9.
Applying Diamond and Pack and taking account of Rule 58’s purpose, I believe that the district court’s order in this case represents a combined decision and order, not a Rule 58 judgment. As in Diamond, the order differs from Form 32 in two respects. See id. First, it is not labeled “Judgment.” Instead, as were fifty-nine other documents - issued by the district court and the magistrate judge in this case, some of which did no more than grant extensions of time, it was labeled “Order.” Were this the order’s only flaw, I would agree with the court that this *43insignificant departure from the model forms would not prevent us from finding that the order complies with Rule 58. See Maj. Op. at 41 n.2. But the order departs from the Model Forms in a second, “most critieal[]” (Diamond’s words) respect: it contains the district court’s reasoning. Because Kidd had objected to the magistrate’s recommendation and because Fed. R.Civ.P. 72 requires the district court to “make a de novo determination upon the record, or after additional evidence, of any portion of the magistrate judge’s disposition to which specific written objection has been made,” the district court explained its reasoning: “In her objection, Plaintiff still fails to raise any genuine issues of material fact in this recent filing and does not set forth any arguments that would cause the Court to reject Magistrate Judge Robinson’s Report and Recommendation.”
The only difference I can discern between this order and the order in Diamond, which we concluded was a combined decision and order, is that the Diamond order contained three sentences of reasoning, not one, together with several citations. In view of Rule 58’s purpose— providing certainty — and our obligation to interpret Rule 58 “to -prevent loss of the right of appeal, not to facilitate loss,” Bankers Trust, 435 U.S. at 386, 98 S.Ct. 1117 (emphasis added), these two differences cannot serve as a basis for distinguishing this case from Diamond. From a Rule 58 standpoint, the critical point is that the orders in both cases contained the district court’s reasoning. It makes no difference that the district court in this case sets forth its reasoning in one sentence whereas the district court in Diamond used three. Suppose that instead of stating its reasoning in a single sentence, the district court here had broken its one sentence into two: “In her objection, Plaintiff still fails to raise any genuine issues of material fact[. I]n this recent filingf, she] does not set forth any arguments that would cause the Court to reject Magistrate Judge Robinson’s Report and Recommendation.” Would this departure from the “one sentence” rule my colleagues seem to adopt have persuaded them that the order was a combined decision and order?
This court also says that the order is a Rule 58 judgment because it lacks even the single citation it reads Diamond to permit. Although this “single citation” standard presumably means that the order would have passed as a Rule 58 judgment even if the district court had cited, for example, Anderson v. Liberty Lobby, Inc., 477 U.S. 242, 106 S.Ct. 2505, 91 L.Ed.2d 202 (1986), the oft-used authority for Rule 56’s summary judgment standard, suppose the district court had also included a citation to one of our cases for the same proposition or even to Rule 56 itself? Would my colleagues then have concluded this was not a Rule 58 judgment? Rule 58’s amendment was intended to end just this type of uncertainty.
The court quotes Chief Judge Robinson’s admonition that “trivial departures must be tolerated in the name of common sense.” See Maj. Op. at 39 (quoting Diamond, 770 F.2d at 234 (Robinson, C.J., concurring)). I agree. They also point to his statement that an order “does not lose its character as a separate document simply because it includes a one-sentence explanation of what the order is all about, or a recital that a magistrate’s report and recommendation are being adopted.” Diamond, 770 F.2d at 234 (Robinson, C.J., concurring) (footnote omitted). I agree with this too. But the Kidd order goes beyond both “a one-sentence explanation of what the order is all about” and a “recital” that Magistrate Judge Robinson’s report and recommendation are being adopted. Id. It sets forth the district court’s reasons for adopting the report. Indeed, Chief Judge Robinson distinguished between the simple recitals he referred to as “trivial departures,” i.e., “explanation^] of what the order is all about,” and the court’s reasoning: “On the other hand, orders combining the court’s *44directives with its statement of factual findings or legal conclusions plainly cannot pass muster as separate documents.” Id.
Suggesting that Chief Judge Robinson could not possibly have meant what he said, my colleagues conclude that the sentence “cannot be read to preclude a single sentence of explanation because the entire panel, with Judge Robinson in full agreement, had already found some minimal amount of legal reasoning to be consistent with Rule 58.” Maj. Op. at 39. But listen again to Chief Judge Robinson’s words. “[S]tatement[s] of factual findings or legal conclusions,” he said, “plainly cannot pass muster as separate documents.” Diamond, 770 F.2d at 234 (Robinson, C. J., concurring). And far from tolerating “minimal” reasoning, the Diamond court concluded that the order was not a Rule 58 judgment because it “provid[ed] the basis, albeit briefly, of the court’s reasoning, along with citations to legal authorities.” Diamond, 770 F.2d at 229-30.
The standard my colleagues now adopt — an order is a Rule 58 judgment so long as it contains only “sparse” reasoning, Maj. Op. at 39 — will require this court, contrary to the Supreme Court’s warning against “case-by-case tailoring of the ‘separate document’ provision,” Indrelunas, 411 U.S. at 221, 93 S.Ct. 1562, to develop a common law of “sparseness.” This will in turn produce the very uncertainty and “protracted litigation over a technical procedural matter” that Rule 58’s amendment was designed to end. Id. at 220, 93 S.Ct. 1562.
In my view, litigants, district courts, and this court would be better served by a bright line rule: Orders containing the district court’s reasoning, three sentences or one, several citations or none, are not judgments within the meaning of Rule 58. To comply with such a rule, the district court need only instruct its clerk to issue judgments that adhere to the essence of Model Forms 31 and 32. Such a procedure would provide the certainty Rule 58 demands, prevent accidental loss of appeal rights, and ensure that this court will never again have to address this issue.
Because I would hold that Kidd’s appeal is timely, I would reach the merits of her case and reverse. Even a cursory review of Kidd’s pro se filings should have demonstrated to the magistrate judge who disposed of Kidd’s case in three conclusory sentences that Kidd’s primary allegation— that she was denied a promotion by the very District of Columbia officials against whom she had just won a substantial verdict for intentional infliction of emotional distress stemming from sexual harassment and forced sodomy by her supervisor— presented a strong prima facie case of retaliatory failure to promote as well as sufficient evidence of pretext to survive summary judgment.
In response, my colleagues offer still another reason why this court cannot reach the merits of Kidd’s case. They point out that the magistrate judge found that Kidd had failed to comply with Local Rule 108(h); citing Twist v. Meese, 854 F.2d 1421, 1424 (D.C.Cir.1988), they also point out that “[t]he rule expressly provides that in the event of noncompliance material facts identified by the opposing side may be deemed admitted.” Maj. Op. at 41 n.1 (emphasis added). But unlike in Twist, where the district court “relied on Rule 108(h) to hold that the material facts identified by the government were deemed admitted,” 854 F.2d at 1424, the magistrate judge here chose not to deem the District’s facts admitted, much less to rest her summary judgment recommendation on Kidd’s failure to comply with Local Rule 108(h). Instead, the magistrate judge considered Kidd’s evidence: “[T]he undersigned finds that plaintiff has failed to identify even a single material fact as to which a genuine issue for trial exists. Rather, both her motion and her opposition to defendants’ motion are almost entirely comprised of her opinions of defendants’ actions and characterizations of various statutes and decisions.” The dis*45trict court likewise did not rely on Local Rule 108(h): “In her objection, Plaintiff still fails to raise any genuine issues of material fact in this recent filing and does not set forth any arguments that would cause the Court to reject Magistrate Judge Robinson’s Report and Recommendation.” Because neither the magistrate judge nor the district court relied on Kidd’s failure to comply with the rule, nothing bars this court from reaching the merits of her case.
The D.C. Court of Appeals described the facts leading up to Kidd’s sexual harassment lawsuit as follows. Shortly after Kidd began working for the Department of Administrative Services, her immediate supervisor, one Melvin Carter, “began calling her at home and once asked her when she would make love to him.” King v. Kidd, 640 A.2d 666, 659 (D.C.1993). “Carter showed Kidd documents demonstrating that she was a probationary employee” and told her that “because of her probationary status, she could be fired at his recommendation and that no one would question anything.” Id. He then called her and “ordered her to come to a nearby hotel.” Id. Kidd hung up, but he called back, reminding her of her probationary status. She went to the hotel where they had sex. Id. Carter continued to “pur-su[e] her,” becoming “angry and loud” when she refused to have sex with him again. Carter also “took away her computer and her clerical assistant.” Id. at 660. Kidd gave in and again had sex with Carter. “According to Kidd’s testimony, Carter ... forcibly sodomized her, rupturing her anal tissues and causing her to fear AIDS.” Id.
Kidd complained to Carter’s direct supervisor, Robert King, about “stress, harassment and mistreatment.” Id. Receiving no relief, Kidd filed suit in Superi- or Court against Carter, King, Raymond Lambert (the director of the Department of Administrative Services), and the District alleging sexual harassment and intentional infliction of emotional distress. A jury found Carter, King, and Lambert personally liable for intentional infliction of emotional distress and Carter liable for sexual harassment as well. The jury awarded Kidd $300,000 in compensatory and punitive damages.
Kidd then returned to work. Believing the environment of the Department to be hostile — King and Lambert, both of whom had been found personally liable to her, were still there — Kidd applied for a transfer. Her request was denied. Kidd also applied for a promotion to a level DS-12. This request also was denied, and record evidence suggests that Lambert, one of the defendants in the Superior Court lawsuit who at the time was personally liable to Kidd for $260,000, took part in the decision to deny the promotion.
To establish a prima facie case of retaliation, Kidd must show that she engaged in activity protected by Title VII, that the District took an adverse employment action against her, and that the adverse action was causally related to the exercise of her rights. See Cones v. Shalala, 199 F.3d 612, 521 (D.C.Cir.2000). Kidd satisfies all three elements of this standard: Her prior sexual harassment lawsuit is protected activity under Title VII; the District failed to promote her; and her evidence of causal connection — that the person denying her the promotion was personally liable to her for $260,000 at the time he denied the promotion — is overwhelming. Indeed, because Kidd’s prima facie case “strongly suggests intentional discriminationf, it] may be enough by itself to survive summary judgment.” Aka v. Washington Hosp. Center, 156 F.3d 1284, 1289 n. 4 (D.C.Cir.1998) (en banc). See also Texas Dep’t of Community Affairs v. Burdine, 450 U.S. 248, 255 n. 10, 101 S.Ct. 1089, 67 L.Ed.2d 207 (1981) (“[T]here may be some cases where the plaintiffs initial [prima facie] evidence, combined with effective cross-examination of the defendant, will suffice to discredit the defendant’s explanation.”).
*46Even if Kidd’s prima facie case standing alone were insufficient, her evidence that the District’s asserted nondiscriminatory reasons were pretextual, when considered in combination with the strength of her prima facie case, was enough to preclude summary judgment. See Aka, 156 F.3d at 1289 (“Assuming then that the employer has met its burden of producing a nondiscriminatory reason for its actions, the focus of proceedings at trial (and summary judgment) will be whether the jury could infer discrimination from the combination of (1) the plaintiffs prima facie case; (2) any evidence the plaintiff presents to attack the employer’s proffered explanation for its actions; and (3) any further evidence of discrimination that may be available to plaintiff Responding to the District’s contention that she was not promoted from DS-11. to DS-12 because the Mayor had frozen hiring and promotions, Kidd submitted a sworn statement that her supervisor “requested a waiver of the Mayor’s Order to promote another employee.” She also stated that “[t]he freeze apparently was arbitrary because Gary Hicka was brought on board as a DS-12 realty specialist ... to fill one of the vacant positions.” From this evidence that the Department made exceptions to the freeze for other employees, a-jury could have concluded that the District’s explanation was pretext for retaliation. See Cones, 199 F.3d at 519.
The District also claimed that Kidd was unqualified for a promotion to the DS-12 level, but Kidd provided evidence that in her DS-11 position, she performed work normally assigned to DS-12 employees. In fact, the District does not dispute that after a restructuring of the Department, many of the tasks she had been performing were taken from her and reassigned to employees at the DS-12, 13, 14, and 15 levels.
I have no idea whether a jury would have returned a verdict in Kidd’s favor. But viewing the record in the light most favorable to Kidd and drawing all inferences in her favor, as we must at this stage of the case, I have no doubt the case should have gone to a jury.
In reaching this conclusion, I recognize the indications in the record that Kidd may have been a difficult plaintiff — the district court tried repeatedly to appoint counsel, and in each instance, counsel withdrew. I also recognize, as the magistrate judge found, that Kidd’s pro se opposition to defendants’ motion for summary judgment was neither succinct nor particularly successful at separating her legal arguments from her factual assertions. In contrast, we have had the benefit of amicus’s masterful appellate brief. But even without that brief, had the magistrate judge carefully reviewed Kidd’s pleadings, as she is required to do in pro se cases, see Haines v. Kemer, 404 U.S. 519, 520, 92 S.Ct. 594, 30 L.Ed.2d 652 (1972) (per curiam) (holding pro se pleadings to less stringent standards than formal pleadings drafted by lawyers), she too would have seen that Kidd had a strong prima facie case of retaliation and sufficient evidence of pretext to preclude summary judgment.
I respectfully dissent.