concurring and dissenting.
The majority has today, for the first time, found the sentence of death to be disproportionate to the penalty imposed in similar cases and, as a consequence, has modified the defendant’s sentence to life imprisonment without the possibility of parole. I concur in this result. If, however, there are those who would trumpet the majority opinion as proof positive that the proportionality protocol works as it should, I move quickly to temper their voices. This result comes from a protocol I perceive as flawed and unreliable. Consequently, in my view, the holding evidences neither reliability nor consistency. Rather, the protocol remains flawed, and a flawed protocol, by definition, produces a flawed result.
The principle underlying comparative proportionality review is that it is unjust to impose a death sentence upon one defendant when other defendants, convicted of similar crimes with similar facts, receive sentences of life imprisonment (with or without parole). Using the appropriate protocol, a properly conducted proportionality review responds to this problem by permitting the judiciary to engage in a “judicial field leveling” process, for lack of a better description. Thus, proportionality review serves a crucial role as an “additional safeguard against arbitrary or capricious sentencing.” State v. Bland, 958 S.W.2d 651, 668 (Tenn.1997). In a line of dissents beginning with State v. Chalmers, I have identified three perceived flaws in the protocol currently embraced by the majority: (1) the proportionality test is overbroad; (2) the “pool” of cases used for comparison is inadequate; and (3) the review is too subjective. See 28 S.W.3d 913, 923 (Tenn.2000) (Birch, J., concurring and dissenting). I continue to offer my objections to the majority analysis in the hope *794that comparative proportionality review may be reformed to more effectively fulfill the goals for which it was intended.
In order to remedy the first of these perceived flaws, the overbreadth of the protocol, I submit that the protocol should be refined to more accurately identify disproportionate sentences. The majority holds a sentence is disproportionate only if the case under review “is plainly lacking in circumstances consistent with those in similar cases in which the death penalty has been imposed.” Bland, 958 S.W.2d at 665. Even if a defendant can show that others received life sentences for similar crimes and no discernible basis exists to distinguish the cases, the sentence will “not necessarily [be found] disproportionate.” Id. This test, however, does not seem reliably gauged to identify disproportionate sentences. Under the current protocol, a sentence may be found “proportionate” based on minimal similarities to a prior death penalty case even if the defendant can point to similar cases in which a life sentence was imposed. “Proportionality” implies consistency and balance in sentencing, neither of which is accomplished when distinguishable penalties are imposed in indistinguishable cases.
Other jurisdictions provide models for a more meaningful and objective proportionality protocol. In these jurisdictions, the circumstances of each case are analyzed to determine whether its characteristics are more consistent with other capital cases wherein a death sentence has been imposed. Typical of those jurisdictions is New Jersey, whose Supreme Court. has stated that “[a] capital sentence is excessive and thus disproportionate if other defendants with characteristics similar to those of the defendant ... generally receive sentences other than death for committing factually-similar crimes in the same jurisdiction.” New Jersey v. DiFrisco, 142 N.J. 148, 662 A.2d 442, 448 (1995) (citing New Jersey v. Martini, 139 N.J. 3, 651 A.2d 949 (1994)). Essentially the same test has been cited with approval by the Court of Appeals of Maryland. See Tichnell v. Maryland, 297 Md. 432, 468 A.2d 1, n. 18 (1983). Even the Supreme Court of Virginia, which has never reversed a case on comparative proportionality grounds, focuses on “whether generally juries in [the same] jurisdiction impose a death sentence for conduct similar to that of the defendant.” Stamper v. Virginia, 220 Va. 260, 257 S.E.2d 808, 824 (1979). Thus, I would urge the adoption of a similarly gauged test to more accurately assess whether a given case, viewing its circumstances as objectively as possible, is more consistent with the circumstances of similar capital cases and capital defendants wherein the death penalty was not imposed.
Reforming the test used to identify disproportionate sentences, however, is but one part of the analysis. If proportionality review is to be effective, this Court also must ensure that relevant cases are not omitted from the comparison pool. Concerning the paucity of the pool, in Bland, I expressed grave concern that the pool of cases used by the majority in comparative proportionality review analysis was too small. 958 S.W.2d at 679 n. 1 (Birch, J., concurring and dissenting). I reiterate that concern today. I would expand the pool to include, at a minimum, all cases in which the defendant has been convicted of first degree murder regardless of penalty. Ideally, however, the pool should include all cases in which the defendant was initially indicted for a capital offense. In determining proportionality, one must compare all similar crimes and defendants, not just those defendants whose prosecution is more vigorously pursued by the State. Omitting cases in which the death penalty was not sought skews the propor*795tionality analysis, for similar life cases upon which the defendant otherwise could rely often will be excluded.1 While it is acceptable and appropriate that the State should have discretion in seeking the death penalty, it is neither logical nor appropriate for the State to indirectly determine, by the use of prosecutorial discretion, the cases this Court ultimately may consider in determining comparative proportionality aspects of an individual defendant’s death sentence.
Other states have approved comparison of all death-eligible cases regardless of whether the death penalty was sought. See, e.g., New Jersey v. Morton, 165 N.J. 235, 757 A.2d 184, 189 (2000) (holding that the court would “consider all death-eligible cases, whether or not they were capitally prosecuted, because the State’s decision not to prosecute the defendant capitally does not necessarily reflect on the defendant’s lack of deathworthiness”); Washington v. Brown, 132 Wash.2d 529, 940 P.2d 546, 561 (1997) (“This pool of similar cases includes those in which the death penalty was sought and those in which it was not.”); see also Tichnell, 468 A.2d at 18 (Md.1983) (including in the pool cases in which the State sought the death penalty, but noting that “we do not preclude any defendant ... [from arguing] that designated non-capital murder cases are similar to the case then under scrutiny and should be taken into account”). Only by implementing such a standard will the Court ensure that our protocol provides a reliable proportionality assessment.
Beyond broadening the pool to more effectively consider all similar first degree murder cases, I submit that one of the primary means for review of that pool of cases, this Court’s Rule 12 database, may also need intensive scrutiny and extensive supplementation. Recently, questions have begun to surface concerning whether the Rule 12 database is sufficiently accurate or complete to justify reliance by this Court in crucial reviews of death penalty cases. Especially significant is a series of articles which appeared in The Tennessean newspaper. One of the articles suggests that glaring and pervasive flaws exist in the Rule 12 database. See John Shiftman, Missing Files Raise Doubts About Death Sentences, The Tennessean (Nashville), July 22, 2001, at Al. The majority’s admission that “Rule 12 reports have not been filed in every prior case” is, at best, a substantial understatement. Shiftman writes, “Three of every five first-degree murder convictions [for which a report was *796required] are missing from the database. So, too, is one of every five death penalty cases, records show. What’s more, hundreds of cases included in the database ... are missing important details about the crime, defendant, and victim.” Id. As irony would have it, State v. Bland, 958 S.W.2d 651 (Tenn.1997), the case in which this Court unveiled the “preparation of a Tennessee CD Rom death penalty database which will be used [for proportionality analysis] by this Court and accessible to litigants,” is not included in the database. More disturbingly, the fact that life cases are much more likely to be omitted from the database than death cases suggests that defendants are placed at a significant disadvantage in their effort to locate those cases upon which claims of disproportion-ality may be based. While “traditional research methods” may, as the majority suggests, supplement the Rule 12 reports, such methods are at once more expensive, time consuming, and difficult than a search of the Rule 12 database. Thus, the errors and omissions from the Rule 12 database, whether as extensive as suggested or not, present a very real obstacle for defendants. Moreover, the majority’s platitudinous assertion that “experienced judgment and institutional knowledge”2 will somehow overcome the flaws plaguing other facets of the proportionality analysis is likely to be of little comfort to defendants facing a review protocol which is already too subjective. In my view, the errors and omissions in the Rule 12 database, whether many or few, inject into this Court’s proportionality review a real possibility of error. Until remedied, this flawed database stands as yet another obstacle to a fair and reliable comparative proportionality review in Tennessee.
Beyond the test used in the majority protocol and the cases to be included in the comparison pool, there remains one additional area of concern in proportionality analysis-the lack of objectivity in identifying which cases are “similar” to the case under review for the purposes of proportionality analysis. I would provide a more objective means for identifying the cases to be used for comparison. As I stated in Chalmers, “[without some objective standard to guide reviewing courts, ‘proportionality’ becomes nothing more than a statement that the reviewing court was able to describe the case before it in terms comparable to other capital cases.” 28 S.W.3d at 924 (Birch, J., concurring and dissenting). The approach taken in Washington may be instructive. In Washington v. Pirtle, the Washington Supreme Court noted that quantifiable details such as the number of aggravating circumstances, victims, and prior convictions should be emphasized “in order to be objective as possible.” 127 Wash.2d 628, 904 P.2d 245, 276 (1995). Although the Pirtle court noted that proportionality is not a task reducible to statistics, it nonetheless emphasized that “numbers can point to areas of concern.” Id.3
*797Within the framework provided by Bland, objective criteria could be used to more clearly select which cases should be used for comparison, thus lessening the inclination to subjectively characterize cases in order to reach desired results. Many factors recorded in the trial court’s Rule 12 report in first degree murder cases, such as the number and type of aggravating and mitigating factors found by the jury, the defendant and victim’s age, race, and gender, the existence or lack of prior convictions, and the number of victims, may be objectively quantified. A case which does not share a significant number of these objective criteria should be used only in rare instances. While more subjective factors such as motive and manner of death may further refine the comparison, these factors are too malleable to justify primary reliance. Emphasis on objective factors would ensure consistency in the review protocol.
The scope of the analysis employed by the majority appears to be rather amorphous and undefined-expanding, contracting, and shifting as the analysis moves from case to case. The difficulties inherent in applying this protocol become evident when applied to the case under submission. The majority lists several factors that it considered in its proportionality analysis. Inter alia, it notes that (1) the defendant reacted violently when the victim would not stop crying; (2) the defendant inflicted serious, and ultimately fatal, injuries; (3) the defendant’s conduct was not provoked or justified; (4) the victim was helpless; (5) the killing did not appear to be premeditated; (6) the defendant was caring for the victim while the victim’s mother was at work; and (7) the defendant delayed seeking assistance for the victim. See Majority op. at 790. All of these factors, however, are also present in State v. Torres,4 one of the “death” cases the majority includes in its comparison pool. Indeed, except for one additional aggravating circumstance found by the jury5 and evidence that the victim in Torres suffered prior abuse, the two cases are remarkably similar.
While the victim in Torres may have suffered more serious abuse, the majority’s conclusion that this case is “plainly lacking in circumstances” similar to Torres seems inconsistent with prior cases in which the majority has sometimes upheld death sentences by comparison to cases so widely divergent that they share no similarity but a single aggravating circumstance found by the jury. In State v. Bane,6 for example, the defendant was convicted of choking and stabbing an elderly victim during a planned robbery; the Court upheld the death sentence on a finding that the case was not “plainly lacking in circumstances” with such cases as State v. Vann,7 involving an eight-year-old victim killed during the perpetration of *798aggravated rape and incest; State v. Hall,8 involving a defendant who burned his ex-girlfriend to death after pouring gasoline on her while she was lying in the front seat of her car; and State v. Mann,9 involving the aggravated rape and murder of an elderly woman. See Bane, 57 S.W.3d at 416 n. 2 (Birch, J., concurring and dissenting) (noting that the majority’s “notion of similarity appears to be highly malleable”). Given the variety and dissimilarity of cases relied on in the Bane proportionality analysis, the majority’s finding that the pending case is “plainly lacking in circumstances” similar to Torres seems difficult to justify. The majority’s analysis, it would seem, may have been guided more by subjective, result-oriented judgment than by an objectively measurable evaluation of the pool of comparison cases.
In sum, then, I would reform the Court’s review protocol as follows: First, in order to more reliably identify disproportionate sentences, I would ask whether the case under review was more consistent with “life” or “death” cases, rather than requiring that the case be “plainly lacking in circumstances” comparable to death penalty cases. Second, I would expand the pool of comparison cases to include all first degree murder cases, not just those cases in which the State chose to seek the death penalty, and I would revamp the Court’s Rule 12 database to ensure that it is complete, rehable, and accurate. Finally, I would more heavily emphasize objectivity in selecting which cases are “similar” to the case under review for the purposes of comparison. The soundness of the changes I propose may be illustrated through application to the case under submission. This case may not plainly lack circumstances similar to Torres, but its objectively measurable aspects certainly are more consistent with the numerous “life” cases cited by the majority. In addition to those cases, I would also include several “life” cases in which the State did not seek the death penalty, thus further reinforcing the conclusion that the death sentence in the case under submission is comparatively disproportionate.
My research and review of the current Rule 12 database of capital cases reveals a number of first degree murder cases involving child abuse which were excluded from comparison by the majority because the State opted not to seek the death penalty. In State v. Davis, No. 02C01-9511-CR-00343, 1997 WL 287646 (Tenn. Crim.App.1997), the 22 month old victim, who had been left with the defendant while the victim’s mother was at work, died of extreme injuries including organ contusions, broken ribs, broken blood vessels and cuts, bruises, and marks on the neck, face, and abdomen. The defendant originally claimed the victim was injured in a fall, but later admitted spanking and kicking the victim on multiple occasions. The defendant received a life sentence. In State v, Rhodes, No. M1999-959-CCA-R3-CD, 2000 WL 264327 (Tenn.Crim.App. 2000), the 18 month old victim was brought to the hospital unconscious and in cardiac arrest, with extensive bruising to the arms, legs, and face. The victim died of blunt force trauma to the head. The defendant later confessed to beating the victim with a switch and striking him in the head. The defendant received a life sentence. In State v. Anthony Hodges, 7 S.W.3d 609 (Tenn.Crim.App.1999),10 and State v. Kena Hodges, No. 01C01-9804-CR-00170, 1999 WL 618861 (Tenn.Crim.App.1999), the 2 *799year old victim was found dead in the Hodges’ residence, already in rigor mortis. The victim had died of blunt force trauma to the head and torso causing intracranial bleeding, swelling, and laceration of the liver. Though the victim was alone with Anthony Hodges for much of the day before her death, there was evidence to indicate that both parents had participated in beating and abusing the victim. Anthony Hodges received a sentence of life without the possibility of parole; Kena Hodges received a life sentence. In State v. Lyons, No. M1999-00249-CCA-R3-CD, 2000 WL 218131 (Tenn.Crim.App.2000), the defendant babysat the victim while the victim’s mother was at work. When the victim’s mother returned home, the victim appeared sick. The victim stopped breathing the next day and was transported to the hospital, where she died. An autopsy revealed extensive contusions and scrapes to the head and multiple tears of the mes-entery caused by blunt trauma to the abdomen. While left alone during a police interrogation, the defendant was videotaped talking to himself about punching the victim in the stomach. The defendant received a life sentence. In State v. Burgess, No. M1999-02040-CCA-R3-CD, 2001 WL 43216 (Tenn.Crim.App.2001), the 16 month old victim was briefly left unsupervised with the defendant by the victim’s mother. The mother subsequently discovered the victim was unresponsive, lethargic, and unable to sit or stand. The victim was taken to the hospital, where she died; the cause of death was identified as massive internal bleeding caused by blows of considerable force to the abdomen. The defendant later confessed to hitting the victim in the stomach. The defendant received a life sentence. In State v. Lacy, 983 S.W.2d 686 (Tenn.Crim.App.1997), the defendant beat and abused the 5-year-old victim over an extended period of time. The victim was found dead after being left alone for the day with the defendant. The cause of death was determined to be multiple blunt force injuries which caused severe internal bleeding; the victim also had evidence of burns, scratches, abrasions, scars, and fractures. One doctor testified that the victim had suffered the most severe blunt trauma injuries he had ever seen. The defendant received a sentence of life without the possibility of parole. In State v. Shephard, No. E2000-00628-CCA-R3-CD, 2001 WL 767010 (Tenn.Crim.App.2001),11 the victim began crying uncontrollably and showing signs of sickness after being left unsupervised with the defendant. The victim’s sickness grew progressively worse. The victim’s mother then left the victim unsupervised with the defendant again the following night, and at some point in the evening, the defendant appeared at a hospital with the victim, who was not breathing and had no pulse. The defendant claimed he “accidentally hit [the victim] on the back of the head with a little [toy] TV.” The victim died from a cerebral hemorrhage and extensive bilateral hemorrhage in the back layer of both eyes. The victim also suffered from two broken ribs caused by “enormous pressure of squeezing.” The defendant received a sentence of life without the possibility of parole.
When the facts of the case under submission are compared to the cases listed above and to those cases listed by the majority in which the defendant received a death sentence,12 it is clear that the State *800typically does not seek the death penalty, and juries typically do not impose the death penalty, in cases similar to Godsey’s case. Indeed, most of the “life” cases listed above exhibit significantly more egregious circumstances than the case before the Court. Thus, Godsey’s death sentence appears to be aberrant. Because Godsey’s case is more consistent with cases for which defendants typically receive a sentence of life or life without the possibility of parole, I concur in the majority’s decision to modify his sentence to life imprisonment.
In light of the issues I have discussed above, I continue to be dissatisfied with the comparative proportionality review protocol embraced by the majority. Until the flaws in our review protocol are corrected, I am constrained to hold that comparative proportionality review, as applied by the majority, “creates an impression of enormous regulatory effort but achieves negligible regulatory effects.” Cf. Carol S. Steiker and Jordan M. Steiker, Sober Second Thoughts: Reflections on Two Decades of Constitutional Regulation of Capital Punishment, 109 Harv. L.Rev. 355 (1995) (generally discussing capital punishment). Accordingly, I respectfully dissent.
. Moreover, one of the greatest sources of inconsistency in capital sentencing may arise at the prosecutorial level, where the State possesses almost unbridled discretion to choose which cases will be prosecuted as death penalty cases. For example, a cursory review of the Rule 12 reports filed in first degree murder cases seems to suggest that prosecutors in some counties frequently seek the death penalty in cases for which prosecutors of other counties do not seek the death penalty. While this conclusion is admittedly unscientific, there exists at least the possibility that a given defendant’s chances of receiving the death penalty may depend more upon the county where the crime occurred than upon the nature of the crime itself. The legislature has expressed its intent, in the Criminal Sentencing Reform Act of 1989, that defendants be fairly and consistently treated and that "unjustified disparity in sentencing” be eliminated. See Tenn.Code Ann. § 40-35-102(2), (3) (2000); see also id. § 40-35-103(3)(noting that "[(Inequalities in sentences that are unrelated to a purpose of [the Sentencing Reform Act] should be avoided”). Thus, under principles espoused in the Sentencing Reform Act, this Court should strive to ensure that inconsistencies in death sentencing such as regional disparities in prosecutorial decision-making are corrected. It is difficult, however, for this Court to squarely address whether the death penalty is being inconsistently applied if comparative proportionality review cannot take into account those cases in which prosecutors choose not to seek the death penalty.
. Majority op. at 785.
. This is not to say that the Court must engage in complex statistical analysis in order to conduct a reliable proportionality review. A statistical analysis likely would result in a significant increase in expense and complexity without appreciably increasing the overall reliability of the review. New Jersey, for example, has implemented a statistical protocol based upon a complex, subdivided database of capital cases, but the small sample size of the comparison pool and problems inherent in quantifying such abstractions as motive and type of murder have forced the Court to focus upon a precedent-seeking process similar to that used in Tennessee. See generally New Jersey v. Loftin, 157 N.J. 253, 724 A.2d 129, 148-52 (1999) ("Because frequency analysis is statistically based, and because conclusions drawn from small sample sizes are inherently unreliable, we have not had confidence in the results produced by the models.”).
. No. E1999-00866-CCA-R3-DD, 2001 WL 245137 (Tenn.Crim.App.2001). The death sentence imposed in Torres is presently before this Court on automatic appellate review pursuant to Tenn.Code Ann. § 39-13-206(a)(l) (2000). That review is currently pending. I note in passing that it may be problematic for this Court to rely upon Torres in its proportionality analysis when we have not yet decided whether the death sentence imposed in Torres is proportionate.
. The jury in Torres found the aggravating factors listed at Tenn.Code Ann. § 39-13-204(f)(1) (1993) (“victim was less than twelve years of age, and the defendant was eighteen years of age or older”) and Tenn.Code Ann. § 39-13-204(i)(5) ("especially heinous, atrocious or cruel in that [the killing] involved torture or serious physical abuse beyond that necessary to produce death”). Torres, 2001 WL 245137 at *1.
. 57 S.W.3d 411 (Tenn.2001).
. 976 S.W.2d 93 (Tenn.1998).
. 958 S.W.2d 679 (Tenn.1997).
. 959 S.W.2d 503 (Tenn.1997).
.This case is missing from the Rule 12 database.
. This case is missing from the Rule 12 database.
. Moreover, I question whether some of the "death” cases relied upon by the majority are sufficiently similar to merit comparison to the case under submission. For instance, when the objectively measurable details of State v. Middlebrooks and State v. Teel are examined, it becomes apparent that these cases bear *800little relationship to the circumstances of Godsey's case. Both of these cases involved teenage victims rather than infants, the jury in both cases found the "heinous, atrocious, or cruel” aggravating circumstance that was not found in this case, both cases involved brutal rapes, and neither case involved circumstances typically found in child abuse cases. As the majority correctly notes, no two cases are exactly alike, and therefore any one of these distinctions in isolation might not render a case "dissimilar.” The multitude and significance of the dissimilar factors in Mid-dhbrooks and Teel, however, are sufficient to make those cases inappropriate in light of the statutory requirement that proportionality analysis be restricted to "similar cases.” See Tenn.Code Ann. § 39-13-206(c)(l)(D) (2000).