Chief Justice BENDER
delivered the Opinion of the Court.
I. Introduction
T1 Jason LaRosa, the defendant, confessed to his wife, his mother, his pastor, a police dispatcher, and an investigating police officer [570]*570that he had sexually assaulted his two-and-a-half-year-old daughter. He was charged with various crimes, and a jury convicted him of all charges. On appeal, a division of the court of appeals reversed LaRosa's convie-tions under the corpus delicti rule.1 That rule requires the prosecution to prove that a crime occurred using evidence other than a defendant's confession. The court of appeals reasoned that the prosecution had presented only opportunity evidence (other than the confessions) establishing that LaRosa had an opportunity to commit a crime, not that the crimes in fact occurred. The People appealed to this court, and we granted certiorari to address the viability of the corpus delicti rule.
T2 This case requires us to decide whether to abandon our judicially created corroboration requirement, the corpus delicti rule, and with it over one hundred years of precedent. If we abandon the corpus delicti rule, then we must decide another issue: what corroboration requirement, if any, to articulate in its place. The People argue that we should abandon the corpus delicti rule in favor of the trustworthiness standard, which requires corroborating evidence that proves that a confession is reliable, or, in the alternative, the sufficiency of the evidence test, which requires no corroborating evidence. Instead, the sufficiency of the evidence test would treat confessions like any piece of evidence to be analyzed in the light most favorable to the prosecution during a motion for a judgment of acquittal.
T3 We abandon the corpus delicti rule because we hold that sound reasons exist for doing so. In its place, we articulate the trustworthiness standard, which requires the prosecution to present evidence that proves that a confession is trustworthy or reliable. To determine whether corroborating evidence proves the trustworthiness or reliability of a confession, we hold that the trial court must find that corroboration exists from one or more of the following evidentiary sources: facts that corroborate facts contained in the confession; facts that establish the crime which corroborate facts contained in the confession; or facts under which the confession was made that show that the confession is trustworthy or reliable.
1 4 Applying the trustworthiness standard to this case raises another issue. We must decide whether applying the trustworthiness standard here would violate LaRosa's due process right to have fair warning of any judicial decision altering a common law doctrine of criminal law. We hold that, because we have consistently applied the corpus de-licti rule as a substantive principle of law for over one hundred years, LaRosa did not have fair warning of our decision to abandon it. Thus, we are constitutionally prohibited from applying the trustworthiness standard in this case.
15 Hence, we affirm the court of appeals' decision reversing LaRosa's convictions. We remand the case to that court with directions to return it to the trial court to enter a judgment of acquittal.
II. Facts and Proceedings Below
T6 While in Florida, LaRosa called an emergency dispatch operator in Douglas County and told her that he had had inappropriate sexual contact with his two-and-a-half year-old daughter. LaRosa explained that he had already called his wife, mother, and pastor and told them the same thing. The dispatcher told LaRosa that an investigating officer would call him back. When called back, LaRosa told the investigating officer that he had taken his daughter swimming at a community recreation center and had performed oral sex on her in a private family shower while he masturbated. LaRosa told the investigating officer that he would return to Colorado and turn himself in to authorities, which he did. Upon arrival, LaRosa was arrested and charged with sexual assault on a child, sexual assault on a child by one in a position of trust, and aggravated incest.2
[571]*57117 Before trial, LaRosa filed a motion to dismiss the charges under the corpus delicti rule. He argued that his confessions to the dispatcher and the investigating officer were insufficient evidence to sustain a conviction because there was no physical evidence that a crime had occurred. There were no eyewitnesses, and his daughter could not remember the incident. The prosecution contended that it would present evidence at trial corroborating LaRosa's confessions, including medical evidence and the daughter's statements to a social worker. Ultimately, the prosecution presented neither. The trial court denied LaRosa's motion, and the case went to trial.
18 The prosecution's case depended on LaRosa's confessions to the dispatcher and the investigating officer, both of which were admitted into evidence. The dispatcher testified that she had training and experience dealing with callers who suffer from mental illness, and LaRosa had seemed "very lucid" during their conversation. The investigating officer testified that he also had experience dealing with mentally ill callers, and during his confession LaRosa had provided a coherent story, explained his motivation for confessing, described his expectations of what would happen to him, and at no point suggested "he was out of touch with reality." The prosecution's other evidence consisted of the recreation center's visitor logs, which showed that LaRosa had visited the center several times, and photographs of the family shower rooms.
19 Following the prosecution's case-in-chief, LaRosa moved for a judgment of acquittal, again based on the corpus delicti rule. LaRosa argued that the prosecution's evidence was opportunity evidence showing only that he had the opportunity to commit a crime, not that the crimes occurred. The trial court denied LaRosa's motion.
'T 10 The trial continued, and LaRosa testified that, when he confessed, he was suffering from marital and financial problems, was "extremely tired," and had confessed to fictitious events. The jury convicted him of all charges.
T11 On appeal, a division of the court of appeals reversed LaRosa's convictions based on the corpus delicti rule. That court reasoned that the prosecution had not presented evidence other than his confessions to establish that the crimes occurred. Rather, the prosecution's evidence established that La-Rosa had an opportunity to sexually assault his child, "which every custodial parent has on a virtually continuing basis." The court of appeals therefore reversed LaRosa's convictions and directed the trial court to enter a judgment of acquittal. The People appealed to this court, and we granted certiorari to examine the viability of the corpus delicti rule in Colorado.3
III. Applicable Law
T 12 Before addressing the parties' contentions, we provide an overview of the applicable law. Specifically, we discuss the judicially created corroboration requirement and its two formulations at issue here, the corpus delicti rule and the trustworthiness standard.
1 13 Almost all courts adhere to a corroboration requirement, which requires the prosecution to present corroborating evidence of a defendant's confession to either allow for its admission into evidence or sustain a conviction. Kenneth S. Broun, McCormick on Evidence § 145, at 592-98 (6th ed. 2006). We adhere to the conventional formulation of the corroboration requirement, the corpus delicti rule. See Downey v. People, 121 Colo. 307, 819, 215 P.2d 892, 898 (1950). Federal courts and a growing number of state jurisdictions adhere to a newer formulation of the corroboration requirement, the trustworthiness standard. See State v. Mauchley, 2003 UT 10, ¶ 19, 67 P.3d 477, 482-83.
The Corpus Delicti Rule
114 To establish guilt in a eriminal case, the prosecution must prove the corpus delicti, or "body of the crime." This means [572]*572that the prosecution must prove that the crime occurred.4
{15 From this corpus delicti concept, we have derived the corpus delicti rule. That rule requires the prosecution to present evidence other than a defendant's confession that proves that the crime occurred. Downey, 121 Colo. at 320, 215 P.2d at 899. Without corroborating evidence, a defendant's confession is insufficient evidence to sustain a conviction. Roberts v. People, 11 Colo. 213, 216, 17 P. 637, 639 (1888). The corroborating evidence "need only be slight," but it must be sufficient "to convince the jury that the crime is real and not imaginary." Neighbors v. People, 168 Colo. 319, 322, 451 P.2d 264, 265 (1969).
T 16 The People state, and our independent research confirms, that the earliest reported case applying the corpus delicti rule in Colorado is from 1872. See Dougherty v. People, 1 Colo. 514, 524 (1872). Since then, we have consistently applied this rule. See People v. Rankin, 191 Colo. 508, 510, 554 P.2d 1107, 1108 (1976); People v. Smith, 182 Colo. 31, 33, 510 P.2d 893, 894 (1978); Meredith v. People, 152 Colo. 69, 71, 380 P.2d 227, 227-28 (1963).
117 There seems to be little consensus concerning the rationale behind the rule, but courts typically rely on some amalgam of the following: protecting defendants from conviction based on confessions to imaginary crimes; avoiding reliance on coerced confessions extracted under police pressure; and promoting better police investigations by ensuring that they "extend beyond the words of the accused." Smith v. United States, 348 U.S. 147, 153, 75 S.Ct. 194, 99 L.Ed. 192 (1954); People v. McMahan, 451 Mich. 543, 548 N.W.2d 199, 204 (1996) (Boyle, J., dissenting); see also McCormick on Evidence § 145, at 595-96. The rule's objective is "relatively modest": to reduce the possibility that a person is convicted based on a confession to a crime that never happened. McCormick on Evidence § 145, at 595; see also People v. Jones, 17 Cal.4th 279, 70 Cal.Rptr.2d 793, 949 P.2d 890, 902-03 (1998).
18 Due in part to its "extremely limited function," Smith, 348 U.S. at 153, 75 S.Ct. 194, the rule has been subject to widespread criticism.5 This criticism has led federal courts and a growing number of state jurisdictions to abandon the corpus delicti rule in favor of the trustworthiness standard.
The Trustworthiness Standard
19 The trustworthiness standard derives from three United States Supreme Court cases announced the same day in 1954. Opper v. United States, 348 U.S. 84, 75 S.Ct. 158, 99 L.Ed. 101 (1954); Smith v. United States, 348 U.S. 147, 75 S.Ct. 194, 99 L.Ed. 192 (1954); United States v. Calderon, 348 U.S. 160, 75 S.Ct. 186, 99 L.Ed. 202 (1954).
¶ 20 In Opper, the Court rejected the corpus delicti rule and adopted, without extensive explanation, the "better rule" that "corroborative evidence need not be sufficient, independent of the statements, to establish the corpus delicti." Opper, 348 U.S. at 93, 75 S.Ct. 158. Instead, the prosecution must "introduce substantial independent evidence which would tend to establish the trustwor[573]*573thiness of the statement." Id. Smith elaborated on the quantum of evidence necessary to establish the trustworthiness of a confession, noting that "[alll elements of the offense must be established by independent evidence or corroborated admissions, but one available mode of corroboration is for the independent evidence to bolster the confession itself and thereby prove the offense 'through' the statements of the accused." Smith, 348 U.S. at 156, 75 S.Ct. 194.
121 What these eryptic pronounce ments mean is unclear. Not surprisingly, federal courts have struggled to interpret Opper and Smith's trustworthiness standard consistently. What is clear is that the trustworthiness standard is different in its focus than the corpus delicti rule. It focuses on whether corroborating evidence establishes the trustworthiness or reliability of the confession, whereas the corpus delicti rule focuses on whether corroborating evidence establishes that the crime occurred. See United States v. Johnson, 589 F.2d 716, 718-19 (D.C.Cir.1978).
IV. Analysis
[22 Having outlined the applicable law, we now turn to the parties' contentions. LaRo-sa contends, and the court of appeals agreed, that this case involves opportunity evidence that does not corroborate his confessions under the corpus delicti rule, which has been a substantive principle of this court's precedent for more than a century. On that basis, he urges us to affirm the court of appeals. The People argue that we should abandon the corpus delicti rule because it does not serve the purposes justifying its original existence. LaRosa counters that we should retain the corpus delicti rule because it still operates to protect defendants from the consequences of false confessions and because stare decisis compels us to. If we abandon the corpus delicti rule, the People argue that we should do so in favor of the trustworthiness standard, or, in the alternative, the sufficiency of the evidence test. Under either alternative, the People urge us to reverse the court of appeals and reinstate LaRosa's convictions.
128 To analyze the parties' contentions, we first examine the corpus delicti rule and its criticisms, consider whether we can abandon the corpus delicti rule in light of stare decisis principles, and conclude that we can. We next discuss the People's alternative arguments in favor of the trustworthiness standard and the sufficiency of the evidence test and conclude that the trustworthiness standard is the better rule because it provides defendants with some minimal protection from convictions based on false confessions. We then turn to articulating the trustworthiness standard, conclude it is confusing, and articulate our own version.
A.
T24 The People contend that we should abandon the corpus delicti rule and with it over one hundred years of precedent, despite the doctrine of stare decisis. In support, the People refer us to the rising tide of criticism directed at the rule.
T 25 The rule has been criticized for inadequately serving its admittedly "limited funetion." See Smith, 348 U.S. at 153, 75 S.Ct. 194. It exists to detect false confessions but does so in only one cireumstance: when a person confesses to an imaginary crime. It does nothing to protect a person who confesses to a crime committed by someone else. See Mauchley, ¶ 22, 67 P.3d at 483. Courts have questioned the logic of that distinction. See id.; State v. Lucas, 30 N.J. 37, 152 A.2d 50, 60 (1959) (stating that "(there seems to be little difference in kind between convicting the innocent where no crime has been committed and convicting the innocent where a crime has been committed, but not by the accused").
126 The rule has also been criticized as outdated. Since its inception, the United States Supreme Court has recognized additional constitutional and procedural safeguards concerning the voluntariness of confessions that have led some courts to question whether the rule is obsolete. See State v. Parker, 315 N.C. 222, 337 S.E.2d 487, 494 (1985) (noting that the rule's concern with coercive police tactics in obtaining confessions has been undercut by Miranda v. Arizona, 384 U.S. 436, 86 S.Ct. 1602, 16 L.Ed.2d 694 (1966)). Additionally, [574]*574since courts first began applying the corpus delicti rule, criminal statutes have become more numerous and complex, making the corpus delicti difficult, if not impossible, to define for certain crimes. See Mauchley, ¶ 31, 67 P.3d at 485 n. 4.
127 Finally, the rule has been criticized for its potential to obstruct justice in cases where, as here, the victim is too young to testify and no tangible injury results from the alleged criminal act. See State v. Ray, 130 Wash.2d 673, 926 P.2d 904, 905, 907 (1996) (applying the corpus delicti rule to reverse the conviction of a defendant who confessed to forcing his three-year-old daughter to fondle his penis). In such situations, the rule may operate to reward defendants who target young or mentally infirm victims who are unable to testify and commit crimes that do not result in tangible injuries or do so carefully and leave no evidence. See McMahan, 548 N.W.2d at 207 (Boyle, J., dissenting) (criticizing the rule for creating this "socially aberrant result"). That the rule may operate to bar conviction for crimes committed against the most vulnerable vie-tims, such as infants, young children, and the mentally infirm, and for crimes that are especially egregious, such as sexual assault and infanticide, has been described as "especially troublesome." See Mauchley, ¶ 29, 67 P.3d at 485; see also Maria Lisa Crisera, Comment, Reevaluation of the California Corpus Delicti Rule: A Response to the Invitation of Proposition 8, 78 Calif, L. Rev. 1571, 1583 (1990) (discussing the rule's potential to obstruct justice in cases involving child abuse and infanticide because it can be difficult, if not impossible, to establish that such injuries resulted from eriminal acts).
¶ 28 With these criticisms in mind, we must analyze whether to abandon the corpus delicti rule in light of stare decisis. Stare decisis is a judge-made doctrine that promotes uniformity, certainty, and stability of the law. Friedland v. Travelers Indem. Co., 105 P.3d 639, 644 (Colo.2005). It requires a court to follow the rule of law it has established in earlier cases unless "sound reasons exist." People v. Blehm, 983 P.2d 779, 789 (Colo.1999). Stare decisis is not an inflexible or immutable rule, but it requires us to apply precedent (here, the corpus de-licti rule) unless we are convineed that it was originally erroneous or is no longer sound due to changed conditions, and more good than harm will come from departing from it. See Friedland, 105 P.3d at 644; Blehm, 983 P.2d at 788.
129 Determining whether the rule was originally erroneous requires us to examine if the rule accomplishes its purpose of protecting defendants from false confessions. As noted, the corpus delict rule seeks to accomplish this purpose by requiring the prosecution to prove the corpus delicti with evidence other than a defendant's confession. By focusing on the corpus delicti, however, the rule draws an untenable distinction between defendants who confess to imaginary crimes and those who confess to crimes committed by others. Because the rule operates to protect defendants in only one specific cireumstance, when a defendant confesses to an imaginary crime, it fails to comport with its purpose of detecting false confessions. This incongruity has existed since the rule's inception by virtue of its inherently flawed design. Thus, we conclude that the corpus delicti rule was originally erroneous. See Mauchley, ¶¶ 27, 46, 67 P.3d at 484, 488.
¶ 30 In the alternative, we may also consider whether the rule is no longer sound due to changed conditions. As noted, Mi-ranmda and similar constitutional doctrines now exist to protect defendants from the overzealous interrogation techniques of police officers. Thus, insofar as one of its original purposes was to protect defendants from coercive police tactics, the rule is no longer necessary. Further, the rule has become difficult, if not impossible, to apply to certain crimes, in part because statutory crimes have proliferated and become more complex. See People v. Trujillo, 860 P.2d 542, 545 (Colo.App.1992) (noting that no tangible injury can be isolated as the corpus delicti for inchoate crimes such as conspiracy, attempt, or solicitation). Finally, our cases have stressed the insubstantial quantum of proof necessary to establish the corpus delicti, rendering its value as a corroboration requirement suspect. See Hampton, 146 Colo. at 574, 362 P.2d at 866 (stating that the corroborating evidence [575]*575"need be only slight"). Given these legal developments, we conclude that the rule is no longer sound.
T31 Last, we must determine whether abandoning the rule will do more good than harm. We are troubled that the rule works to bar convictions in cases involving the most vulnerable victims, such as infants, young children, and the mentally infirm. We are also aware that the rule operates disproportionately in cases where no tangible injury results, such as in cases involving inappropriate sexual contact, or where criminal ageney is difficult or impossible to prove, such as in cases involving infanticide or child abuse. Indeed, in Colorado, LaRosa's case is not the first of its type in which the rule has been invoked to bar conviction for sexual assault against a young child. See Meredith, 152 Colo. at 72, 380 P.2d at 228 (applying corpus delicti rule to reverse the conviction of a defendant who confessed to molesting a five-year-old boy); Robson, 80 P.3d at 913-14 (applying corpus delicti rule to affirm the trial court's dismissal of charges against a defendant who confessed to sexually assaulting his infant daughter). Because the rule may operate to obstruct justice, we conclude that abandoning it will do more good than harm. We are convinced that the corpus delieti rule is too rigid in its approach, too narrow in its application, and too capable of working injustice in cases where, as here, evidence of the corpus delicti is not only nonexistent but impossible to uncover. See Parker, 337 S.E.2d at 494; Mauchley, ¶ 46, 67 P.3d at 488. Thus, we abandon the corpus delieti rule because we hold that sound reasons exist for doing so.
B.
¶ 32 Having agreed with the People that we should abandon the corpus delicti rule, we next address what corroboration requirement, if any, we should articulate in its place. On this issue, the People present alternative arguments. The People argue in favor of the trustworthiness standard because it better accomplishes the purpose that the corpus delicti rule is meant to serve. In the alternative, the People argue that any corroboration requirement, including the trustworthiness standard, conflicts with the sufficiency of the evidence test announced in People v. Bennett, 188 Colo. 125, 515 P.2d 466 (1973), because it treats confessions as inherently unreliable and prohibits the trial court from considering confessions in the light most favorable to the prosecution. As such, the People argue that we should abandon any corroboration requirement and analyze confessions under the sufficiency of the evidence test articulated in Bennett.
133 There is support for the People's contention that Bennett conflicts with the corroboration requirement. Benneit requires a court, when ruling on a motion for a judgment of acquittal, to analyze the evidence in the light most favorable to the prosecution and determine whether it is "substantial and sufficient to support a conclusion by a reasonable mind that the defendant is guilty of the charge beyond a reasonable doubt." Bennett, 183 Colo. at 130, 515 P.2d at 469. A corroboration requirement, by contrast, would require a reviewing court to invalidate a conviction that satisfies Bennett if it rested solely on an uncorroborated confession. See United States v. Brown, 617 F.3d 857, 862 (6th Cir.2010). The People's contention highlights the substantive and procedural overlap between the sufficiency of the evidence test and the corroboration requirement, but they are not in irreconcilable conflict.
¶ 34 For one, they serve different purposes. The sufficiency of the evidence test is constitutionally mandated to ensure that the prosecution proves every element of an offense beyond a reasonable doubt. People v. Gonzales, 666 P.2d 123, 127 (Colo.1983). By contrast, the corroboration requirement is not constitutionally mandated but requires the prosecution to present corroborating evidence of a defendant's confession to assuage our long-standing concern about false confessions.
185 Further, both doctrines are satisfied by different evidentiary requirements. The sufficiency of the evidence test is an elemental test. Its focus is on the "substantive elements of the criminal offense." Jackson v. Virginia, 443 U.S. 307, [576]*576324 n. 16, 99 S.Ct. 2781, 61 L.Ed.2d 560 (1979). It requires the court to consider whether a reasonable mind could conclude that "each material element of the offense was proven beyond a reasonable doubt." Bennett, 183 Colo. at 132, 515 P.2d at 470; see also Gonzales, 666 P.2d at 128 (applying the Bennett test to the specific statutory elements of possession of contraband). By contrast, the corroboration requirement is not an elemental test because it does not require the court to review the evidence as it relates to the specific elements of the crime. Instead, it focuses on whether the prosecution presented evidence to corroborate a confession.
36 These differences are best illustrated by considering the facts of this case. LaRo-sa's confessions, when viewed in the light most favorable to the prosecution, would invariably support a conclusion by a reasonable person of guilt beyond a reasonable doubt. This is so because, by definition, a "confession" is an admission of guilt. Any confession would therefore meet Bennett's sufficiency of the evidence test, irrespective of its reliability. As such, a corroboration requirement may require the prosecution to present additional and different evidence than that required under Bennett if it seeks to rely on a defendant's confession to secure a conviction. Although similar, the sufficiency of the evidence test and the corroboration requirement are not in irreconcilable conflict. Thus, Bennett does not prohibit us from articulating a different corroboration requirement if necessary to protect defendants from false confessions.
137 Without engaging in an empirical battle over the frequency with which false confessions occur, we recognize that some defendants on occasion do confess to nonexistent crimes or crimes committed by others.6 Despite its "extremely limited function," Smith, 348 U.S. at 153, 75 S.Ct. 194, the corroboration requirement thus serves a valid purpose: to prevent "errors in conviec-tions based upon untrue confessions alone," Warszower v. United States, 312 U.S. 342, 347, 61 S.Ct. 603, 85 L.Ed. 876 (1941). This problem, though possibly overstated in the caselaw, is compounded by the fact that confessions "stand high in the probative hierarchy of proof." See Lucas, 152 A.2d at 61; see also Stephen A. Saltzburg, Standards of Proof and Preliminary Questions of Fact, 27 Stan. L. Rev. 271, 293 (1975). Given the persuasive power a confession may hold over a jury, courts have imposed safeguards, including various corroboration requirements, to restrict the jury's power to convict on the basis of a confession alone.7 To this end, we [577]*577are persuaded that some corroboration requirement is necessary to protect defendants from false confessions.
38 As our analysis has shown, the corpus delicti rule inadequately addresses that problem by focusing on the corpus delicti rather than on the confession itself. The trustworthiness standard is more effective than the corpus delicti rule because its focus comports with its purpose: it seeks to detect false confessions by focusing on whether a confession is true or false. The trustworthiness standard also responds to the criticisms of the corpus delicti rule. It is easier to apply to complex or inchoate crimes because the corpus delicti does not have to be defined, and it is less likely to work injustice in cases where no evidence of the corpus delicti exists. See Opper, 348 U.S. at 93, 75 S.Ct. 158; McCormick on Evidence § 148, at 605. Finally, the trustworthiness standard is not duplicative of Mirando and similar constitutional safeguards because it protects defendants from false confessions, not involuntary ones.8 Thus, because the trustworthiness standard better accomplishes the objectives the corpus delicti rule seeks to accomplish, we abandon the corpus delicti rule in favor of the trustworthiness standard.
C.
T 39 We now turn to articulating that standard. Although easy to repeat in principle, "Itlhe doctrinal nature and procedural concomitants of the trustworthiness requirement announced in Opper are not entirely clear." Singleterry, 29 F.3d at 737. Courts have not interpreted Opper and Smith consistently, resulting in different approaches to the trustworthiness standard that can be confusing, complicated, or at odds with the flexibility that is one of its hallmarks.9 Our task, then, is to articulate a standard that is both comprehensible and capable of consistent application.
140 Under the trustworthiness standard, the prosecution is not required to present evidence other than a defendant's confession to establish the corpus delicti. Rather, the prosecution must present evidence that proves the trustworthiness or reliability of a confession. See Opper, 348 U.S. at 93, 75 S.Ct. 158. The evidence is sufficient if the corroborating evidence "supports the essential facts admitted sufficiently to justify a jury inference of their truth." Id. "[Thhe corroborating facts may be of any sort what[578]*578ever, provided only that they tend to produce a confidence in the truth of the confession." Wigmore on Evidence § 2071, at 511 (emphasis in original).
¶ 41 To determine whether corroborating evidence proves the trustworthiness or reliability of a confession, we hold that the trial court must find that corroboration exists from one or more of the following evidentiary sources: facts that corroborate facts contained in the confession; facts that establish the crime which corroborate facts contained in the confession; or facts under which the confession was made that show that the confession is trustworthy or reliable.
.¶42 A related but different issue arises as to when the trial court should make this determination. There is confusion in the caselaw as to whether the trustworthiness standard is a rule affecting the admissibility of evidence or the sufficiency of the evidence.10 We think the better approach is to treat the trustworthiness standard, at least for procedural purposes, like a rule affecting the sufficiency of the evidence to be analyzed by the court following a motion for a judgment of acquittal. This approach is consistent with the United States Supreme Court's treatment of the trustworthiness standard 11 and our treatment of our previous corroboration requirement, the corpus delicti rule. See Robson, 80 P.3d at 913-14 (noting that we have consistently "treated the [corpus delicti rule] as a substantive rule of law relating to the quantum of proof necessary to sustain a conviction" and not a rule affecting admissibility). As a practical matter, this procedure also makes sense because the trial court will be better equipped to evaluate the evidence and make its determination after the prosecution has presented its case.
V. Application
¶ 43 Having outlined the trustworthiness standard, we would ordinarily apply it to this case. LaRosa contends, however, that applying the trustworthiness standard here would violate the constitutional prohibition against ex post facto laws. The People counter that the Ex Post Facto Clause applies to legislative acts, not judicial decision making. As such, the People argue that LaRosa's contention should be analyzed under the Due Process Clause and rejected because he had "fair warning" of our decision to abandon the corpus delicti rule.
144 The parties' contentions reflect confusion about whether the United States Supreme Court's Ex Post Facto Clause jurisprudence applies to judicial decision making. In Rogers v. Tennessee, 532 U.S. 451, 121 S.Ct. 1693, 149 LEd.2d 697 (2001), the Supreme Court held that it does not. The Court held that judicial ex post facto claims must be analyzed under the Due Process Clause and "in accordance with the more basic and general principle of fair warning." Id. at 452, 121 S.Ct. 1693. The principle of fair warning is violated when a "judicial alteration of a common law doctrine of criminal law" was "unexpected and indefensible." Id. at 462, 121 S.Ct. 1693 (quoting Bouie v. City of Columbia, 378 U.S. 347, 354, 84 S.Ct. 1697, 12 L.Ed.2d 894 (1964)). Before Rogers was decided, we similarly held that the "key test in determining whether the due process clause precludes the retrospective application of a judicial decision in a criminal case is whether the decision was sufficiently foreseeable so that the defendant had fair warning." Aue v. Diesslin, 798 P.2d 436, 441 (Colo.1990). Accordingly, we analyze LaRosa's [579]*579contention under the Due Process Clause and its guarantee of fair warning.
1 45 Because Rogers involved similar facts, it provides guidance for our analysis. In Rogers, the Court considered whether retroactive application of the Tennessee Supreme Court's decision abolishing the common law "year and a day rule" violated the Due Process Clause. Rogers, 532 U.S. at 453, 121 S.Ct. 1693. The Court concluded that it did not because the rule was "widely viewed as an outdated relic of the common law" dating back to the thirteenth century. Id. at 462, 121 S.Ct. 1693. Perhaps most importantly, the Court described the paucity of Tennessee caselaw applying the rule, noting that it had never served as a ground of decision in any homicide prosecution in the state and had only been mentioned in three cases, each time in dicta, Id. at 464, 121 S.Ct. 1693. Although the rule was a "substantive principle" of Tennessee law, the Court characterized it as "a principle in name only" because it had never been enforced. Id. at 466, 121 S.Ct. 1693. The Court therefore concluded that the Tennessee Supreme Court's decision was a routine exercise in common law decision making so that applying its decision retroactively did not violate due process. Id. at 467, 121 S.Ct. 1693.
" 46 The corpus delicti rule, although widely criticized, is still followed in many state jurisdictions. Indeed, several state courts have reaffirmed it, presumably rejecting similar arguments to those we have found persuasive here. See McMahan, 548 N.W.2d at 201; Ray, 926 P.2d at 906. Further, unlike the "year and a day rule" in Rogers, we have applied the corpus delicti rule as a substantive principle of Colorado law for over one hundred years. See Smith, 182 Colo. at 34, 510 P.2d at 895. The rule has been regularly invoked to bar convictions, occasionally in cases similar to this one. See Robson, 80 P.3d at 914. Thus, we conclude that LaRosa did not have "fair warning" of our decision to abandon the corpus delicti rule in favor of the trustworthiness standard. Because La-Rosa did not have fair warning of our decision, we hold that applying the trustworthiness standard here would violate his due process rights.
[ 47 Henee, because we are constitutionally prohibited from applying the trustworthiness standard here, and because the People concede that the evidence is insufficient to sustain LaRosa's convictions under the corpus delicti rule, we affirm the court of appeals' decision reversing LaRosa's convictions.
VI. Conclusion
48 For the reasons stated, we affirm the court of appeals and remand the case to that court with directions to return it to the trial court to enter a judgment of acquittal.
Justice COATS dissents, and Justice EID joins in the dissent.