dissenting:
I am unable to agree with my colleagues that the District’s actions in this unfortunate case are protected by the doctrine of sovereign immunity. Accordingly, I respectfully dissent.
I. GENERAL LEGAL PRINCIPLES
A. Burden of Proof.
Before discussing the evidence of record, I think it useful to state my understanding of the applicable legal principles. I begin with an issue which the trial judge found it unnecessary to decide, namely, the proper assignment of the burden of proof.
It is, I think, undisputed that the District is the only party which had unrestricted access to all of the information relevant to its claim of immunity. What the District’s employees did, why they did it, and what factors they considered, are facts readily known only to the District. “[I]t has been established as a general rule that the burden of proof lies on the person who wishes to support his case by a fact which lies peculiarly within his knowledge, or of which he is supposed to be cognizant.” Selma, R & D.R. Co. v. United States, 139 U.S. 560, 568, 11 S.Ct. 638, 640, 35 L.Ed. 266 (1891) (citations omitted); see also ITSI TV Productions, Inc. v. Agricultural Ass’n. 3 F.3d 1289, 1292 (9th Cir.1993). “The ordinary rule, based on considerations of fairness, does not place the burden upon a litigant of establishing facts peculiarly within the knowledge of his adversary.” Browzin v. Catholic University of America, 174 U.S.App.D.0. 60, 66 n. 12, 527 F.2d 843, 849 n. 12 (1975) (quoting United States v. New York N.H. & H.R. Co., 355 U.S. 253, 256 n. 5, 78 S.Ct. 212, 214-15 n. 5, 2 L.Ed.2d 247 (1957)).1
Applying this principle to a governmental claim of immunity, the Supreme Court of Michigan has explained that
whether a governmental agency was engaged in a governmental function when performing the act complained of is a question best known to the agency and best asserted by it. It naturally follows that plaintiffs need not plead facts in avoidance of immunity, but that it is incumbent on the agency to assert its immunity as an affirmative defense.
McCummings v. Hurley Medical Center, 433 Mich. 404, 446 N.W.2d 114, 117 (1989) (per *455curiam). The Supreme Court of New Jersey has held that the burden is on the public entity both to plead and prove its immunity. Kolitch v. Lindedahl, 100 N.J. 485, 497 A.2d 183, 189 (1985) (construing New Jersey statute). The federal appellate courts which have considered the question have unanimously concluded that when the “discretionary function” exception to the Federal Tort Claims Act is invoked as a defense against a facially sufficient complaint, the United States bears the burden of proving that the particular governmental action falls within the scope of that exception. See Prescott v. United States, 973 F.2d 696, 701-02 (9th Cir.1992) (citing authorities). “If the government desires to rely upon any of [the] provisions [exempting it from liability for the exercise of a discretionary function], it has a right to do so in defense of the action, providing such defense is aptly pleaded and proven.” Stewart, supra note 1, 199 F.2d at 520 (emphasis added).2
“In selecting and adopting a general plan of public improvement ... the municipal corporation exercises judicial discretion, but in carrying out the plan it acts ministerially, and must perform the work in a reasonably safe and skillful manner.” District of Columbia v. Caton, 48 App.D.C. 96, 104-05 (1918); see also Elgin v. District of Columbia, 119 U.S.App.D.C. 116, 118-19, 337 F.2d 152, 154-55 (1964). The dispositive question is whether the District proved by a preponderance of the evidence that the acts and decisions of which Aguehounde complains were discretionary rather than ministerial.
B. Discretionary and Ministerial Functions.
The majority has outlined in some detail the distinction between discretionary and ministerial governmental functions. I have no quarrel with much of its discussion. I cannot agree, however, that because the determination of the duration of the clearance interval at Wisconsin and Fessenden was a de facto part of the District’s overall traffic plan, the decision to set it at four seconds was necessarily a protected discretionary act. In my opinion, this proposition, followed to its logical conclusion, would insulate the District from liability for negligent conduct under circumstances which the discretionary function doctrine was never designed to reach.
As Judge (later Attorney General) Griffin Bell wrote for the court in Fowler v. Southern Bell Telephone & Telegraph Co., 343 F.2d 150 (5th Cir.1965), “the defense of sovereign privilege imposes a drastic impingement on personal liberty, and is recognized only because the impingement is considered justified in order to encourage public officials to fearlessly discharge the duties of their office.” Id. at 154. Not every governmental decision which involves some exercise of discretion is insulated from review, for “it would be difficult to conceive of any official act, no matter how directly ministerial, that did not admit of some discretion in the manner of its performance, even if it involved only the driving of a nail.” 18 Eugene McQuillin, the Law of Municipal CORPORATIONS, § 53.04.10, at 157 (3d ed.1993) (quoting Ham v. Los Angeles, 46 Cal.App. 148, 189 P. 462, 465 (1920)).
The goal of the discretionary function exception is to “prevent judicial ‘second-guessing’ of legislative and administrative decisions grounded in social, economic and political policy through the medium of an action in tort.” United States v. Varig Airlines, 467 U.S. 797, 814, 104 S.Ct. 2755, 2764-65, 81 L.Ed.2d 660 (1984). “The discretionary function exception applies only to conduct that involves the permissible exercise of policy judgments.” Berkovitz v. United States, 486 U.S. 531, 537, 108 S.Ct. 1954, 1959, 100 L.Ed.2d 531 (1988). Accordingly, when that exception is invoked, the court must make “a particularized and fact-specific inquiry to determine whether the acts or omissions in question flowed from a choice based on economic or social policy.” Prescott, supra, 973 *456F.2d at 700. That inquiry is designed to determine not whether the government employee has a choice, but whether that choice is a policy judgment. Id. (citations omitted).
Moreover, there is persuasive authority for the proposition that governmental conduct is insulated from liability only “where the official or employee actually exercised some discretion.” 18 McQuillin, supra, § 53.04.10, at 157 (emphasis added).
Immunity for “discretionary” activities serves no purpose except to assure that courts refuse to pass judgment on policy decisions in the province of coordinate branches of government. Accordingly, to be entitled to immunity the state must make a showing that such a policy decision, consciously balancing risks and advantages, took place. The fact that an employee normally engages in “discretionary activity” is irrelevant if, in a given case, the employee did not render a considered decision.
Johnson v. State, 69 Cal.2d 782, 73 Cal.Rptr. 240, 249 n. 8, 447 P.2d 352, 361 n. 8 (1968) (emphasis added); 18 McQuillin, supra, § 53.04.10 at 160 n. 7.3 Although some decisions reflect a different focus,4 proof that an administrative decision was not based on economic or social considerations is surely probative on the question whether the “discretionary function” exception applies, especially where, as in this case, the testimony of the District’s own representatives, described infra at pp. 458-459, demonstrates that the failure to weigh such factors was the norm rather than the exception.
There is no dispositive District of Columbia precedent on the precise question presented to us, but I believe that the approach of the courts in this jurisdiction has been generally consistent with the authorities which I have cited. It has long been established that the District has a ministerial duty to maintain the streets in a reasonably safe condition for travel, and that this is an exception to the doctrine of sovereign immunity. Wagshal v. District of Columbia, 216 A.2d 172, 173 (D.C.1966); see also District of Columbia v. Pace, 498 A.2d 226, 230 (D.C.1985). In Wagshal, this court, citing Urow v. District of Columbia, 114 U.S.App.D.C. 350, 316 F.2d 351 (1962) (per curiam), cert. denied, 375 U.S. 826, 84 S.Ct. 69, 11 L.Ed.2d 59 (1963), recognized that the District’s determination whether or not to place a traffic light at a particular intersection was discretionary and protected by sovereign immunity. Wagshal, 216 A.2d at 173. We held, however, that there can be no immunity “if the plan that is adopted creates a hazard on the road, either because of its inherent unreasonableness or because of negligence in its administration.” Id. at 173-74 (emphasis added). More recently, we observed that “[t]he fact that in a particular case an [employee] might have alternative courses of action from which to choose, and this choice might involve a certain degree of judgment, does not elevate the [employee’s] decision to the level of ‘basic policy.’ ” WMATA v. O’Neill, 633 A.2d 834, 839 (D.C.1993) (quoting Lopez v. Southern California Rapid Transit, 40 Cal.3d 780, 221 Cal.Rptr. 840, 849, 710 P.2d 907, 916 (1985)).
Only a few months ago, in Cope v. Scott, 310 U.S.App.D.C. 144, 45 F.3d 445 (1995), the United States Court of Appeals unanimously held that the decision where to place “slippery when wet” signs in Rock Creek Park was not “grounded in the policy of the regulatory regime” and did not implicate “political, social, or economic policy choices of the *457sort that Congress intended to protect [by granting immunity.]” Id. at-, 45 F.3d at 452. The court rejected the government’s contention that the balancing of safety considerations against the aesthetic appearance of Beach Drive represented a protected discretionary function. My colleagues contend that Cope was erroneously decided,5 and that Urow, not Cope, controls. The 2:1 decision in Urow is, however, distinguishable in a decisive respect. See note 12, infra, and preceding text. Moreover, although Cope is not binding on us, we ought at least to pause before we reject a very recent decision of our federal appellate colleagues across the street and place the contemporary law of sovereign immunity in the District of Columbia in unnecessary disarray. See Hornstein v. Barry, 560 A.2d 530, 536-37 n. 15 (D.C.1989) (en banc).6
II. SETTING THE CLEARANCE INTERVAL — A DISCRETIONARY FUNCTION?
The trial judge found it to be “clear” that setting traffic signal intervals “is the type of function that the discretionary function rale is designed to shield.” This is so, according to the judge, because the timing of the light
involves considerations of safety not only for pedestrians but for travellers, and it involves a balancing of safety needs against the need to assure adequate traffic flow, which itself involves considerations of safety as well as commerce and convenience. Balancing these facts also requires the ascertainment of facts, such as numbers of vehicles and pedestrians, and ways in which drivers and pedestrians behave in the aggregate, which are peculiarly subject to study and expertise. Subjecting the decisions of traffic engineers to litigation and to seeond-giiessing by jurors would deter effective government. For these and other reasons, our courts have held that with respect to functions like the one here, the function is discretionary.[7]
My colleagues “fully agree” with this analysis, but I do not believe that the record bears it out.
A. The State of the Record.
I think it important to note at the outset that we are not dealing here with the question whether to install a traffic signal at the intersection of Wisconsin Avenue and Fes-senden Street. Cf. Urow, supra. The decision to place a signal at that location had been made many years before. The determination by the District’s engineers which is challenged on this appeal concerns solely the length in seconds of the “clearance interval;” i.e., how much time vehicular traffic should be given to clear the intersection before pedestrians and drivers are given a green light to cross the street.
According to Aguehounde, there is no support in the record for the trial judge’s assertion that this narrow question required the District’s engineers to balance pedestrian safety against traffic flow, commerce, or convenience, or to include in their calculus the various other considerations enumerated by the trial judge in the passage from his opinion which I have quoted. The District does not really challenge Aguehounde’s reading of *458the record,8 and my own search has revealed nothing to suggest that Aguehounde is mistaken. Indeed, the testimony of Ague-hounde’s expert witness reveals, and the District’s “Required Yellow Interval in Seconds” chart9 confirms, that the length of the clearance interval is set — and is supposed to be set — on the basis of a three-part formula which does not include the factors to which the judge referred.
Moreover, the testimony of the District’s own witnesses reflects that social, economic and political factors were probably not considered at all when the decision in question was made. In most instances, according to the District’s expert, a four-second interval is selected almost automatically as suitable for any intersection. Finally, a traffic engineer who was the principal witness for the District effectively admitted in testimony which I quote in note 10, infra, that on this particular occasion, it was a faulty measurement of the intersection, rather than a high-level choice between policies, that led to what proved to be an erroneous decision as to how long the clearance interval should be.
Aguehounde called two expert witnesses, Richard Pivnik, a traffic engineer, and Clyde Richard, an expert on accident reconstruction. Pivnik testified that the correct clearance interval is determined by a formula based on the “perception reaction time,” the “accepted deceleration rate for a controlled comfortable stop,” and the “time it would take a car to clear the intersection.” The third of these factors obviously depends on the width of the intersection. Pivnik believed that this formula — which the District applied in its “Required Yellow Interval in Seconds” chart — was in use in “every jurisdiction ... that I know of, that I’m familiar with,” and that it had been an accepted formula for at least thirty years. Pivnik did not include in the calculus the other eonsider-ations enumerated by the trial judge, such as “balancing of safety needs against the need to assure adequate traffic flow,” and it is readily apparent from his testimony that these factors are not a part of the formula, and that they therefore play no role in this particular decision. The testimony of Clyde Richard, an accident reconstruction expert, was essentially identical to that of Mr. Pivnik with respect to the manner by which the formula is calculated.
Raj Ghaman, the District’s traffic engineering expert, testified that “there are various tests that suggest and are commanded that clearances — signal clearances — be between three and six seconds.” He stated that “[aljmost everybody, and I included, have used about four seconds as a magic number that works at just about every intersection.” Ghaman did not contradict the testimony of Pivnik and Richard as to the factors which go into the “clearance interval” calculus. Moreover, his testimony indicates that, far from weighing competing policy considerations applicable at the particular intersection, District of Columbia engineers routinely set the interval at four seconds at virtually all intersections.
Saraj Gyani, a traffic engineer who was the District’s principal witness, testified primarily that he was unaware of any statute, regulation or policy that required a six-second interval at the Fessenden traffic light. He also expressed the opinion that four seconds was a safe interval. Gyani did not, however, contradict the evidence presented by Ague-hounde as to the factors which engineers consider in setting the clearance interval. Indeed, his testimony does not suggest that the clearance interval is set at any given intersection, or was set at four seconds in this case, on the basis of a choice between policy alternatives.10
*459Like most judges, I am no expert on traffic engineering. There may perhaps appear to be some intuitive merit in the trial judge’s assessment of what factors engineers do or should consider in determining the duration of the clearance interval. But if the factors enumerated by the judge, or any of them, are a part of the prescribed or actual calculus, the District had ample opportunity to adduce evidence proving this to be the case. The District failed altogether to do so, however, and instead presented testimony that the interval was either arbitrarily fixed at four seconds as a matter of routine (according to Ghaman) or selected, in this case, as a result of a measuring error (according to Gyani).11
I do not think the trial court or this court can properly come to the District’s rescue in light of this failure of proof, especially given what I view as the effective destruction of the District’s “choice between policy alternatives” theory by its own witnesses. We have recognized that “where a fact is well known by all reasonably intelligent people in the community,” courts may judicially notice that fact without requiring formal proof. Poulnot v. District of Columbia, 608 A.2d 134, 141 (D.C.1992) (citation omitted). Any connection, however, between the duration of the clearance interval at Wisconsin and Fessen-den and the flow of traffic, both there and at other intersections, is not sufficiently well-known or obvious to permit a court to take judicial notice of it.
B. The “Clearance Interval” Case Law.
Given the state of the record, the cases in this jurisdiction on which the judge relied are readily distinguishable. In each of them, the governmental decision in question was different from, and more traditionally discretionary than, the setting of the “clearance interval.” Moreover, as I read these cases, none of them involved a situation in which the evidence showed that no “policy” discretion was exercised in fact.12
*460As my colleagues have pointed out, “jurisdictions which have directly addressed the issue ... whether the setting of a traffic light clearance interval is a discretionary function have gone both ways.” Op. at 448, n. 9. In my judgment, however, the decisions which are said to support the District’s position are readily distinguishable, or unpersuasive when compared to the dissents, or both.13 The eases rejecting the “discretionary function” defense, on the other hand, appear to me to constitute well-considered, albeit non-binding, precedents.14
*461 C. Other Considerations.
It is also worth noting that two Superior Court judges ruled, at earlier stages of this litigation, that the timing of the clearance interval was ministerial rather than discretionary in nature. In denying the District’s motion for summary judgment, Judge Colleen Kollar-Kotelly held:
[T]he plaintiff claims that the District failed to “execute” [its] own policy of using a set formula by improperly measuring the width of the road. Since the District does not have immunity for ministerial acts such as measuring distances and calculations after a policy is already determined, the District cannot claim immunity [as] to this basis for negligence.
Subsequently, at trial, Judge Burgess denied the District’s motion for a directed verdict at the conclusion of Aguehounde’s case because “the figure was too small” for the width of the intersection, resulting in a clearance time that was too short. Judge Burgess later reversed himself, however, and granted the District’s motion for judgment n.o.v.
III. CONCLUSION
Richard Pivnik testified on behalf of Ague-hounde that by incorrectly measuring the intersection, and by setting the clearance internal on the basis of the erroneous measurement, the District’s engineers made the light turn green before the cars had cleared the intersection. He stated that this premature authorization to walk across the street created a danger which would not be readily apparent to a pedestrian.15 If this testimony is credited, then employees of the District effectively (though inadvertently) lured Aguehounde into the accident by measuring the wrong distance.
I do not believe that such a measuring error is the kind of governmental action to which principles of sovereign immunity can fairly be held to apply. This is not a situation in which the jury is being asked to second-guess an authentic discretionary choice between policy alternatives. If the District had presented the testimony of the decision-maker to the effect that he weighed political, economic, or social considerations, or if it had adduced expert testimony showing that the setting of a clearance interval affects traffic at other intersections, that pedestrian safety was or should have been weighed against the movement of traffic, or that other recognized discretionary factors were or should have been considered, we would have a different ease. In the case which we do have, however, I vote to reject the “discretionary function” defense.16
. There are obvious limits to the applicability of this rule, especially in light of the availability of modern pretrial discovery. See Edward W. Cleary, McCormick on Evidence, § 337, at 950 & n. II (1984). Nevertheless, it appears to me unreasonable in the present context to require Ague-hounde to prove the circumstances under which a decision was made within the District government, or the basis on which District employees made such a decision. Cf. Stewart v. United States, 199 F.2d 517, 520 (7th Cir.1952), in which the court concluded, using some strong language, that it would be altogether inappropriate to place such a burden on the plaintiff.
. In District of Columbia v. Banks, 646 A.2d 972 (D.C.1994), this court held that "governmental immunity must be affirmatively asserted in the defendant's pleadings.” Id. at 979 n. 9. Although the question of the burden of proof as to sovereign immunity was not presented in Banks, our treatment of the District's claim of immunity as analogous to an affirmative defense leads logically to the conclusion that the District must carry that burden.
. The decision of the Supreme Court of California in Johnson and the views of the leading commentator on the law of municipal corporations cannot accurately be characterized as "no support.” Cf. Op. at 449. No reasonable person can quarrel with the truism, relied on by the majority, that "where there is room for policy judgment and decision, there is discretion.” Id. at 449 (quoting Dalehite v. United States, 346 U.S. 15, 36, 73 S.Ct. 956, 968, 97 L.Ed. 1427 (1953)). This proposition, however, does not negate the Supreme Court’s explanation that "discretionary function” exception was designed to insulate from judicial second-guessing "decisions grounded in social and economic policy." Varig Airlines, supra, 467 U.S. at 814, 104 S.Ct. at 2764 (emphasis added).
. See, e.g., Crumpton v. Stone, 313 U.S.App.D.C. 412, --, 59 F.3d 1400, 1403 (1995) ("we look not to what the decision-maker in a particular case was thinking, but to whether the type of decision being challenged implicates policy judgments”) (citation and internal quotation marks omitted).
. The majority also points out that Cope was decided under the Federal Tort Claims Act. This court's decision to adopt "the distinction between discretionary and ministerial decisions was [, however,] influenced by similar distinctions drawn under the [FTCA].” District of Columbia v. North Wash. Neighbors, Inc., 367 A.2d 143, 148 n. 7 (D.C.1976), cert. denied, 434 U.S. 823, 98 S.Ct. 68, 54 L.Ed.2d 80 (1977); see also Spencer v. General Hospital of the District of Columbia, 138 U.S.App.D.C. 48, 53, 425 F.2d 479, 484 (1969) (en banc). Accordingly, I do not think that Cope can be persuasively distinguished on that ground.
. Oddly, the court in Cope did not mention Urow. Cf. United States v. Doe, 235 U.S.App.D.C. 99, 101 n. 2, 730 F.2d 1529, 1531 n. 2 (1984) (one panel of United States Court of Appeals lacks the authority to overrule decision of a previous panel; only the full court, sitting en banc, may do so).
7.The judge cited Urow, supra, 114 U.S.App.D.C. at 352, 316 F.2d at 353; North Wash. Neighbors, Inc., supra, 367 A.2d at 148 n. 7; District of Columbia v. Pace, supra, 498 A.2d at 229; and McKethean v. WMATA, 588 A.2d 708, 715 (D.C. 1991) as involving "functions like the one here.” I address these decisions in note 12, infra. The judge also relied on three decisions from other jurisdictions specifically involving traffic signal intervals. I discuss these cases in note 13, infra.
.The District claims in its brief that a decision with respect to the clearance interval at a particular intersection
does not simply affect that intersection. It affects the flow of traffic on the entire street and hence, affects the larger system-wide traffic control structure, which is clearly entitled to discretionary immunity.
It is revealing that the District has no citation to the record following this assertion.
. See Op. at 446.
. Specifically, as I read the record, Gyani acknowledged that in calculating the width of the intersection for the purpose of identifying "the time it would take a car to clear the intersection,” the District measured the wrong distance. In its answer to an interrogatory, the District had represented that the width of the intersection *459was taken curb to curb, a distance of 65 feet. Gyani, however, testified as follows:
Q. Backing up one second in terms of measuring the intersection, when we know the speed is 25 miles an hour, the way you were supposed to measure the width of the intersection to set the clearance interval as of April 1990 — correct me if I’m wrong — was from the stop bar one side, across the first crosswalk, across the middle of the intersection, and then across the crosswalk on the other side; am I right?
A. You are right.
Q. ... And, the reason for that is that you need to measure the width of the intersection in order to have enough time for the cars, not just to clear the box in the middle, but also to clear the crosswalks where the pedestrians might be; correct?
A. Correct.
Q. ... And, this measurement here, 105 feet 9 inches, is an accurate measurement of stop bar to the far side of the far crosswalk on Wisconsin and Fessenden; correct?
A. I haven't measured it, but, yes, it sounds reasonable.
Q. ... So, what you should have had on this intersection as of April 23, 1990, is a clearance interval of at least 5.7, or rounded off, six seconds; correct?
A. Should have.
The District contends that this testimony was based on the unverified assumption that the formula in the District’s chart was supposed to be applied. My colleagues in the majority apparently agree. See Op. at 453. The transcript reveals, however, that Gyani made his concession without any qualification. No qualification was necessary, for Gyani testified that although the formula was not mandated, this chart was in use both at the time of his testimony and at the time the decision as to the clearance interval was made.
The reader should note that the indented lines on pp. 22 of the majority opinion, beginning with the words "If that formula is followed,” are not a quotation from Gyani's testimony. Rather, they reflect the majority's conception of what Gyani meant. The actual testimony which Gyani gave is quoted in this footnote.
. If the Districts engineers had been making a "policy” decision as between pedestrian safety and maintaining the flow of traffic, as the District suggests, then the clearance interval could have been varied according to traffic volume at different times of day in order to improve traffic flow. This was not done, however. In fact, Aguehounde's evidence showed that the proportion of red and green time for Fessenden Street and Wisconsin Avenue was set differently for “Off Peak,” "A.M. Peak," and "P.M. Peak” times, but that the Fessenden Street yellow light remained constant at four seconds throughout the day.
. In Urow, the court, with one judge dissenting, held that the decision whether to install a traffic light was discretionary where the applicable statute authorized that the commissioners to provide such traffic control devices "as are deemed advisable.” 114 U.S.App.D.C. at 351-52, 316 F.2d at 352-53. In North Wash. Neighbors, Inc., a *460case involving the question whether homeowners in the District must bear the cost of repairing certain pipes, the question of sovereign immunity was not raised by the parties, and the court simply commented that “the designing of streets and the control of the flow of traffic over them [are] discretionary in nature.” 367 A.2d at 148 n. 7. In Pace, supra, we held that the planning of highways and the setting of priorities for their improvement were "policy determinations” which “callfed] for a delicate balancing of competing considerations,” and were thus discretionary. 498 A.2d at 228-29. In McKethean v. WMATA, 588 A.2d 708 (D.C.1991), we concluded that the failure to relocate a bus stop, like other aspects of the "design and planning of a transportation system,” id. at 714, was a "quasi-legislative policy decision which [is] discretionary in nature and should not be second-guessed by a jury,” Id. The court recognized in McKethean, however, that only a "policy decision” is discretionary, and that a decision implementing a policy is ministerial. Id. at 713.
Once a decision to place a traffic signal at an intersection has been made, the setting of a clearance interval could reasonably be viewed as ministerial implementation of that decision. In my view, the District failed altogether to prove that the clearance decision was not ministerial. In any event, none of the cases on which the trial judge relied decided that question.
. The leading such case is Weiss v. Fote, 7 N.Y.2d 579, 200 N.Y.S.2d 409, 167 N.E.2d 63 (1960). In Weiss, the New York Court of Appeals, by a vote of 4:3, held that the setting of a four-second interval at a particular intersection by the City of Buffalo’s Board of Safety was discretionary because the Board relied on extensive studies and because "there is nothing to suggest that its decision was either arbitrary or unreasonable.” Id. 200 N.Y.S.2d at 413, 167 N.E.2d at 66. In the present case, on the other hand, Richard Pivnik testified that the setting of the light at four seconds was unreasonable because it invited pedestrians like Aguehounde to walk into the intersection, with the light in his favor, before the vehicular traffic had cleared. Moreover, Chief Judge Desmond wrote a compelling dissent in Weiss. He rejected the notion that "the timing of a single traffic light could be considered the kind of high-level policy decision not reviewable by the courts,” and he complained that "[b]y this decision we are taking a long and surprising step backward into the old, abandoned area of governmental immunity.” Id. 200 N.Y.S.2d at 416, 167 N.E.2d at 68.
In Bjorkquist v. City of Robbinsdale, 352 N.W.2d 817 (Minn.App.1984), the court held, relying on Weiss, supra, that the timing of a traffic signal is a discretionary duty because "[tjhere is no obligation to time the lights a particular way,” and because "that decision is arrived at after weighing competing interests.” Id. at 819. In the present case, on the other hand, the testimony of Aguehounde's expert witnesses, which was bolstered by the District’s witnesses, indicated that there was no such weighing here.
In Davis v. City of Cleveland, 709 S.W.2d 613 (Tenn.App.1986), a divided court held, by 2:1, that "the original setting of the yellow caution interval and any failure to reset such timing sequences in this instance represented a [discretionary] judgment call.” Id. at 615. Judge Franks wrote a scholarly dissent which, in my view, is more persuasive than the majority opinion. After noting that “a wide array of 'operational' acts or decisions of public employees have been held 'non-discretionaiy’ and therefore not exempt from tort liability,” id. at 616, (citations omitted), Judge Franks wrote as follows:
The affidavit of appellant’s expert witness advances the opinion that the 3.6 second interval was deficient under current standards, particularly in light of the traffic volume and approach speed at the intersection. The witness concluded "[t]he proper yellow time ... would have allowed the Jackson truck to clear the intersection before Mr. Davis was shown a green signal.” Thus, a disputed issue of material fact as to whether the city or county, through its employees, acted negligently in setting the caution light interval under these conditions was established.
Id. at 617.
. In Delosovic v. City of New York, 143 Misc.2d 801, 541 N.Y.S.2d 685 (Supreme Ct.N.Y. County 1989), a trial court decision, the judge addressed the common sense of a situation which was quite similar in principle to the present one:
Notwithstanding the fact that no case has been found discussing governmental liability relating to "Walk” — "Don’t Walk” signals, and although such signals do not exist at all intersections where there are traffic lights, I find that once these signals are installed there is a duty to pedestrians to see that they perform in a proper manner to inform pedestrians when it is appropriate to proceed, and there may be governmental liability to a person injured by a vehicle if the timing is such that there is an inadequate period provided at any time the "Walk” signal appears for a person, proceeding at an average *461rate of speed, to complete the crossing before vehicular traffic is authorized to proceed.
Id. 541 N.Y.S.2d at 688. (Emphasis added). I do find it odd, however, that the judge did not cite or consider the decision of his state's highest court in Weiss v. Fote, supra.
In Fraley v. City of Flint, 54 Mich.App. 570, 221 N.W.2d 394 (1974), the court squarely rejected the city's contention that "a municipality cannot, as a matter of law, be held liable in tort for the alleged improper timing of a traffic control signal.” Id. 221 N.W.2d at 397. The court noted that Michigan law (like the District's) imposes a duty upon the government agency to maintain and design highways with reasonable care, and that the negligent setting of the interval, even within the range recommended by the Michigan Manual of Uniform Traffic Control Devices, was not shielded from liability. Id. 221 N.W.2d at 397.
. Because my colleagues dispose of this case on the grounds of sovereign immunity, I do not reach the question whether Aguehounde was contributorily negligent as a matter of law. I wonder, however, whether judgment n.o.v. is appropriate when, if the evidence is viewed in the light most favorable to Aguehounde, the alleged contributory negligence was itself induced by the defendant.
. Because I am of the opinion that, on this record, the setting of the clearance interval was a ministerial act rather than a discretionary one, I would not reach the question whether, assuming that such a decision is a discretionary one, it was rendered ministerial by the existence of a specific directive which left no room for an official to exercise a policy judgment. See Berkovitz, supra, 486 U.S. at 547, 108 S.Ct. at 1964. My colleagues do reach that question, however, and I feel constrained to express some reservations regarding their disposition of it.
Although there was plainly support in the record for the judge's finding that the formula in the *462“Required Yellow Interval in Seconds” chart was not mandated, it is not at all clear to me that the same result would properly be reached if the District were assigned the burden of proof. Gya-ni testified that, although there was no written policy requiring the use of the chart, the person who sets the clearance interval uses the chart "because he feels that is the chart he should use.” The word “required" speaks for itself.
Moreover, Mr. Moore, the engineer who set the clearance interval (and who is not otherwise identified in the record) was not called by the District to testify. “The production of weak evidence when strong is available can lead only to the conclusion that the strong would have been diverse.” Interstate Circuit, Inc. v. United States, 306 U.S. 208, 226, 59 S.Ct. 467, 474, 83 L.Ed. 610 (1939); accord, Murphy v. McCloud, 650 A.2d 202, 215 (D.C.1994). Although we do not know why Moore did not testify, the weakness of the evidence which the District did present leads me to question whether it can fairly be viewed as having met its burden of proof on the "specific directive" issue. I would be inclined to remand on that question with directions to the trial court to apply the correct burden of proof.