(dissenting). After reading the majority’s opinion, one is left with the impression that rules of law are merely matters of policy preferences to be invoked, modified or simply ignored when their consequence are, in the eyes of four members of this Court, inconvenient or undesirable. Indeed, the word "policy” appears no less than 11 times in the majority’s opinion and at least seven references are made to balancing society’s and defendant’s interests, as though the mere repetition of those concepts alone has the power to persuade. What constitutes good policy, however, is within the eyes of the beholder. Since the practical effect of the Court’s *651decision here will be to make relief for Rosario violations virtually unavailable in postconviction proceedings, the result can hardly be characterized as "equitable” or good policy from the viewpoint of an aggrieved defendant unfortunate enough to be relegated to such proceedings (see, majority opn, at 649). Further, despite the majority’s efforts to place the burden of responsibility for its policy choice on the language of the applicable statute, there is nothing in the language of CPL 440.10 (1) (f) that requires such a dramatic departure from the clear rule that has governed the disposition of Rosario claims in direct appeals for the past 15 years (see, People v Jones, 70 NY2d 547, 553; People v Novoa, 70 NY2d 490, 499; People v Ranghelle, 69 NY2d 56, 63; People v Perez, 65 NY2d 154, 160; People v Consolazio, 40 NY2d 446, 454, cert denied 433 US 914). To the contrary, the logic and values that inform that rule compel its application in CPL 440.10 proceedings. Accordingly, I dissent.
I.
In People v Rosario (9 NY2d 286, 289), this Court held that "a right sense of justice” requires that before cross-examinatian the People must turn over to the defense the pretrial statements of prosecution witnesses pertaining to the subject matter of their testimony. Fifteen years later, in People v Consolazio (40 NY2d, at 454, supra), we made clear that a failure on the part of the People to disclose such material could not "be excused on the ground that [it] would have been of limited or of no use to the defense.”
To read the majority’s opinion, one would think that this rule was nothing more than a bit of judicial legislation grounded in policy choice rather than logic and reason (see, majority opn, at 641 ["(i)n originally adopting the per se error rule, we balanced the rights of the defendant against the rights of society and arrived at a policy designed to foster” the Rosario goals]; id., at 644-645 ["(a)pplying a per se error rule * * * was * * * a policy decision”]; see also, id., at 643 ["our decision 'turn(ed) largely on policy considerations’ ”]). However, nothing could be further from the truth.1 As the Chief Judge *652himself explained in People v Perez (65 NY2d, at 160, supra), the "essence of the Rosario requirement * * * is that a judge’s impartial determination as to what portions [of a witness’s pretrial statements] may be useful to the defense, is no substitute for the single-minded devotion of counsel for the accused.” For that reason, the Perez Court concluded, a harmless error analysis based on the relative probative worth of the withheld evidence is inappropriate (accord, People v Ranghelle, 69 NY2d 56, supra).
The subject was once again explored in People Jones (70 NY2d 547, supra), in which this Court squarely and firmly rejected an argument by the People that the Consolazio-Perez-Ranghelle line of cases should be overruled. Echoing and expanding upon the analysis of the Perez Court, we stated that "[t]he nature and purpose of Rosario” make it "impossible” to apply traditional harmless error analysis based on the quantum and nature of the People’s proof and the causal effect of the error on the outcome (70 NY2d, at 551). We explained: "When, as a result of the prosecutor’s violation of the Rosario rule, defense counsel has been deprived of material of which he or she is unaware or cannot otherwise obtain, there is no way, short of speculation, of determining how it might have been used or how its denial to counsel might have damaged defendant’s case” (id., at 552). Since the Jones Court believed that such speculation would "undermine the very basis of our prior decisions,” it declined to apply a harmless error analysis based on the evidentiary value of the undisclosed material (id., at 552; see also, People v Novoa, 70 NY2d, at 499, supra).
A bare four-Judge majority has chosen a different conclusion in this case, not because the procedural posture of CPL 440.10 proceedings makes the harmless error inquiry any less "impossible,” but rather because, in its view, a showing of particularized prejudice is specifically required by the language of the applicable provision, CPL 440.10 (1) (f). In this regard, however, the present majority’s rationale is nothing more than an exercise in result-directed statutory construction that simply does not withstand scrutiny.
*653II.
The term "prejudice],” as used in CPL 440.10 (1) (f) and elsewhere in the Criminal Procedure Law, is fraught with ambiguity. It is a word that has no single meaning; rather, it takes on whatever significance logic and reason would give it in the context in which it is being used. Thus, in cases involving errors in the admission or exclusion of evidence, the term "prejudice” may fairly be used to refer to the probable effect of the error on the outcome, and the degree of "prejudice” may be assessed by considering "the nature and the inherent probative worth of the evidence” (People v Crimmins, 36 NY2d 230, 241). In cases involving other classes of errors, however, the nature and quantum of proof is irrelevant in determining what constitutes "prejudice” (see, People v Felder, 47 NY2d 287, 296), and the inquiry requires a more subtle analysis.
For example, errors in the trial court’s closing instructions that deprive the defendant of his right to jury consideration of the crime elements are deemed prejudicial " '[n]o matter how conclusive the evidence’ ” (People v Lewis, 64 NY2d 1031, 1032, quoting People v Walker, 198 NY 329, 334). Similarly, errors affecting the jury’s deliberative process have been held prejudicial to the extent they tend to "compromise” that process, without regard to the strength of the People’s case (see, e.g., People v Brown, 48 NY2d 388, 395). Cases involving deprivations of the basic right to representation by counsel are also generally not subjected to traditional harmless error analysis (see, e.g., People v Hilliard, 73 NY2d 584; People v Felder, supra).
People v O’Rama (78 NY2d 270, 279-280) is particularly pertinent in this context. In that case, the trial court had violated its statutory duty to notify defense counsel of the precise contents of a juror inquiry (see, CPL 310.30). We concluded that reversal was required notwithstanding the defendant’s inability to show "specific prejudice” because the court’s error had deprived defendant of "the opportunity to have input, through counsel * * * into the court’s response to an important, substantive juror inquiry” and that, as a consequence, the error was "inherently prejudicial” (78 NY2d, at 279-280). Significantly, the opinion in O’Rama drew an analogy to the recent Rosario cases in which reversal was held required without regard to a showing of specific prejudice (78 *654NY2d, at 279-280, citing People v Jones, supra; People v Perez, supra).
The teaching of O’Rama and the other above-cited cases is that certain kinds of errors occurring during trial are intrinsically prejudicial because they either detract from the process or impair the defendant’s ability to present a defense. In such instances, a so-called "per se” rule of reversal is applied. The use. of the term "per se” in this context does not denote a complete absence of prejudice; rather, it represents a shorthand way of saying that errors within that class are prejudicial by their very nature and that, accordingly, nothing further need be shown to compel reversal.
That is precisely what this Court had in mind when it applied a "per se” rule of reversal in Jones, Perez, Consolazio, Novoa and Ranghelle. When a prosecutor fails to disclose Rosario material during trial, the defense is deprived of the opportunity to make the necessary strategic judgments and do "the careful preparation required for planning and executing an effective cross-examination” (People v Jones, 70 NY2d, at 552, supra). Further, the defense has been prevented from exploring potential areas of weakness in the prosecution’s evidence. This deprivation goes to the heart of counsel’s ability to provide a meaningful defense and, consequently, is "prejudicial” in the same sense that the error in People v O’Rama (supra, 279-280) was prejudicial to the defense in that case.
Thus, contrary to the majority’s central argument, CPL 440.10 (1) (f)’s requirement of "prejudice” does not provide a justification for abandoning in this context the Rosario per se rule that has been applied for the past 15 years in direct appeals. Indeed, if "prejudice” had not been present in Jones, Ranghelle, Perez and Consolazio, we would have been guilty of exceeding our powers as an appellate court, since CPL 470.05 (1), which applies to direct appeals, dictates that we determine appeals "without regard to technical errors or defects which do not affect the substantial rights of the parties” (emphasis supplied). Surely, the majority cannot seriously suggest that the term "prejudice]” in CPL 440.10 (1) (f) means something different from the impairment of the "substantial rights” of the accused that CPL 470.05 (1) requires.
In the final analysis, the majority’s choice to réach a different result under CPL 440.10 (1) (f) than under CPL 470.05 (1) cannot be justified by statutory language, precedent or a priori *655logic. Thus, in reality, it must be premised solely on what the majority terms a choice of "policy” — the majority’s current code word for its four members’ unwillingness to apply a well-settled rule of law in such a way as to produce vacaturs of previously settled criminal convictions. In my view, this choice is unsound and, indeed, indefensible.
III.
The majority’s policy choice is objectionable not only because it ignores recent and settled precedent, but also because it is so one-sided. The particularized prejudice standard that the majority has embraced requires a "reasonable possibility that the failure to disclose the Rosario material contributed to the verdict” (majority opn, at 649). The majority gives no particular reason for choosing this standard, other than its use in the original Rosario decision (9 NY2d 286, cert denied 368 US 866, supra) and its congruence with the rule now applied in New York in Brady violation cases (Brady v Maryland, 373 US 83; see, People v Vilardi, 76 NY2d 67). In view of the substantial differences between the goals and purposes of the Rosario and Brady rules, however, this congruence alone can hardly provide a sound basis for such a profound change in the law. And, the use of the standard in the original Rosario decision is certainly not a persuasive point, since that standard has long been superseded by the per se rule of reversal developed in Consolazio and its progeny.
Most significantly, the use of this standard, which was tailored for Brady cases, will, as a practical matter, eliminate the possibility of a remedy for most defendants who, through no fault of their own, do not discover Rosario violations until after sentence is imposed. The majority’s rule mandates an assessment of the factual significance of the undisclosed material in light of the admitted trial evidence. However, as stated in People v Jones (70 NY2d 547, 551-553, supra), it is simply impossible for a reviewing court to determine whether the outcome of a particular case may have been affected by the prosecution’s failure to turn over a pretrial statement of one of its witnesses (accord, People v Perez, 65 NY2d, at 160, supra). Indeed, the evidentiary value of Rosario material is often not even discernible to defense counsel until after he or she has had an opportunity to use it as a basis for cross-examination, since even the most seemingly innocuous material may open new avenues of impeachment that could not *656have been foreseen before the witness’s responses were heard. Thus, as the majority is well aware, the cases in which the requisite showing can be made will be few and far between. In fact, since the majority opinion does not even attempt to explain why the inquiry it mandates will be any more fair or any less "impossible” in postconviction proceedings than it would have been in the appeals in Perez (supra), Consolazio (supra), Ranghelle (supra) or Jones (supra), the conclusion is inescapable that the Court’s previously expressed concern for fairness in the adjudicative process has, at least in this context, been supplanted with a concern to protect existing convictions, particularly where the passage of time "affects the People’s ability to retry [the] defendant” (majority opn, at 647; see, id., at 646).2
IV.
That the majority’s central concern lies with the age, rather than the underlying fairness, of the convictions challenged in this and similar cases is evident not only from its many explicit references to that consideration (see, majority opn, at 640, 645, 646, 647, 650), but also from the peculiar exception it has created for cases where the Rosario claim "is the subject of both a direct appeal and a postconviction motion brought under CPL 440.10” (majority opn, at 649). Presumably, this exception does not run afoul of the majority’s "policy” concerns because it implicates only the most recent convictions in the system, convictions which are vulnerable in any case as long as a direct appeal is pending.
The ostensible justification for recognizing what might otherwise appear to be an arbitrary "exception” is the authority of People v Novoa (70 NY2d 490, supra), in which, according to *657the majority, the Rosario claim "was the subject of both direct appeal and a CPL 440.10 motion” and "was treated as a direct appeal for procedural purposes” (majority opn, at 641; see also, id., at 649). This characterization of Novoa, however, reflects a fundamental misunderstanding of appellate procedure.
In fact, the material that formed the basis of the Rosario claim in Novoa was not discovered until more than a year after judgment was entered (see, 70 NY2d, at 494, supra). Accordingly, it was not part of the record on the direct appeal and could not have been considered on that appeal, absent a discretionary order by the Appellate Division to "expand” the record (cf., People v Perez, supra). Further, since the existence of the material was not known during the trial, there was neither a "protest” nor a judicial "ruling” on the record underlying the judgment and, thus, the Rosario claim presented no "question of law” for this Court to review on the direct appeal (see, CPL 470.05 [2]).
In truth, the only ruling on which this Court could have reversed in Novoa was the trial court’s order denying the convicted defendant’s postjudgment motion for relief following a full hearing (see, 70 NY2d, at 495-496, supra) — an order which was not cognizable on the direct appeal. Indeed, as its primary reason for reversing, the Novoa Court specifically cited its disagreement with the trial court’s analysis on the postjudgment motion. Accordingly, unless the Novoa Court was seriously mistaken in its appreciation of the procedural posture of the case before it, it simply cannot be said that the Rosario claim was either "the subject of [the] direct appeal” or "treated as a direct appeal for procedural purposes” (see, majority opn, at 641; see also, id., at 649).3
When stripped of its supposed precedential support in Novoa, the exception that the majority has created for cases where the Rosario claim "is the subject of both a direct appeal and a postconviction motion” is exposed as an arbitrary and unrealistic attempt at expediency. It is arbitrary because it makes the availability of the per se reversal rule turn on the fortuity of whether the undisclosed Rosario material is discovered before or after the intermediate appellate court has *658heard and determined the direct appeal.4 It is unrealistic for the simple reason that, as discussed above, the procedural principles governing appeals make the existence of the conditions for the exception, at best, unlikely.
V.
Finally, if policy considerations are to be the touchstone, the Court should also consider the effect that its decision will have, on both the administration of the Rosario rule and on the behavior of prosecutors. Viewed against both of these yardsticks, the majority’s holding falls short.
First, by establishing a separate standard for reversal in CPL 440.10 cases, the majority has left little room for the Consolazio-Perez-Ranghelle-Jones per se rule of reversal to operate. In cases where the undisclosed Rosario material comes to light before the close of evidence, the error will likely be treated as a “mere delay” and appellate reversal will depend upon “whether the defense was substantially prejudiced” (People v Ranghelle, 69 NY2d, at 63, supra). In cases where the failure to disclose becomes apparent only after entry of judgment, the majority’s new rule will apply and reversal will also be contingent on a particularized showing of actual prejudice.5 Significantly, in such cases, the Rosario doctrine itself will have little utility, since a defendant who can meet the majority’s new test can, by definition, also obtain vacatur under the Brady-Vilardi standard, without regard to Rosario. What remains of the Consolazio-Perez-Ranghelle-Jones per se reversal rule is, then, its availability in that narrow category of cases in which the defense learns of the *659prosecution’s failure to fulfill its Rosario obligations sometime after the close of evidence and before the entry of judgment (see, People v Ranghelle, supra, at 62). Such a narrowing of the rule’s reach is hardly consistent with the majority’s expressed continuing "commitment” to the Rosario doctrine.
Second, the rule the majority adopts will lead to an unacceptable disparity in the administration of the Rosario rule. On the one hand, defendants who are able to demonstrate on direct appeal the existence of a complete failure to disclose Rosario material will be entitled to a new trial without further inquiry. In contrast, defendants who have equally meritorious Rosario claims but are relegated to CPL 440.10 challenges will be required to make a particularized showing of actual prejudice and, for the reasons discussed above, are unlikely to succeed. This disparity is especially troubling because it makes the outcome depend largely upon the timing of disclosure, a matter that is solely within the People’s control.
Third, because the timing of disclosure will now be all büt dispositive, the rule the majority adopts creates a strong incentive for prosecutors who belatedly discover potential Rosario material to postpone disclosure until after sentencing, when the record available for direct appeal is closed and the only remedy the defendant has is to seek CPL 440.10 postjudgment relief. Such a result is objectionable because it rewards prosecutorial gamesmanship and runs directly counter to the underlying Rosario policy of encouraging prompt and full disclosure.
VI.
In closing, I would note my fundamental disagreement with this majority’s approach to judicial decision-making, which emphasizes policy considerations over logic and precedent. One obvious problem with this approach is that it renders the law susceptible to sudden directional changes based upon nothing more than a change in the prevailing judicial sentiment or even "the accident of a change in [the Court’s] composition” (Simpson v Loehmann, 21 NY2d 305, 314 [Breitel, J., concurring]). Today, for example, the Court has reduced the Consolazio-Perez-Ranghelle-Jones principle from a logically necessary corollary of the Rosario rule to a mere policy choice. While today’s majority professes its continued devotion to the *660basic principle,6 this demotion certainly opens the way for its further curtailment — or even its reversal — since a rule based on policy alone is readily susceptible to attack by a court that wishes to emphasize other values or policy considerations (compare, People v Bing, 76 NY2d 331, with People v Bartolomeo, 53 NY2d 225).
To be sure, we in the judiciary must operate in the real world, and our decision-making must reflect sensitivity to the practical consequences, as well as the legal and theoretical implications. Ultimately, however, our responsibility in the judiciary is to apply reasoning and precedent with a view toward fairness, both to society and the accused. While society’s interest in the finality of criminal convictions is, no doubt, an important value, so too is society’s interest in the "fundamental objective” of providing the accused with a fair opportunity to test the People’s witnesses through the crucible of cross-examination — the value that informs the Rosario rule (see, People v Jones, supra, at 550, quoting People v Perez, supra, at 158). To the extent that we remain committed to that value, we are also duty bound to uphold it even in contexts where the consequences may be viewed by some as undesirable. Indeed, as has been noted in another context, if a court is "really committed to” a particular principle, "its subsequent decisions * * * [should] nourish” rather than undermine it (People v Bing, supra, at 343).
In sum, I find no reason to depart from the principle that this Court established in People v Consolazio (supra) and has followed without deviation for some 15 years, despite numerous challenges (see, People v Jones, supra; People v Novoa, 70 NY2d, at 499, supra). In the absence of a persuasive statutory or precedential ground for altering the standard based on the procedural context in which the Rosario challenge is raised, I conclude that a CPL 440.10 motion to vacate a judgment because of the prosecution’s failure to deliver Rosario material must be granted — without regard to particularized prejudice— to the same extent that such a failure would constitute error requiring reversal on direct appeal. Accordingly, I vote to affirm.
*661Judges Simons, Kaye and Bellacosa concur with Chief Judge Wachtler; Judge Titone dissents and votes to affirm in a separate opinion in which Judges Alexander and Hancock, Jr., concur.
Order reversed, etc.
. The majority’s reliance on the use of a policy justification in Rosario is misplaced (see, majority opn, at 643), since the issue in this case is not whether the Rosario doctrine itself should be available in postconviction challenges, but rather, whether the per se rule of reversal, which was extrapolated as a natural and logical corollary of Rosario, should also be *652available. As to this question, the rationale utilized in the line of cases establishing the latter rule is the proper source of guidance. Similarly the discussions delineating certain substantive limitations on the scope of the prosecutor’s Rosario duty (People v Poole, 48 NY2d 144; People v Consolazio, 40 NY2d 446, 454), which the majority cites (majority opn, at 644), are completely irrelevant in this context.
. The majority notes that no prior Rosario challenges have reached this Court in precisely the same posture as this appeal and then extrapolates that "applying the rule we articulate today * * * will as a practical matter not affect the vast majority of defendants who raise Rosario claims” (majority opn, at 648). This apparent effort at apology will be cold comfort for those defendants relegated to CPL 440.10 proceedings. Futhermore, there is irony in the majority’s present attempt to minimize the significance of its decision as inapplicable to "the vast majority of defendants who raise Rosario claims.” That assertion is directly contrary to the People’s claim that if the per se rule were applied in CPL 440.10 proceedings, "[e]very criminal conviction, regardless of its age or * * * fairness * * * would thereby be placed at risk” — a contention that members of the majority have found very persuasive.
. While the Novoa Court may have overlooked the procedural posture in which the Rosario issue was presented because it was "not asked to consider [it] at that time” (majority opn, at 641), that circumstance cannot excuse the majority’s present misapprehension about that question.
. Even where the Rosario material has been discovered before disposition of the direct appeal, there remains the possibility that the direct appeal will be determined before the CPL 440.10 motion based on the Rosario violation can be heard and determined. While the defendant in such situations can seek to have the disposition of the direct appeal held in abeyance, it seems capricious and unfair to make the availability of a more favorable standard of review turn upon the defendant’s success in navigating such a series of procedural hurdles.
. I note that the prejudice required by the majority’s decision here is not the same as the prejudice that the Ranghelle decision requires when disclosure is delayed but nonetheless made during trial. The latter includes any prejudice flowing from the impairments to the defense’s trial strategy that were caused by the delayed disclosure. The former, in contrast, appears to permit relief only when the undisclosed evidence has self-evident probative worth that might have affected the guilt determination.
. I reserve a degree of skepticism about the majority’s declared support for the Consolazio-Perez-Ranghelle-Jones rule, particularly since one of its four members has so recently called it an "errant footstep” and urged its reversal by the Legislature (People v Jones, 70 NY2d 547, 553-557 [Bellacosa, J., concurring]).