with whom MATTHEWS, Justice, joins, dissenting.
I respectfully dissent from that part of the court's opinion that declines to subtract the amount received in the prior settlement from the damages award for this 1998 accident. The result the court reaches is contrary to the text of the statute that controls, AS 09.17.080(c) as it read in 1998. It is also contrary to our decisions that determine how we apply that statute and to the principle of compensation that has governed a negligent tortfeasor's IHfability in Alaska.
1. Introduction. Our disagreement arises because AS 09.17.080(c) continued to include the words "subject to a reduction under AS 09.16.040" after the 1988 initiative eliminated joint liability by amending AS 09.17.080(d). That initiative did not delete those words or otherwise amend subsection .080(c), even though it repealed AS 09.16.
The court's opinion concludes that former AS 09.17.080(c) does not apply here and that the "proportionate share rule" (or "pure several liability regime") embodied in AS 09.17.080(d) does.1 This conclusion is problematic for several reasons.
Because the two subsections coexisted side-by-side when this claim arose in 1998, the court must reconcile them-and the principles they embody-if it can. For reasons I discuss below, it can. The court should therefore apply AS 09.17.080(c) and subtract the settlement from the award.
A more subtle problem may underlie the court's opinion. It does not seem to hold that subsection .080(d) directly applies; it instead seems to resolve the reduction dispute by choosing to "adopt" the principle of pure several liability embodied in subsection .080(d) after the 1988 initiative became effective.2 Because the express words of former AS 09.17.080(c) apply here, the only justification for not applying them must be that a different subsection applies more directly or that the "subject to a reduction" language has been at least implicitly repealed or amended. The court does not clearly engage in the analysis needed to follow either approach. If subsection .080(c) applies, we do not have the option of choosing to "adopt," even by analogy to a principle expressed in a different statute, a different method of offsetting the prior settlement.
The facts are complex, but they can be simplified to this model: Two negligent tort-feasors caused a person to suffer personal injury in 1998. The plaintiff settled with one tortfeasor for $80,000 and proceeded to trial against the second. At trial the jury found *1033for the plaintiff, found plaintiff's compensatory damages to be $100,000, found the settling tortfeasor forty percent responsible for the accident, and found the nonsettling tortfeasor sixty percent responsible.3 These findings resulted in a jury verdict against the nonset-tling tortfeasor for sixty percent of $100,000. Were it not for the prior settlement, the liability of the nonsettling tortfeasor per AS 09.17.080(d) would have been $60,000.
The question here is whether the nonset-tling tortfeasor's liability may exceed the difference between the jury's award and the amount plaintiff received in settlement from the settling tortfeasor. The plain words of the controlling statute-former 'AS 09.17.080(c)-answer that question. The answer is "No." But per today's opinion the answer is "Yes." Invoking the principle of several-only liability and the 1988 amendment to AS 09.17.080, it holds that the award must be reduced "pro rata" (by the percentage of responsibility the jury allocated to the settling tortfeasor), and not "pro tanto"(by the amount of the prior settlement).
The opinion relies on principles and policy choices that were not reflected in the text of AS 09.17.080(c) as it existed in 1998. In my view, the text of that subsection as it then read and our pertinent case law preclude the choice the court makes here.
The court correctly observes that the ap-pellee has not cited subsection .080(e),4 but we must reach the issue anyway. The issue is inherent because we are being asked to apply subsection .080(d), and its application turns on the applicability of a co-existing statutory neighbor of which we are aware.5 Also, the appellee argues that subsection .080(d) does not apply, argues that the settlement must be subtracted from the award, invokes the tort principle (avoidance of double recovery) underlying subsection .080(c), and cites our offset decisions in support. At least one of those cases applies AS 09.16.040,6 the reduction statute to which subsection .080(c) expressly referred until 1997. We must therefore decide whether AS 09.17.080(c) applies.
2. The Applicable Statutory Scheme. Alaska Statute 09.17.080 concerns apportionment of damages. As of October 6, 1998, when the accident occurred, section .080 provided:
Apportionment of Damages. (a) In all actions involving fault of more than one party to the action, including third-party defendants and persons who have been released under AS 09.16.040, the court, unless otherwise agreed by all parties, shall instruct the jury to answer special interrogatories or, if there is no jury, shall make findings, indicating
(1) the amount of damages each claimant would be entitled to recover if contributory fault is disregarded; 'and
(2) the percentage of the total fault of * all of the parties to each claim that is allocated to each claimant, defendant, third-party defendant, and person who has been released from lability under AS 09.16.040.
(b) In determining the percentages of fault, the trier of fact shall consider both the nature of the conduct of each party at fault, and the extent of the causal relation between the conduct and the damages claimed. The trier of fact may determine that two or more persons are to be treated as a single party if their conduct was a cause of the damages claimed and the separate act or omission of each person cannot be distinguished.
(c) The court shall determine the award of damages to each claimant in accordance with the findings, subject to a reduction under AS 09.16.040, and enter judgment against each party liable. The court also ghall determine and state in the judgment each party's equitable share of the obli*1034gation to each claimant in accordance with the respective percentages of fault.
(d) The court shall enter judgment against each party liable on the basis of several liability in accordance with that party's percentage of fault.
(Emphasis added.)7 The emphasized passage contains the critical words. They provide that the award is "subject to a reduction under AS 09.16.040."
The version of AS 09.16.040 to which AS 09.17.080(c) referred provided:
Release or covenant not to sue. When a release or covenant not to sue or not to enforce judgment is given in good faith to one or two or more persons liable in tort for the same injury or the same wrongful death
(1) it does not discharge any of the other tortfeasors from liability for the injury or wrongful death unless its terms so provide; but it reduces the claim against the others to the extent of any amount stipulated by the release or the covenant, or in the amount of the consideration paid for it, whichever is the greater; and
(2) it discharges the tortfeasor to whom it is given from all liability for contribution to any other tortfeasor.
(Emphasis added.)8 The emphasized words called for a reduction (pro tanto) by the amount of the settlement. In 1997 the legislature amended section .080(c) by deleting this text: ", subject to a reduction under AS 09.16.040." 9 This is the text that requires the reduction in this case.
Unless context requires greater specificity, I refer to the version of section .080 set out above as AS 09.17.080 or "former" AS 09.17.080. Likewise, unless context requires otherwise, my references to AS 09.17.080(c) are to the version quoted above.
The court's opinion relies on the principles reflected in subsection .080(d). But subsection .080(c) contains the language that controls this case, which concerns the topic of reducing a plaintiff's recovery by the amount plaintiff received in a prior settlement. Subsection .080(c) expressly addresses that topic. Subsection .080(d) does not address it expressly or impliedly, and is irrelevant to the issue presented here.
There are several different ways to carry out the reduction required by subsection .080(c). Each requires that the amount of the prior settlement be subtracted (thus, reduced pro tanto) from the jury's award at trial, and each precludes the pure pro rata reduction approach chosen by this court.
The words of subsection .080(c), as it read in 1998, required the superior court to engage in a three-step reduction process. It was first to "determine the award of damages to each claimant in accordance with the findings." This award was then "subject to a reduction under AS 09.16.040." Finally, the court was to "enter judgment."
As applied to Robles, the words "the award of damages to each claimant in accordance with the findings" seem to mean the total damages found by the first jury as reduced by the percentage of fault found by the second jury. These collectively are the "findings" that determined the damages "award" (whether they were returned by one jury or two). Subsection .080(c) next required that the "award of damages" so caleu-lated be "subject to a reduction under AS 09.16.040." Under former AS 09.16.040, amounts paid in settlement reduced the claim against the other tortfeasors. After making this reduction, the court was required by AS 09.17.080(c) to enter judgment. Thus, subsection- .080(c) required the court to subtract *1035the amount of the prior settlement from the damages award before the court entered judgment against the trial defendant. In this case, that meant the trial award had to be reduced by the full amount Gannaway previously received in settlement, and that the superior court did not err in making that reduction.
Read most rigorously, the words of subsection .080(c) required that damages be reduced first by the percentage of responsibility allocated to anyone (including the settling tortfeasor) other than the trial defendant, and then by the amount of the prior settlement. Although this equld in some cireum-stances result in judgments that arguably under-compensated plaintiffs, this is the process the controlling statute specified.
Section .080 can alternatively be interpreted to call for a comparative reduction approach that respects the different principles underlying the two subsections. I read those subsections to reflect two different, but reconcilable, principles: avoidance of overcompensation or double recovery, achieved by requiring pro tanto (subtractive) reduction for amounts previously recovered in settlement (per subsection .080(c)); and avoidance of joint liability, achieved by requiring that the judgment against each party be based on several liability (per subsection .080(d)). The two subsections thus impose different limitations having different purposes. And because they impose limitations on the judgment, applying the subsection requiring the greater limitation in a given case does not conflict with the other subsection. The advantage of this approach is that it does not ignore statutory text (here, "subject to a reduction under AS 09.16.040") and considers the effect of both tort reform limitations.
The best way to reconcile the two provisions is to apply them comparatively in cases to which they both apply. Thus, a pro rata reduction would occur if it were needed to satisfy the several-liability-only policy of subsection .080(d). And a pro tanto reduction for a prior settlement would occur if it were needed to satisfy the requirement of subsection .080(c) that makes an "award of damages to each claimant ... subject to a reduction under AS 09.16.040." To ensure that the judgment as entered only reflected each defendant's several liability and thus satisfied subsection .080(d), a trial court following the comparative reduction approach would have compared the result of the pro tanto reduction with the result of a pro rata reduction. A pro tanto reduction required by subsection .080(c) insufficient to reduce the nonsettling defendant's liability below its proportionate share of the total damages would have the effect of imposing joint liability on the non-settling defendant, contrary to subsection .080(d). To avoid that result, the final judgment in that situation would have to be based on the pro rata calculation. This calculation would also avoid overcompensation, because the judgment would necessarily be no greater than the amount calculated under subsection .080(c). Similarly, if a. judgment based on pure several Hability nonetheless exceeded the amount calculated after pro tanto reduction for a prior settlement, calculating the judgment per subsection .080(c) would best reconcile the demands of both subsections.
In effect, the two subsections can be reconciled by applying them in a way that carries out the policies that underlie them.
A hypothetical example illustrates this approach: plaintiff settles with one of two joint tortfeasors for $80,000. A jury finds plaintiff suffered damages totaling $100,000 and finds each tortfeasor fifty percent responsible. The approach first discussed above requires that the $80,000 settlement be subtracted from the nonsettling tortfeasor's proportionate share of the damages, resulting in no recovery for the plaintiff.10 Under the comparative approach next discussed, the $80,000 settlement amount is subtracted from the $100,000 total damages finding. Because the nonsettling tortfeasor's resulting post-offset $20,000 liability is less than that tortfeasor's $50,000 proportionate share of the total damages, the purpose underlying subsection *1036.080(d)-protecting the tortfeasor from paying for harm caused by other tortfeasors-is fulfilled and no further reduction is required. The final recovery is $20,000.11
And if the settling tortfeasor instead paid $20,000, a pro tanto reduction from the total amount of damages would only reduce the damages award to $80,000. Because subsection .080(d) prohibits courts from requiring tortfeasors to pay more than their proportionate share of damages, the nonsettling tortfeasor's liability cannot exceed $50,000.12
This comparative approach trades the statute's plainer reading first discussed above for a process that respects both reductive principles reflected in the two subsections, and leads to an arguably more equitable reduction formula. It is also the method the superior court applied in Navistar International Transp. Corp. v. Pleasant.13
But either method of calculation I discuss has the same effect on today's case: the claimant recovers nothing. And either method requires a pro tanto reduction. The problem with the court's opinion is not that it adopts the incorrect method for applying the reduction required by subsection .080(c)'s reference to AS 09.16.040, but that it declines to apply subsection .080(c) at all.
Subsection .080(c)'s continued reference to AS 09.16.040 until 1997 might seem problematic at first glance; AS 09.16.040, after all, was repealed in 1988.14 But we are not free to ignore the express words of AS 09.17.080(c) on a possible theory they had no meaning. We dealt with a similar interpretive problem involving AS 09.16.040 and AS 09.17.080 in Benner v. Wichman.15 That case required us to interpret a reference in AS 09.17.080(a) to AS 09.16.040 after section 040 (and, indeed, all of Chapter 16) had been repealed. We there looked to the repealed statute in attempting to carry out the drafters' intentions because we regarded the repealed statute as "integral" to understanding the disputed language in AS 09.17.080(a).16
Today's opinion advances several reasons for declining to apply subsection .080(c). Its primary reason seems to be that the "subject to a reduction under AS 09.16.040" clause "simply means" that a judgment is subject to reduction under that statute "provided that the statute applies." 17 And, per the court, AS 09.16.040 does not apply because it was repealed before the settlement was reached.18
The court is looking at the wrong statute. Section .040 was repealed in 1988, but the critical text of subsection .080(c) was not. We cannot simply ignore the words of subsection .080(c). Until 1997 they provided that the award is "subject to a reduction under AS 09.16.040."
The court's opinion reads subsection .080(c) as though it contains qualifying words such as "provided that the statute applies." It does not, of course. But if it had, it would still apply here, because it addresses the topic in issue here, reduction for a prior settlement. The court's opinion also reads the statute as though it merely states that an award is "subject to AS 09.16.040." But it *1037actually states that the award is "subject to a reduction" under AS 09.16.040. (Emphasis added.) The difference is significant.19
The court's opinion advances another theoretical explanation for the existence of the "subject to" clause until 1997: The clause usefully protected contractual expectations arising out of settlements reached before the 1988 initiative became effective, until it was clear by 1997 that the clause was vestigial and could be repealed.20 Given that imputed purpose, the court seems to reason that there is no need for reduction for post-1988 settlements.
There is no basis for thinking that initiative adopters or the legislature, by retaining the "subject to a reduction" language for about nine years after voters approved the initiative, intended to protect contract expectations. No legislative history supports that notion, and reality seems to rebut it. The contracting parties in a tort settlement are the tort plaintiff and the settling tortfeasor. The latter has no interest in how section .080 is applied and the former derives no protecti-ble contractual benefit as against nonsettling tortfeasors (who are not party to the contract). No relevant expectation benefitting the plaintiff arises from the contract itself, (Even if a statute might give rise to expectations, the continued existence of the "subject to a reduction" clause would have rendered unreasonable any expectation that settlements would no longer reduce trial awards in Alaska after 1988.) Settlement incidentally benefitted the nonsettling tortfeasor, who had the statutory right to a reduction for the amount of the settlement. But that right derived from the fact of payment, not the contract itself. So it is hard to imagine that the reduction language was retained for the limited purpose of protecting contractual rights. -It is therefore equally implausible to think that the date of settlement is relevant to the statutory analysis required here. The relevant date is the date when the cause of action accrued. By accruing before 1997, this cause of action is governed by the "subject to a reduction" clause.
The court's opinion, in alternatively, suggesting that subsection .080(c)'s reference to the repealed AS 09.16.040 may have been a "drafting mistake," 21 may assume that the 1988 initiative was intended to repeal the reduction language in subsection .080(c). There was no such express repeal and there was no expression of any such intention.22 Or the opinion may assume a degree of inconsistency between subsection .080(c) and subsection .080(d) that can be resolved only by holding that there was an implied repeal. Any such assumption would be problematic.
Because we can reconcile the text and the effect of subsections .080(c) and (d), we cannot assume the implied repeal of the pro tanto reduction language in subsection .080(c). Reconciliation turns on recognition that the two subsections have different purposes and that they foeus on different aspects of tort damages. Subsection .080(d), read literally, focuses on the parties who are being held liable at trial. It does so by addressing judgment "against each party liable" and by adopting several-only liability, fundamentally limiting defendants' liability. In comparison, the first two clauses of subsection .080(c) focus on what claimants may recover. Thus, those clauses address the award "to each claimant" as calculated "in accordance with the findings," and therefore take into account any fault of the plaintiff as well as the fault of the tortfeasors. The two subsections can be reconciled by reading subsection .080(c) to limit what a plaintiff can recover (because it requires a reduction of the plaintiff's award per AS 09.16.040) and by reading subsection .080(d) to limit what a *1038defendant can be made to pay (because entering judgment "on the basis" of several liability under that subsection does not nee-essarily foreclose any other limitation, such as a reduction for amounts previously received). The two subsections can be read together to limit the amount a plaintiff can recover from the defendants at trial, both by limiting a plaintiff's total compensation and by limiting a defendant's individual exposure. The two subsections can be most clearly reconciled by applying the comparative reduction methodology discussed above on pages 7-10.
If we ignore this method of reconciling the two subsections, they can be read to conflict, on a theory the pro tanto reduction subsection .080(c) required was inconsistent with the pro rata reduction subsection .080(d) seemed to require. But this conflict is only implied, and it is not obvious how it should be resolved. To resolve it as the court's opinion implicitly seems to do (by declining to apply subsection .080(c)) reads words out of the statute, and renders them not merely innoeu-ously superfluous, but actively firreconcilable.23 This requires us to ignore specific statutory text dealing with a specific topic (subtractive reduction for amounts previously received) in favor of general language that unspecifically calls for entry of judgment "on the basis of" several liability.24 It also raises questions about the effect of the 1997 express repeal.
It is theoretically possible to surmise that an intention to adopt several-only liability cannot be reconciled with the subtractive reduction subsection .080(c) requires. But the only (or main) perceived evil being addressed in 1988 was any form of joint lability, i.e., any principle by which a defendant at trial could be required to pay more than its pro rata share of the damages. Avoiding joint liability does not necessarily conflict with a reduction for prior settlements, nor does the methodology conflict if the comparative approach is applied.
I am not prepared to hold that the 1988 amendment somehow repealed the critical words of subsection .080(c). The court's opinion may reason that what it calls "pure" several liability 25 somehow nullifies subsection .080(e)'s reduction language. To the extent the opinion's analysis turns on its characterization of the amendment's purpose, I am unconvinced. The 1988 amendment had the purpose and effect of repealing any remaining possibility that defendants would be exposed to joint liability.26 Ending defendants' exposure to joint liability is not inherently and thematically at odds with reducing a plaintiff's award by the amount of a prior settlement or recovery. Thinking that pure several liability had a purpose of benefitting injured plaintiffs ignores the history of tort reform in Alaska. Several-only lability was adopted in 1988 to prevent modified joint liability; 27 plaintiffs' loss of an opportunity to hold defendants jointly Hable was regarded as detrimental to plaintiffs and beneficial to defendants.28 Throughout its history, Alaska tort reform has focused on limiting defendants' liability and reducing the cost of *1039liability insurance.29 Supporters of tort reform memorialized no concern about whether tort defendants at trial might pay less than their share of damages or that trial awards might under-compensate tort plaintiffs.30
In any event, several-only liability is not necessarily at odds with a policy choice of requiring reductions per the language of subsection .080(c). The concept of several liability does not preclude a reduction for prior settlements. Likewise, pure several liability was adopted because even modified joint liability was considered undesirable.31 Joint and several liability (and, later, several and modified joint liability) were features of Alaska tort law for many years.32 Defendants' exposure to joint liability provoked several statutory amendments. Consequently, as of 1988 (and thus as of October 6, 1993) a drafter could have both rationally intended that subsection .080(d) would simply abolish joint liability and logically reasoned that several liability would determine the maximum amount (not the final amount) of a judgment against a defendant. A drafter likewise could have reasoned that subsection .080(c), by requiring the specified reduction before entry of judgment, was not inconsistent with foreclosing judgments based on joint liability. The history of tort reform in Alaska reflects a series of partial measures addressing a few aspects of tort law at a time. The 1988 interest in addressing one perceived problem-joint liability-does not prove an interest in addressing all other possible problems, as well.
Moreover, reading subsection .080(c) to require a subtraction of the amount re*1040ceived notwithstanding subsection .080(d) is consistent with our decisions interpreting analogous statutes. The superior court here applied what we said about reduction in Universal Motors, Inc. v. Neary (which arose under the same statutory regime).33 As the next part of my dissent discusses, subtractive reduction for prior tort settlements was long a feature of Alaska law. A drafter aware of our decisional law could have rationally intended to retain the reduction feature expressed in former subsection .080(c).
Indeed, as of 1998 (and until 1997), AS 09.17.080 also contained two other references to AS 09.16.040 notwithstanding that seetion's repeal in 1988.34 Those other two references and our treatment of one of them in 1994 argue against any implied repeal.35
In short, we are not free to disregard the statutory reduction language.
3. Our Precedent. Despite periodic statutory changes, our decisions discussing the applicable statutory scheme and its predecessors consistently recognized that an award against a negligent tortfeasor had to be reduced by amounts the plaintiff previously received as damages from other tortfeasors. This is so whether the prior recovery was by settlement or judicial award.
Navistar International Transp. Corp. v. Pleasant illustrated how the statutory reduction was to work.36 It confirms that in Alaska, at least before 1997, a tort award had to be reduced at least by the amount of a prior settlement. Pleasant suffered personal injury in July 1986 and sued several alleged tortfeasors.37 After the jury returned a verdict for the plaintiff against some of the tortfeasors but before judgment was entered, the plaintiff settled with one of the defendants for a large amount.38 The settlement agreement allocated some of the settlement to punitive damages and the remainder to compensatory damages.39 To calculate the judgment against the nonsettling tortfeasor, the superior court applied the comparative reduction approach I describe in Part 1. Thus, it compared the result of a pure pro tanto reduction per AS 09.17.080(c) with the result of a proportionate share calculation under the modified joint and several liability regime per AS 09.17.080(d) as it read when Pleasant was injured.40 Because it found that the modified joint and several Hability result was less than the pro tanto result,41 it *1041entered judgment based on the former result.42 We reversed because the superior court, in reducing pro tanto, did not subtract the part of the settlement amount allocated to punitive damages; it therefore incorrectly concluded that the pro tanto reduction result was larger than the modified proportionate share result and therefore incorrectly entered judgment based on the latter method.43 We confirmed that it was appropriate to reduce pro tanto; indeed, we reversed because the pro tanto reduction did not subtract the full amount of the prior settlement.44 We raised no question about the superior court's use of the comparative reduction approach, and tacitly seemed to approve it. Nothing about the amendment to subsection .080(d) effective in 1989 renders the comparative reduction approach less appropriate or justifies complete failure to carry out the subsection .080(c) reduction approved in Navistar.
In Norcon, Inc. v. Kotowski,45 Kotowski, claiming she had suffered sexual harassment in June and July, 1989, sued Norcon and Exxon for, among other things, compensatory damages for emotional distress and lost earnings.46 She settled with Exxon before trial for $20,000.47 The trial jury then awarded her compensatory damages of about $10,000 against Norcon,48 which argued on appeal that the amount of the Exxon settlement had to be subtracted from the jury award to avoid a double recovery.49 We agreed, citing Navistar. 50
In Neary, we considered whether the "one-action rule" barred Neary's second personal injury lawsuit for a 1994 accident.51 We held that the statutory regime then in effect (the same regime that applies here) did not mandate a single action for each injury or accident.52 We also reasoned that it was not necessary to adopt a one-action rule to prevent inconsistent results or double recoveries.53 We so concluded because we recognized that the damages award at a second trial had to be offset by what a plaintiff had already received at a first trial.54 In support, we cited Norcon and Navistar.55
In Bohna v. Hughes, Thorsness, Gants, Powell & Brundin, we held that "consideration paid" in AS 09.16.040(1) included amounts paid per a loan receipt even if plaintiff had to repay the amount so received.56 " '[Alnything received by way of a covenant not to sue operates as a payment pro tanto upon any judgment obtained against the others'" 57 We recognized that the reduction was pro tanto.58
In Chenega Corp. v. Exxon Corp., we again recognized that non-collateral source payments received by a plaintiff must be deducted from a jury award against a tortfeasor.59
*1042The court's opinion today distinguishes these opinions on the theory they turn on statutes that predated the "pure several liability regime enacted by initiative in 1988." 60 But the 1988 initiative did not repeal the language in AS 09.17.080(c) that required a reduction in accordance with AS 09.16.040.
Our decisions therefore confirm that the applicable statutes require a pro tanto reduction, at least for purposes of applying the comparative reduction approach that the superior court used and that we tacitly approved in Navistar.61 Because the pro tanto statutory reduction requirement recognized in those cases cannot be reconciled with the pro rata reduction approach adopted here, stare decisis forecloses the result the court's opinion would reach unless the court addresses and overrules this precedent.
The court's opinion does not discuss or apply the analytical standard that we use in deciding whether to overrule our past precedent. But it is hard to see how that standard could be satisfied here.62 There is no basis for thinking our past decisions were wrong when we issued them; they appropriately relied on AS 09.17.080(c) as it read until 1997. The policy choice between pro tanto and pro rata reduction is' not so one-sided that we can now say that our decisions applying a pro tanto approach were wrong when decided. After we issued those opinions no conditions changed relative to when this cause of action accrued in 1993. It likewise cannot be said that more good than harm would result from departing from our precedent: The statutory issue now before us seems unlikely to arise again, given that the 1997 amendment has now been in effect for nine years. It seems unlikely the issue will recur. There is no pressing reason to reinterpret statutory text that has not existed since 1997. The harshness the court's opinion perceives in applying a pro tanto reduction cannot justify overruling precedent that originates in the policy choices the statute reflects. But to the extent pro tanto reduction may lead to a harsh result, harshness is avoided by reading the statute to provide for the comparative reduction approach discussed in Part 2, above, and applied in Nav-istar. The avoidable possibility of a harsh result cannot justify overruling our prior decisions.
4. Tort Principles of Compensatory Damages. Several value judgments seem to underlie the court's analysis. The opinion seems to reason that a "central purpose" of the 1988 initiative was to ensure that tortfea-sors are held accountable for the harm they cause.63 It perceives signs of this policy in expressions of an intention to make a tortfea-sor pay for the harm he or she has caused.64 It concludes that the potential for overcompensation is justified by "the reciprocal risk of settling with a defendant for less than that defendant's comparative share of the harm." 65 Its conclusion therefore turns on the weight it gives the policies it identifies.
I discuss public policy only reluctantly here. In my view the express words of AS 09.17.080(c) before amendment in 1997 both specify what courts must do in this case and *1043prevent courts from invoking public policy to make a policy choice different than that reflected in the statute's words.66 I discuss public policy here only to demonstrate that for claims that arose in 1998, the reduction the statute required until 1997 was consistent with a valid and important principle of tort law that had long been recognized and applied in Alaskas, and that was not foreclosed by any public policy identified by the court today.
The policy choice underlying the court's analysis today confuses the pertinent policies and misconceives tort law in context of claims by victims of negligence, at least as to claims accruing before 1997. Thus, it is a controlling and fundamental principle of tort law, notwithstanding its Dickensian flavor of unreality, that negligently inflicted personal injuries can be fully compensated with money.67 Money can make the personal injury plaintiff legally whole, whether or not medical science can. This means that the harm to the plaintiff is measured in dollars. And it means that when the plaintiff has received dollars equal to the monetary value of the harm suffered, the plaintiff is no longer suffering harm.68 He is "cured." And if he receives dollars equal to half the monetary value of his harm, he is "half-cured." This means that his harm, in the eyes of the law, is only half what it was before he was. paid. Thus, speaking hypothetically, if the plaintiff's harm is valued at $400,000 and he receives $100,000 from one tortfeasor in a settlement, by definition only $800,000 of harm remains uncompensated. Things are more complicated in dealing with a settlement and then a subsequent trial at which a jury finds the amount of damages, but the complication is conceptually unimportant. If the jury finds that the plaintiff suffered damages worth $400,000, his receipt of the prior $100,000 settlement proceeds means that the post-settlement value of plaintiffs harm is only $300,000.
This means that applying a policy of requiring a tortfeasor to pay for the harm he has "caused" necessarily raised this question (before 1997); what harm has the plaintiff suffered after he has been made partly whole? In the above hypothetical the uncompensated harm is $300,000, not $400,000. Public policy demands that the trial defendant pay for the harm attributable to him; 69 it does not demand that he pay for harm that has already been remedied, and no legislative history or tort reform history applicable to this case supports a conclusion to the contrary.70 Policy does not require him to pay *1044for harm that is already cured by partial settlement with another tortfeasor.71 After the settlement, the second tortfeasor is no longer responsible for the first $100,000 worth of harm; as to that harm, plaintiff has already been fully compensated.72 Because dollars define the harm, it is the uncompensated harm, and not the entire harm, for which the wrongdoer must pay. That is as far as the public policy-of making the tort-feasor pay-goes.73
The second problem is that the draft's perception of a broader public policy is almost certainly incorrect given the improbability that any tort reforms from 1986 onward had any purpose of requiring tort-feasors to pay plaintiffs more than the amount of their uncompensated damages.
Public policy in Alaska has long required a setoff (or reduction) for prior settlements. In Luth v. Rogers & Babler Construction Co., we stated that common law principles prohibit double recovery:
A cardinal common law principle establishes that in the absence of punitive damages a plaintiff can recover no more than the loss actually suffered. The doctrine prohibiting double recovery supports the rule that a payment received by the plaintiff for a covenant not to sue someone potentially liable in tort must be deducted from the damages recoverable from persons whose tort liabilities arise out of the same cireumstances.[74]
Alaska Statute 09.16.040, adopted after Luth's accident, embodied the same rule. Nothing in the history of chapter 17 suggests that its proponents intended to make it possible for personal injury plaintiffs to recover more money than it took to make them whole. Instead, it would appear that the legislature intended to reduce the possibility of overcompensation when, in 1986, it broadened the categories of collateral benefits that require that a judgment be reduced.75 Chapter 17 reflects a policy choice of preventing plaintiffs from recovering from tortfeasors any unsubrogated health benefits the plaintiffs have received from the plaintiffs' own insurance companies for the same injuries.76 Nothing that happened in connection with the 1988 initiative suggests a more plaintiff-oriented policy.
In short, the subtractive reduction that former subsection .080(c) requires in this case is consistent with public policy. The broad public policies discussed in the court's opinion do not permit us to disregard the words of the statute.
5. Conclusion. I therefore conclude that we should affirm the reduction made by the superior court.
. Op. at 1020, 1021, 1023.
. Op. at 1019.
. In Robles's case, two different juries returned the corresponding findings; the first determined the amount of damages and the second allocated fault. This circumstance is irrelevant to the issue presented.
. Op. at 1023.
. More accurately, it appears that the appellant asks us to apply the 1988 initiative that amended AS 09.17.080(d) and adopted pure several liability.
. Bohna v. Hughes, Thorsness, Gantz, Powell & Brundin, 828 P.2d 745, 758 (Alaska 1992).
. As first enacted in 1986, subsection .080(c) referred to "AS 09.17.090." Ch. 139, § 1, SLA 1986. The legislature substituted "AS 09.16.040" in 1987. Ch. 14, § 16, SLA 1987.
. The 1988 initiative repealed chapter 16, including AS 09.16.040, but did not delete the three references to section .040 in AS 09.17.080(a) and (c). 1987 Initiative Proposal No. 2, adopted in 1988. The initiative's amendments became effective March 5, 1989.
. Ch. 26, § 13, SLA 1997. The legislature also amended subsections .080(a) and (b), but not subsection .080(d). Ch. 26, §§ 11-12, SLA 1997. The 1997 changes applied to causes of action accruing on or after August 7, 1997. Ch. 26, § 55, SLA 1997. The text of subsection .080(c) deleted in 1997 therefore applies to this case because the cause of action accrued in 1993.
. The math is as follows: ($100,000 x 0.50) $80,000 = -$30,000. Because no negative award is possible here, there is no recovery.
. The math for the comparative approach requires dual calculations, The first calculates the maximum amount that the nonsettling tortfeasor should be required to pay as follows: $100,000 x .50 = $50,000. The second calculates the maximum amount that the plaintiff may receive without receiving a partial double recovery as follows: $100,000-$80,000 = $20,000. Taking the lower of the dual calculations will ensure that both policies-the avoidance of joint liability and the avoidance of a double recovery-are satisfied.
. The dual calculations here would be as follows. The first would be: $100,000 x .50 = $50,000. The second would be: $100,000-$20,000 = $80,000. Again, selecting the lower of the two results will ensure that the nonsettling tortfeasor pays only for the damages he has caused while also ensuring that the plaintiff does not receive a double recovery.
. Navistar Int'l Transp. Corp. v. Pleasant, 887 P.2d 951, 956 (Alaska 1994).
. The voters approved the 1987 Initiative Proposal No. 2 repealing chapter 16 in 1988.
. Benner v. Wichman, 874 P.2d 949 (Alaska 1994).
. Id. at 957 n. 18 (interpreting AS 09.17.080(a)'s reference to AS 09.16.040).
. Op. at 1023.
. Id.
. The clause is not a mere redundancy, unlike the generic phrase ("subject to the provisions of this Act') condemned by Brian Garner. Bryan A. Garner, A Dictionary or MopEen Lzoar UsacE at 840 (Oxford University Press 1995). The clause here is specific: it specifies the topic and the required action (reduction) and it also provides a specific statutory cross-reference.
. Op. at 1024.
. Op. at 1024.
. 1987 Initiative Proposal No. 2, The initiative amended AS 09.17.080(d), repealed AS 09.16, and made these changes applicable to all causes of action accruing after the effective date of the act. Id. The initiative did not amend, or mention, AS 09.17.080(c).
. See In re Adoption of Missy M., 133 P.3d 645, 650 (Alaska 2006) (holding that effect should be given to all words of statute so that none are rendered superfluous); Mech. Contractors of Alaska, Inc. v. State, Dep't of Pub. Safety, 91 P.3d 240, 248 (Alaska 2004) (presuming " 'that the legislature intended every word, sentence, or provision of a statute to have some purpose, force, and effect, and that no words or provisions are superfluous'" (quoting Kodiak Island Borough v. Exxon Corp., 991 P.2d 757, 761 (Alaska 1999))).
. See City of Cordova v. Medicaid Rate Comm'n, Dep't of Health & Soc. Servs., State, 789 P.2d 346, 352 (Alaska 1990) ("It is a maxim of construction that specific statutes should be given precedence over more general ones.").
. Op. at 1019, 1020.
. See McLaughlin v. Lougee, 137 P.3d 267, 277 (Alaska 2006) ("The scant legislative history demonstrates that the initiative's supporters were concerned almost exclusively with ensuring that defendants would be liable only for their share of fault." (quoting Alaska Gen. Alarm, Inc. v. Grinnell, 1 P.3d 98, 106 (Alaska 2000))).
. See Election Pamphlet for 1987 Initiative Proposal No. 2 (87TOR2) (statement in support).
. See id. (statement in opposition).
. See House Floor Debate on SB 337 (May 8, 1986) (statement of Rep. Pearce) (noting that "the real victims that I think we're all here working for are the people who can't get insurance because their rates have gone out of sight" and that "society has gone overboard for the victims"), transcribed in Respondent's Brief, FMC Corp. v. Craig Taylor Equip. Co., No. S-2261 (Alaska 1988), App. B, at 24-25; id. at 57 (statement of Rep. Pignalberi) (rebutting criticism that tort reform proposal would have no effect on insurance rates); id. at 109 (statement of Rep. Miller) (arguing that anything less than pure several liability is unfair to defendants); id. at 111 (statement of Rep. Boucher) (arguing that tort reform is necessary because of "general greed of our society"); id. at 116 (statement of Rep. Hanley) (noting that "I think there are members among us who, perhaps, even think that the simple several liability ... is ... a fair approach"); Senate Floor Debate on SB 337 (May 5, 1986) (statement of Sen. Faiks) (noting that Director of Division of Insurance believed that limitations on joint and several liability would allow division "to gain rate reductions, new markets and increased capacity"), transcribed in Respondent's Brief, FMC Corp. v. Craig Taylor Equip. Co., No. S-2261 (Alaska 1988), App. A, at 33; id. at 67 (statement of Sen. Kelly) (arguing that the bill would lower insurance premiums and increase insurance availability); see also Letter from Governor Bill Sheffield to Sens. Faiks & Sackett, Co-Chairs, Senate Finance Committee and Rep. Miller, Chairman, House Judiciary Committee, at 1 (Apr. 28, 1986) (noting that "the real goal of tort reform is affordable insurance'"); Election Pamphlet for 1987 Initiative Proposal No. 2 (87TOR2) (statement in support) (arguing that joint and several liability is "unfair" because it "forces people to pay for damages caused by somebody else"); id. (statement in opposition) (arguing that if several lability is enacted, "the insurance company wins and the victim loses"); cf. ch. 26, § 1, SLA 1997 (explaining that purposes of 1997 tort reform act included "discouraging. frivolous litigation," "al-leviat[ing] the high cost of malpractice insurance premiums," and "fostering an environment likely to control the increase of liability insurance rates").
. Some legislators reluctantly supported the 1986 tort reform bill that introduced partial, or "modified," joint liability, because this represented a compromise seen as less harmful to plaintiffs than pure several liability ultimately adopted in 1988. See Senate Floor Debate, supra note 29, at 34 (statement of Sen. Josephson) (explaining that "many of us who would prefer to continue to have the doctrines of tort law developed in the courts arising out of concrete fact situations with advocacy on both sides are yielding, in a sense, our better judgment and saying we're willing to legislate because there js that crisis or the perception of a crisis"); id. at 25 (statement of Sen. Rodey) (describing his proposal for modified joint and several liability as a "middle ground"). One legislator reasoned that a benefit of tort reform would be that more potential tort-feasors could afford liability insurance, and thus might increase the chances tortfeasors would be able to respond to tort claims. See id. at 67-68 (statement of Sen. Eliason).
. See Election Pamphlet for 1987 Initiative Proposal No. 2 (87TOR2) (statement in support).
. See ch. 80, § 1, SLA 1970 (codified as amended at AS 09.16.010-.060) (codifying joint and several liability); ch. 139, § 11, SLA 1986 (establishing several and modified joint liability); see also Kodiak Island Borough v. Roe, 63 P.3d 1009, 1013 (Alaska 2003) (explaining history of joint and several liability in Alaska).
. Universal Motors, Inc. v. Neary, 984 P.2d 515 (Alaska 1999).
. See former AS 09.17.080(a) and .080(a)(2) (1993) (both referring to parties and persons released from liability "under AS 09.16.040").
. See Benner, 874 P.2d at 958 n. 18 (consulting repealed AS 09.16.040 in interpreting AS 09.17.080(a)).
. Navistar Int'l Transp. Corp. v. Pleasant, 887 P.2d 951 (Alaska 1994).
. Id. at 953-54. When Pleasant was injured, Alaska had a system of modified joint and several liability, which capped joint liability against a tortfeasor less than fifty percent at fault at twice its percentage of fault. See former AS 09.17.080(d) (1987).
. Navistar, 887 P.2d at 954.
. Id.
. Id. at 956. The specific facts of Navistar are complex because of the number of alleged tort-feasors. Here is how the superior court calculated the judgment. It first subtracted the portion of the settlement allocated to compensatory damages ($1,525,046) from the jury's total damages verdict and prejudgment interest ($3,431, 459.61). Id. The court then further reduced the award by the amount of a settlement with a second tortfeasor ($200,000) and plaintiff's ten percent comparative negligence ($342,510.10) for a total of $1,357,544.94. Id. The court compared this total to the amount called for by the modified joint and several liability system in force at the time. Id. For the modified joint and several liability calculation, the court deducted the enhanced prejudgment interest on the two settlements from $3,431,459.61 and multiplied this amount by twice the proportion of Navistar's fifteen percent share of the fault (15 x 2 = .30) for a total of $1,027,530.31. Id. Because this modified pro rata approach yielded an amount less than the $1,357,544.94 called for by the pure pro tanto approach, the court entered judgment for the smaller amount. Id. Had the superior court subtracted the entire $2.1 million in making the first calculation, the result would have been lower than $1,027,530.31 and the superior court presumably would have entered judgment for that lower amount.
. Id. It "tested" the pro rata reduction result against the pro tanto reduction result and based the judgment on the lower of the two. Id. It thus applied both AS 09.17.080(c) and AS 09.17.080(d) (as i then read) by reading each as *1041setting the remaining defendant's maximum liability.
. Id.
. Id. at 958.
. Id. Because the superior court also failed to make a second, unrelated reduction, the result of our holding was that the defendant owed nothing to the plaintiff. Id.
. Norcon, Inc. v. Kotowski, 971 P.2d 158 (Alaska 1999).
. Id. at 163, 171.
. Id. at 171.
. Id. at 163.
. Id. at 171.
. Id.
. Universal Motors, Inc. v. Neary, 984 P.2d 515, 516 (Alaska 1999).
. Id. at 516.
. Id. at 518.
. Id.
. Id. at n. 13.
. Bohna v. Hughes, Thorsness, Gantz, Powell & Brundin, 828 P.2d 745, 758-59 (Alaska 1992).
. Id. at 758 (quoting Cullen v. Atchinson, Topeka & Santa Fe Ry. Co., 211 Kan. 368, 507 P.2d 353, 362 (1973) (alteration in Bohna )).
. Id. at 758-59.
. Chenega Corp. v. Exxon Corp., 991 P.2d 769, 790 (Alaska 1999) ("[Playments made to the victim from non-collateral sources-such as the tortfeasor's insurance company or a joint tortfea-sor-reduce the tortfeasor's liability.").
. Op. at 1020. The regime was enacted in 1988 when voters approved the 1987 initiative proposal; the regime became effective in 1989. See 1987 Initiative Proposal No. 2.
. Navistar, 887 P.2d at 956.
. We overrule our decisions only when "clearly convinced that the rule was originally erroneous or is no longer sound because of changed conditions, and that more good than harm would result from a departure from precedent." State v. Fremgen, 914 P.2d 1244, 1245 (Alaska 1996) (quoting State v. Dunlop, 721 P.2d 604, 610 (Alaska 1986) (internal quotation marks omitted)); see also Thomas v. Anchorage Equal Rights Comm'n, 102 P.3d 937, 943 (Alaska 2004) ("The stare decisis doctrine rests on a solid bedrock of practicality: no judicial system could do society's work if it eyed each issue afresh in every case that raised it.") (quoting Pratt & Whitney Canada, Inc. v. Sheehan, 852 P.2d 1173, 1175 (Alaska 1993) (internal quotation marks omitted)). "In recognizing the importance of this doctrine, we have consistently held that a party raising a claim controlled by an existing decision bears a heavy threshold burden of showing compelling reasons for reconsidering the prior ruling." Id. at 943.
. Op. at 1019.
. Op. at 1019 (citing Election Pamphlet for 1987 Initiative Proposal No. 2 (87TOR2) (statement in support)).
. Op. at 1021.
. The history of the initiative, including the explanation on the ballot, is not so clear on the reduction topic as to nullify the words of AS 09.17.080(c). .
. See McDougald v. Garber, 73 NY.2d 246, 538 N.Y.S.2d 937, 536 N.E.2d 372, 374-75 (1989) (noting that recovery of noneconomic damages "rests on the legal fiction that money damages can compensate for a victim's injury. We accept this fiction, knowing that although money will neither ease the pain nor restore the victim's abilities, this device is as close as the law can come in its effort to right the wrong." (internal quotation marks omitted)).
. See Livingstone v. Rawyards Coal Co. (1880) 5 App. Cas. 25, 39 (U.K.) (famously defining the measure of damages as "that sum of money which will put the party who has been injured, or who has suffered, in the same position as he would have been in if he had not sustained the wrong for which he is now getting his compensation or reparation"), quoted in Harvey McGrEcor, McGrEcor on Damages 10 (14th ed.1980).
. See W. Pack Keeton, et al , Prosser and KeETON on Torts § 1, at 6 (5th ed. 1984) (''The purpose of the law of torts is to adjust [losses or injuries sustained as a result of the activities of others], and to afford compensation for injuries sustained by one person as the result of the conduct of another." (quoting Cecil A. Wright, Introduction to the Law of Torts, 8 Cameripes L.J. 238 (1944))).
. See Noffsinger v. State, 850 P.2d 647 (Alaska App.1993) (suggesting that legislature may. have switched to several liability system because of "inequity" that results from joint and several liability, where "among several independently negligent parties whose conduct combines to cause an injury, the party who is capable of paying-although only partially responsible-will end up paying the full award"); Laubach v. Morgan, 588 P.2d 1071, 1075 (Okla.1978) (noting that legislative enactment of comparative negligence statute required judicial abolishment of joint and several liability because "[hJolding a defendant tortfeasor, who is only 20 percent at fault, liable for entire amount of damages is obviously inconsistent with the equitable principles of comparative negligence"), quoted in 3 Stem on Personat Insury Damaces Treatises § 14:24 (3d ed.1997).
. As the United States Court of Appeals for the District of Columbia has explained:
[A] cardinal principle of law is that in the absence of punitive damages a plaintiff can recover no more than the loss actually suffered. For when the plaintiff has accepted satisfaction in full for the injury done him, from whatever source it may come ... he is so far affected in equity and good conscience, that the law will not permit him to recover again for the same damages. The office of compensatory damages is to make the plaintiff whole, but certainly not more than whole.
Kassman v. American Univ., 546 F.2d 1029, 1033 (D.C.Cir.1976) (internal footnotes and quotation marks omitted).
. See Keetow supra note 69, § 1, at 7 ('The civil action for a tort ... is commenced and maintained by the injured person, and its primary purpose is to compensate for the damage suffered, at the expense of the wrongdoer.").
. See id. § 1, at 9 ("The idea of punishment, or of discouraging other offenses, usually does not enter into tort law. ...").
74. Luth v. Rogers & Babler Constr. Co., 507 P.2d 761, 766 (Alaska 1973).
. See ch. 139, § 1, SLA 1987 (codified as amended at AS 09.17.070).
. In Chenega Corp. v. Exxon Corp., 991 P.2d 769, 791 (Alaska 1999), we explained that AS 09.17.070 "limits the circumstances in which a victim can receive double recovery, while enhancing the chances that a tortfeasor may not be held fully accountable."