dissenting in part: I respectfully dissent from that portion of the majority opinion which holds that the recommitment of this defendant for a period of five years would be unconstitutional as an ex post facto application of the most recent amendment of RSA 651:ll-a (Supp. 1983).
To explain my disagreement, I begin with a consideration of the earlier proceedings in the case. In 1980, the defendant pleaded not guilty by reason of insanity to an indictment charging that he had committed the criminal offense of second-degree assault that same year. See RSA 631:2 (Supp. 1983). The State and the superior court accepted his plea under RSA 651:8-a (Supp. 1983).
The acceptance of the plea established that the defendant had been insane at the time he committed the acts charged in the indictment and, for that reason, was not guilty of committing the criminal offense of second-degree assault. See id. The acceptance of the plea did not, however, authorize the court or the State to commit the defendant to an institution in derogation of his liberty. Rather, under the law then and now, the defendant was entitled to be released unless a hearing was held to determine his present mental *314condition, and unless the State then proved beyond a reasonable doubt that the defendant would be dangerous if allowed to go at large. RSA 651:9 (Supp. 1983). It is crucial to recognize that the issue at such a hearing is not whether the defendant had been insane in the past; that legal conclusion has already been established. Nor is the issue whether the defendant had been dangerous at any time in the past, however relevant that may be in determining how he will probably act in the future. The issue, and the only issue, is whether at the time of the hearing he is and will remain dangerous. RSA 651:9 (Supp, 1983).
In the present case, the court held an evidentiary hearing to determine the defendant’s mental condition at that time. It found beyond a reasonable doubt that at the time of the hearing the defendant’s mental condition was such that it would be dangerous for him to go at large. It therefore committed him to the New Hampshire Hospital under RSA 651:9 (Supp. 1983). As it read in 1980, RSA 651:ll-a (Supp. 1983) limited the commitment to a period of two years. If a court had not modified its order and released the defendant during those two years, on their expiration the State was required either to release the defendant or to prove beyond a reasonable doubt that his mental condition would still make it dangerous for him to go at large. If the State did prove that, the statute authorized a recommitment for a further period of two years. Id.
In this case, during the two-year period of the defendant’s original commitment, the legislature amended RSA 651:ll-a (Supp. 1983) to increase the maximum duration of a commitment from two years to five. Laws 1982, 34:2. The amendment did not purport to increase the maximum length of any orders of commitment issued before its effective date. It simply authorized commitments after the effective date of the amendment to be as long as five years. The amendment took effect before the defendant’s original two-year commitment had expired.
When that original commitment did expire, the court held a new evidentiary hearing. The State presented evidence about the defendant’s mental condition at the time of the hearing, after which the court found beyond a reasonable doubt that the condition was still such that the defendant would be dangerous if at large. The court therefore recommitted the defendant for a further period but transferred to us the pending question about its permissible duration. The State claims the amendment should be applied to authorize a recommitment for up to five years. The defendant claims that part I, article 23 of the Constitution of New Hampshire precludes recommitment for more than the two-year period provided by the *315law at the time he committed the underlying acts and at the time of his plea.
To decide between these two positions, we must start with the laconic provisions of article 23:
“Retrospective laws are highly injurious, oppressive, and unjust. No such laws, therefore, should be made, either for the decision of civil causes, or the punishment of offenses.”
While the article does not define its principal term, retrospective, it does describe two categories of laws that may be retrospective, as “laws ... for the decision of civil causes” or cases, and laws for “the punishment of offenses.” The decisions that have interpreted the article have followed the same distinction, and to understand the article’s prohibition, we really must consider two separate categories of retrospectivity.
The early leading case of Woart v. Winnick, 3 N.H. 473 (1826) established the doctrinal distinction between these categories by first considering the characteristics that would make a law for “the punishment of offenses” retrospective. The court did this by equating that kind of penal retrospective law with what article I, section 10 of the National Constitution also prohibits under the name of an ex post facto law. Following this early case, when we speak of the application of article 23 to laws for “the punishment of offenses,” we commonly speak of it simply as forbidding ex post facto laws, or ex post facto applications of laws, even though article 23 does not use that Latin term.
The court in Woart described a prohibited ex post facto law, or ex post facto application of a law, by paraphrasing language from the earlier federal case of Calder v. Bull, 3 U.S. (3 Dall.) 386 (1798), which had construed the federal clause:
“Every law, which makes an action done before the passing of the law, and which was innocent when done, criminal, and punishes such action; or which aggravates a crime, and makes it greater, than it was when committed; or which changes the punishment, and inflicts greater punishment, than the law annexed to the crime when committed; or which alters the legal rules of evidence, and receives less or different testimony, than the law required at the time of the commission of the offence, in order to convict the offender is an ex post facto law.”
Woart v. Winnick, supra at 475.
The identity of the State and national provisions in the context of penal laws, and the content of those provisions, remain today essen*316tially as the court described them in Woart. See State v. Theodosopoulos, 123 N.H. 287, 461 A.2d 100 (1983); State v. Breest, 116 N.H. 734, 367 A.2d 1320 (1976); Rich v. Flanders, 39 N.H. 304 (1859); Weaver v. Graham, 450 U.S. 24 (1980).
The court in Woart then turned to the application of the article to laws for deciding “civil causes,” by quoting Mr. Justice Story, in another earlier federal case that had applied article 23:
“‘[EJvery statute, which takes away or impairs vested rights, acquired under existing laws, or creates a new obligation, imposes a new duty, or attaches a new disability, in respect to transactions or considerations already past, must be deemed retrospective’....”
Woart v. Winnick, supra at 479 (quoting Society v. Wheeler, 22 F. Cas. 756, 767 (C.C.D.N.H. (1814) (No. 13,156))).
These two extended quotations are the sources of the distinction between the penal and civil applications of article 23 that has persisted and increased in breadth to this day. While the application of the article to laws for the “punishment of offenses” has not changed significantly over the years, its application to laws for deciding civil cases has undergone continual development. In the civil sphere, the rule has come to be that a retrospective law is not absolutely unconstitutional, but may be applied if some public benefit can be shown to outweigh the injury said to flow from retrospective application. Compare Estate of Kennett v. State, 115 N.H. 50, 333 A.2d 452 (1975), and Hayes v. LeBlanc, 114 N.H. 141, 316 A.2d 196 (1974), with Smith Insurance, Inc. v. Grievance Committee, 120 N.H. 856, 424 A.2d 816 (1980), and Geldhof v. Penwood Associates, 119 N.H. 754, 407 A.2d 822 (1979).
Such distinctions between the consequences of retrospectivity in the penal and civil spheres should not concern us in this case, however, unless the application of the statute as amended would be either ex post facto or retrospective. It would not be either one.
In considering, first, whether the application of the amendment would be ex post facto in the context of the present case, we must determine whether the law as amended is one for “the punishment of offenses.” That issue turns on whether the application of the amendment would fall within any of the categories described in Calder v. Bull, 3 U.S. (3 Dall.) at 390, as paraphrased in Woart v. Winnick, 3 N.H. at 475. There does not appear to be any basis to argue that the amendment provides for the punishment of a new crime, or that it alters any rules of evidence bearing on proof of guilt, or that it aggravates the criminal character of any prior offense, as distinguished from its penalty.
*317Under the rule in Calder, all that remains is the possibility of arguing that the amendment would inflict a greater punishment for the crime or offense than the law imposed when the defendant committed the offense. This argument would fail. The amendment would be applied to this defendant on the same grounds that the earlier two-year provision may be applied. Under either provision there is no “offense” that justifies commitment. Only proof of dangerousness can justify commitment, and a commitment on grounds of dangerousness is not punishment. Jones v. United States, 103 S. Ct. 3043 (1983).
It is clear right on the face of the commitment statute that it applies not to those who have committed an offense, but to those who have been found to be still dangerous following an insanity plea. RSA 651:9 (Supp. 1983). It is not an offense to be dangerous, and the finding of insanity eliminated the possibility that the acts charged in the indictment could constitute a criminal offense. As the earlier discussion indicated, a plea of not guilty by reason of insanity is a plea in confession and avoidance, and its acceptance is inconsistent with criminal liability. Novosel v. Helgemoe, 118 N.H. 115, 384 A.2d 124 (1978). Thus, a statute regulating commitments of dangerous defendants following findings of insanity does not deal with criminal offenses and, by definition, could not do so.
Nor would the commitment be “punishment” within the meaning of article 23 or Calder v. Bull supra. The duration of any such commitment is not limited to the period of imprisonment provided by the statute for those who are criminally liable for committing such acts, and no such commitment of any duration can be authorized merely on the basis of the defendant’s acts as charged in the original indictment. On the contrary, this defendant can be recommitted on one condition only, that there is no reasonable doubt that his mental condition at the time of recommitment would make it dangerous for him to be at large. Such a commitment is imposed for the sake of safety, not for sake of penal policy, and its duration is limited to the duration of the dangerous condition that justifies it in the first instance.
If, for constitutional purposes, such a commitment were a punishment for an offense, then the majority would have to do more than they purport to do in deciding this case, for every civil involuntary commitment under RSA 135-B:37 based on “likelihood of danger” as defined in RSA 135-B:26 would equally be a punishment for an offense. It would then be necessary to overrule State v. Hudson, 121 N.H. 61, 425 A.2d 255 (1981), which held that an involuntary civil commitment cannot give rise to a double jeopardy claim.
*318In summary, there is no offense because the defendant has been found to be insane. There is no punishment because the defendant is committed for the sole reason that he is dangerous at a time after the time he did the acts charged in the indictment and after the date of the enactment of the statutory amendment in question. As a majority of the Supreme Court of the United States held in Jones v. United States, 103 S. Ct. 3043, 3052 (1983), “[a]s he was not convicted, he may not be punished.” The application of the enactment cannot, therefore, be ex post facto under the State or the National Constitutions.
The majority seek to avoid this analysis by their conclusion that “recommitments pursuant to RSA 651:ll-a (Supp. 1983) are criminal proceedings. See Opinion of the Justices, 122 N.H. at 203, 442 A.2d at 595 (1982) (‘reasonable doubt standard of proof was constitutionally required in criminal recommitment proceedings.’)” One may agree that in and out of court we commonly use the word “criminal” to describe both the insanity of defendants in insanity cases and the proceedings in question. One may agree, too, that the administrative provisions governing commitment and release of such defendants are different in some respects from those that govern the “civil” sphere.
Considering all of this, the majority position remains insufficient to sustain its conclusion, simply because nothing turns under the ex post facto aspect of article 23 on the mere fact that the proceedings may in some sense be labelled as “criminal.” But everything turns on whether the law being applied is a law for the punishment of an offense. Because this law is not, the majority are wrong to conclude that its application can be ex post facto. Indeed, the position taken by the majority runs counter to virtually all cases from other jurisdictions that have considered the application of the ex post facto prohibition in circumstances analogous to the present one. See People v. Buttes, 184 Cal. Rptr. 497 (Cal. App. 1982); People v. Valdez, 79 Ill. 2d 74, 402 N.E.2d 187 (1980); People v. Thiem, 82 Ill. App. 3d 956, 403 N.E. 647 (1980); In re Clark, 86 Kan. 539, 121 P. 492 (1912); Matter of Rogers, 63 N.C. App. 705, 306 S.E.2d 510, appeal dismissed, 309 N.C. 633, 308 S.E.2d 716 (1983); State v. Jackson, 2 Ohio App. 3d 11, 440 N.E.2d 1199 (1981); Matter of Guardianship of Nelson, 98 Wis. 2d 261, 296 N.W.2d 736 (1980); contra Raimondo v. Pavkovic, 107 Ill. App. 3d 226, 437 N.E.2d 712 (1982) (holding of intermediate appellate court of Illinois inconsistent with holding of Supreme Court of Illinois in People v. Valdez supra); see People v. Tedford, 109 Ill. App. 3d 195, 440 N.E.2d 329 (1982).
Though the majority have erred in finding the proposed application of the amendment to be ex post facto, the question remains *319whether it would be retrospective as falling within any of the categories of retrospective applications laws for deciding “civil causes.” Would it impair any “vested right” of the defendant “or attach ... a new disability, in respect to transactions or considerations already past”?
We must approach the answers to these questions by recognizing that there is no existing rule of statute or common law giving the defendant a vested right in the continuance of the earlier statutory provision regulating the length of commitment for dangerousness. Outside the sphere of the ex post facto, generally no one has a vested right to the continuance of prior law. Farnum’s Petition, 51 N.H. 376 (1871).
Any vested right must flow, then, from the circumstances of this defendant’s relationship to the State. Perhaps his most persuasive argument on its face would be that he had relied on the rule in State v. Goodrich, 116 N.H. 477, 363 A.2d 425 (1976), as guaranteeing the continuation of the conditions he expected when he entered his plea of not guilty by reason of insanity. That case held that a defendant may withdraw a plea of guilty when the court declines to follow the State’s recommendation made in accordance with a plea agreement.
The attempted analogy with Goodrich will not hold up, however. In this case there is no record of a plea agreement with terms relating to the disposition of the defendant. There was only an agreement that the State would accept a plea that would relieve the defendant of the possibility of criminal liability. Since the defendant was not a party to any agreement about disposition, there is no basis to elevate his hope or expectation to the level of a right that deserves to be treated on equitable grounds as vested. Were there any doubt on this issue, I believe that a full exposition of the competing concerns involved would support a ruling against the defendant on grounds of paramount public interest under Estate of Kennett v. State, 115 N.H. 50, 333 A.2d 452 (1975), and Hayes v. LeBlanc, 114 N.H. 141, 316 A.2d 187 (1974).
The only arguable possibility remaining for finding retrospectivity under the Woart tests for “civil causes” lies in the category of disabilities imposed “in respect to transactions or considerations already past.” Would the extended period of commitment be ordered on the basis of transactions or considerations antedating the amendment? It clearly would not be.
As I have noted before, no recommitment, be it for five years or two years, may be justified except on the basis of a finding beyond a reasonable doubt about the defendant’s dangerous condition at the time the court orders the recommitment. It is not enough that the defendant did certain acts in the past, or that he was dangerous in *320the past. He may not, in other words, be committed for a past “transaction” or “consideration.” Rather, he must be found to be dangerous at the time of recommitment, and that time is after the enactment of the statute, not before it.
The only conceivable basis for arguing that application of the amended statute would authorize a longer commitment based on “past transactions or considerations” would be a claim that article 23 forbids the consideration of behavior before the statutory change as evidence of dangerousness after the statutory change. However, when the use of evidence antedating the statutory enactment is not alone sufficient to impose a post-enactment burden, its use has never been held to be an example of retrospectivity.
Indeed, In re Moulton, 96 N.H. 370, 77 A.2d 26 (1950), rejected the argument that article 23 precluded the petitioner’s commitment as a sexual psychopath based on evidence of acts committed before the sexual psychopath statute was enacted by Laws 1949, chapter 314. Indeed, the only evidence specifically described in that opinion was evidence of pre-statutory activity. Justice Kenison wrote for the unanimous court that the provision under which the defendant was committed “is not retrospective merely because it considers the past history of the person involved. The determination in 1950 that he was a sexual psychopath in 1950 is not made retrospective in its operation because the examining board and the court considered his life history prior to the passage of its act.” Id. at 374, 77 A.2d at 29.
Nothing in the record before us distinguishes this case from Moulton. In each case a statute has authorized a commitment that was not authorized by prior law. In Moulton the commitment had not been authorized at all; here it was not authorized beyond two years. In each case the commitment has been authorized on the basis of a finding about the subject’s condition at a time after the enactment of the new law. In each case the court has based the finding in part on evidence of facts existing before enactment of the new law. In each case some of the evidence has been furnished by the subject himself before enactment of the new law: in Moulton by a guilty plea to a prior offense, in this case by a plea of not guilty by reason of insanity, which admitted the acts charged in the indictment.
Despite these points of identity, the majority seek to distinguish Moulton on two grounds. They say it upheld the application of a statute that has since been repealed. But the question should be, has article 23 been repealed or amended since Moulton. They say the statute in Moulton authorized a “civil” commitment, as indeed it did. But Moulton lost his liberty just as surely as the present defendant has lost his, and for essentially the same reason. The majority ignores this by suggesting that constitutional distinctions should *321turn on statutory or colloquial labels like “civil” or “criminal” rather than on the realities of liberty and basis for its loss. In the past this court has rejected such an approach, Proctor v. Butler, 117 N.H. 927, 380 A.2d 673 (1977), and we should reject it now.
Moulton simply recognized what is apparent in the present case, that the challenged commitment must rest on a post-enactment finding of post-enactment dangerousness. The application of such an enactment as the present amendment is therefore not retrospective, and article 23 does not forbid a recommitment for as long as five years.
BROCK, J., concurs in the dissent.