Order; Dissent by Judge KOZINSKI.

The panel has voted to deny appellants' petition for rehearing and Judge Schroeder has voted to reject the suggestion for rehearing en banc. Judges Wiggins and Noonan recommend rejection of the suggestion for rehearing en banc.

The full court was advised of the suggestion for rehearing en banc. An active judge requested a vote on whether to rehear the matter en banc. The matter failed to receive a majority of the votes of the nonrecused active judges in favor of en banc consideration. Fed. R.App. P. 35.

The petition for rehearing is denied and the suggestion for rehearing en banc is rejected.

KOZINSKI, Circuit Judge, with whom BRUNETTI, Circuit Judge, joins and O'SCANNLAIN, SILVERMAN and GRABER, Circuit Judges, join with respect to Parts I and III, dissenting from the order rejecting the suggestion for rehearing en banc.

The panel in this case has written an opinion for the purpose of rebuking Congress as to its policy for sentencing violent criminals. None of the legal issues presented is close or difficult: Defendants did, indeed, use firearms to commit five violent robberies; the district court did sentence them in accord with applicable law; the sentences do not violate the constitutional prohibition against cruel and usual punishments-not by a long shot. As the panel readily admits, the only reason it is publishing an opinion is "to urge Congress to reconsider its scheme of mandatory consecutive minimum sentences and to grant district court judges the discretion to set sentences at the level appropriate for the circumstances of a particular defendant and his or her crimes." United States v. Harris, 154 F.3d 1082, 1083 (9th Cir.1998).

* I do not believe it is appropriate to use an opinion of this court as a vehicle for political lobbying. Once a sentencing scheme is found to be constitutional-as this one has been, see United States v. Wilkins, 911 F.2d 337, 339 (9th Cir.1990)-the decision whether to punish harshly, punish lightly, or give the sentencing judge discretion is a matter of legislative judgment.

My colleagues are entitled to state their views often and forcefully; long live the First Amendment. They can do so in an op-ed piece, a law review article, a concurring opinion or congressional testimony. See, e.g., infra note 2. Any such expressions would reflect only the views of the authors and such others as may choose to join them. But an opinion of the court speaks for all of us, including those who don't consider it wise for a court to become embroiled in political controversy, or who disagree with the views expressed. I do not believe it is fair or appropriate for my colleagues to foist their highly controversial-and possibly incorrect-views on the other judges of this court. Since the judges on the panel insist on couching their policy views as the opinion of the court, I must dissent.

Contrary to the panel's insulting suggestion, Congress has not adopted mandatory minimum sentences as a matter of "political expediency." 154 F.3d at 1085. Rather, Congress carefully and over many years considered the views of a wide variety of law enforcement experts and concluded that giving sentencing judges discretion in setting the punishment for certain violent crimes does not serve the interests of our society. See infra pp. 1005-1010. By contrast, the judges on the panel offer nothing but their gut feelings that the sentences here are too harsh.

The panel delivers its sermon to Congress as if it were dispensing received wisdom. The gospel according to the Ninth Circuit is that any rational, moral sentencing scheme must allow for the possibility of reform or rehabilitation of the offender, except in the most atrocious cases. See 154 F.3d at 1085 (quoting Harmelin v. Michigan, 501 U.S. 957, 1028, 111 S.Ct. 2680, 115 L.Ed.2d 836 (Stevens, J., dissenting)). This is one view of the matter, but not the only view. A rational legislator could surely decide that the criminal laws should serve a single purpose: to protect law-abiding citizens from those who commit violent crimes-particularly repeat offenders-and that rehabilitation of the perpetrator should be given no consideration. Beyond that, a legislator might consider a wealth of data showing quite convincingly that rehabilitation simply does not work for violent criminals-and that those who purport to determine when such criminals have been "rehabilitated" are frequently and tragically wrong.

A widely-reported study by the United States Department of Justice shows that in 1991, criminals out on parole or probation committed over 90,000 violent crimes in this country: over 13,000 murders;1 almost 13,000 rapes; almost 40,000 robberies; almost 20,000 assaults. See U.S. Dep't of Justice, Bureau of Justice Statistics, Probation and Parole Violators in State Prison, 1991: Survey of State Prison Inmates, 1991 (visited Jan. 10, 1998) < http://www.ojp.usdoj.gov/bjs/pub/ascii/ppvsp91.txt>. Of the 2,716 inmates on death row in 1993, almost two-thirds had prior felony convictions, and 28% (some 700) were on parole, probation or pre-trial release at the time of their capital offense. See U.S. Dep't of Justice, Bulletin No. NCJ-150042, Capital Punishment 1993, at 10 (1994).

None of these people were released with the expectation that they would commit mayhem; somebody-a judge, a parole board, a probation officer or some other "expert"-determined that these folks were safe. But the experts were wrong, disastrously wrong, and more than 90,000 innocent people a year pay, often with their lives and bodies, for such mistakes. Of course, these figures represent only those who got caught, so we can infer that parole and probation violators were responsible for many more violent crimes than even these staggering figures suggest.

Rehabilitation and reform are worthy goals; we all want to believe that no one is beyond redemption. But that is very far from saying-as the panel does-that a sentencing scheme which does not leave room for rehabilitation, or which gives rehabilitation a very low priority, is irrational and immoral. See 154 F.3d at 1085 (citing Karen Lutjen, Note, Culpability and Sentencing Under Mandatory Minimums and the Federal Sentencing Guidelines: The Punishment No Longer Fits the Criminal, 10 Notre Dame J.L. Ethics & Pub. Pol'y 389, 389 (1996), for the proposition that such a system of punishment is "morally suspect"). It might be different if rehabilitation could be determined with precision. But our bitter national experience with revolving-door justice shows that rehabilitation is both hard to achieve and extremely difficult to detect. Rational, moral lawmakers could well conclude that people who commit violent crimes are so unlikely to be rehabilitated-and so likely to victimize innocent people-that locking them up for a very long time, perhaps for good, is the only way to secure our safety.

A fair amount of scholarly research supports this view. Noted criminologist John J. DiIulio points out that most violent crimes are committed by a tiny percentage of the population who are habitual offenders and have no realistic prospect of reform. See John J. DiIulio, Jr., Help Wanted: Economists, Crime and Public Policy, 10 J. Econ. Persp. 3, 8 (1996); see also Gwenn Ifill, Crime Proposal's Effect on Gun Use Is Questioned, N.Y. Times, May 24, 1991, at A14 (reporting that 6% of criminals commit up to 70% of all serious crimes); Bonin v. Calderon, 59 F.3d 815, 850-51 (9th Cir.1995) (Kozinski, J., concurring) (detailing cases of violent recidivism).

Nor can the panel justify its view on the ground that these crimes were relatively mild-they were not. Harris, Steward and their cohorts committed a series of vicious robberies, sticking guns in their victims' faces while promising to murder them if they didn't do as they were told. One victim described his ordeal as he was forced to open a cabinet containing money:

I started to turn to see what was going on and that's when I got hit the first time with the gun up side the head.... [T]he impact knocked my glasses off ... and [the glasses] landed about 12 feet away.... I told him that I couldn't see the right key [to open the cabinet] because I didn't have my glasses. Unfortunately, I have really bad vision without my glasses, and that prompted being hit with the gun several more times and kicked in the head and about the body almost two dozen times.... At the time I was thinking, "My God, one of these times, he's just going to cave in my skull."

Another was repeatedly pistol-whipped for taking too long to open a safe:

I was trying to spin the safe. He was yelling at me to get it open, and I was trying to get it open, and I don't recall, he hit me or pushed me, but I was telling him to relax, "I'm trying to get the money, I'm trying to get the safe open," and he was hollering at me. I tried it again. I got hit again, and I became so flustered, I lost the safe combination.... I was face down on the floor, and a little pool of blood on the floor. I don't know if [the bleeding was caused by] hitting the floor [after he was struck] or from being hit, but I must have blacked out for a couple of seconds.

A bystander recounted her terror during one of the robberies:

I saw a guy with a gun. He was holding a gun to Augustine, the dishwasher's head, or actually his neck.... I was talking to my friend and I turned my head and I saw the guy with the gun and I looked to the right and saw a guy with a bag and he was taking money out of the cash register. And I said, "Oh, my God. Are we being robbed?" Because it didn't look real at first.... The guy with the gun said, "Don't F'ing move," and I just said, "Oh." I didn't do anything. And I turned my head a little bit and said, "Sue, don't look at him because we are dead because they don't have masks on. So I have seen him, don't look." And the next thing I know is they are coming-walking out of the bar to go into the back and Sue grabs me by the back and throws me under the bar because I couldn't move, couldn't do anything.

Both scholarly research and anecdotal evidence suggest that people who are willing to use physical violence against other people often start out small and work their way to more serious stuff. Not all violent criminals may be repeat offenders, but those who have committed violent crimes on more than one occasion are likely to commit even more violent crimes in the future, if given a chance to do so. Is it irrational, is it immoral for a conscientious legislator to refuse to take that chance with the lives and safety of the public he is sworn to serve by insisting that repeat violent offenders be locked up for a very long time? I should think not.

Contrary to the panel's disrespectful insinuation of "election-eve" wisdom, 154 F.3d at 1085, Congress has been thoughtful and deliberate in its efforts to tighten up sentences for violent crimes, considering evidence of experts on both sides of the issue, and gauging the effects of one piece of legislation before enacting another. When Senator Arlen Specter introduced legislation in 1981 to require mandatory life sentences for certain types of "career criminals" who engaged in armed robbery and armed burglary, he argued that federal mandatory sentences were necessary to ease the burden on states having to deal with repeat offenders:

[T]he criminal justice system of many states is simply not equipped to provide trial, conviction, and sentencing of career criminals. It is for this reason that I have introduced S. 1688, which addresses itself to the problem of career criminals.... [The Bill] calls for a sentence of life imprisonment, recognizing that once an individual gets to the stage of being a career criminal, it is necessary to separate that person from society, in order, simply stated, to protect society.

Career Criminal Life Sentences Act of 1981: Hearings on S. 1688, S. 1689 and S. 1690 Before the Subcomm. on Juvenile Justice of the Senate Comm. on the Judiciary, 97th Cong. 4 (1981) (statement of Sen. Specter). Before passing the Armed Career Criminal Act of 1984, Pub.L. No. 98-473, ch. 18, 98 Stat. 2185, the first major federal statute providing mandatory minimum sentences for repeat armed robbers, Congress had the benefit of extensive testimony from law enforcement officers and other experts on the feasibility and proper role of mandatory minimum sentences. See, e.g., Armed Career Criminal Act: Hearings on H.R. 1627 and S.52 Before the Subcomm. on Crime of the House Comm. on the Judiciary, 98th Cong. 64 (1984) (statement of Stephen S. Trott, then Assistant Attorney General, Criminal Division). Senator Specter subsequently sought an expansion of the Act based on its initial effectiveness and popularity among law enforcement officials:

The career criminal statute has worked precisely as planned and deserves to be expanded. Local prosecutors use the law in two ways. First, district attorneys refer [the] toughest criminals to the U.S. Attorney for prosecution in Federal courts, which can provide more swift and certain trial and sentencing. Second, district attorneys use the mere prospect of such referral and the extremely stiff penalties to get a favorable guilty plea at the local level. I believe that this "leveraging" effect is one of the most important aspects of the statute. We have seen major cases in which local prosecutors have obtained guilty pleas and stiff sentences simply by raising the possibility of referring the case to Federal authorities for prosecution under the armed career criminal statute.

....

As a member of the Senate Judiciary Committee, I have held hearings in many of the country's largest cities in order to encourage prosecution of career criminals. Based on the testimony presented at these Senate hearings by victims, sheriffs, U.S. attorneys, police, local prosecutors, and other criminal justice officials, and in light of the initial success of the statute, I am convinced that we must expand the law....

132 Cong. Rec. S4325-03 (daily ed. Apr. 16, 1986) (statement of Sen. Specter).

The success of Justice Department initiatives such as Project Triggerlock, which subjected thousands of repeat armed offenders to federal mandatory minimums by allowing state prosecutors to transfer cases involving repeat offenders to federal court, firmed up Congress's commitment to mandatory minimum sentences. See, e.g., 141 Cong. Rec. S6487-01 (daily ed. May 11, 1995) (statement of Sen. DeWine) (reporting that Project Triggerlock took 15,000 career criminals off the street in only 18 months). No less influential have been the opinions of federal and state law enforcement officials. Testimony of officials from the Justice Department and the Bureau of Alcohol, Tobacco and Firearms (BATF), as well as a BATF study, served as a basis for adding more mandatory minimum sentences to the Omnibus Crime Control Act of 1991. See H.R.Rep. No. 102-242(I), 1991 WL 206794, at 542 (1991). Another BATF report to Congress found that as of May 25, 1994, minimum mandatory sentences had prevented over 4 million violent offenses, representing nearly $11 billion in savings to intended victims. See S.Rep. No. 103-286, 1994 WL 281952, at 35 (1994).

While mandatory minimums have many detractors,2 people involved in law enforcement have found them a useful tool. In testimony before the House Subcommittee on Crime and Criminal Justice, the Clinton Administration opposed provisions of the Crime Prevention and Criminal Justice Reform Act of 1994 that would have repealed all mandatory minimum sentences from federal law:

There are ... some provisions in this bill ... that the Administration cannot support. For example, the bill's repeal of all mandatory minimum sentences currently in Federal law ... is not something we can support. The Administration believes that in certain cases mandatory minimum sentences are an appropriate crime-fighting tool and response to violent and recidivist criminal behavior.

Crime Prevention and Criminal Justice Reform Act: Hearings on H.R. 3315 Before the Subcomm. on Crime and Criminal Justice of the House Comm. on the Judiciary, 103d Cong., 1994 WL 14168873, at 12 (1994) (statement of Jo Ann Harris, Acting Deputy Attorney General); see also Federal Gun Crimes: Hearings Before the Subcomm. on Crime and Criminal Justice of the House Comm. on the Judiciary, 103d Cong., 1994 WL 510295 (1994) (statement of Jo Ann Harris, Assistant Attorney General, Criminal Division) (describing the Justice Department's prosecution of federal firearms violations and emphasizing mandatory minimum sentences as important tools of federal law enforcement). A former United States Attorney for the Eastern District of Virginia reported broad support for federal mandatory minimums from state law enforcement officials:

[M]y district ranked in the top 3 in arrests and prosecution of criminals using guns during my tenure. The reason for our success is simple: local police, prosecutors-even mayors-brought the U.S. Attorney cases to prosecute under the strict mandatory sentences found in Title 18. Cooperation among levels of government was at an all time high. There were reports of arrested criminals literally weeping when they learned their case would be prosecuted federally.

In Richmond, [federal-state cooperation initiatives] lead to the elimination of many murderous gangs that were involved in the drug trade. State officials were eager to work with the DEA and FBI in a team approach. The magnet for all of us was the tough mandatory sentences under the federal system and the fact that under federal law a convicted felon served substantially all of his sentence. There is no parole.

Juvenile Crime Issues: Hearings Before the Subcomm. on Crime of the House Comm. on the Judiciary, 104th Cong., 1996 WL 368300, at 4-5 (1996) (statement of Richard Cullen, former United States Attorney for the Eastern District of Virginia) (emphasis added); see also Juvenile Crime: Hearings Before the Subcomm. on Youth Violence of the Senate Judiciary Comm., 105th Cong., 1997 WL 10569690 (1997) (statement of Kenneth W. Sukhia, former United States Attorney for the Northern District of Florida) (pointing to the federal sentencing scheme as a model for dealing with violent juvenile crime). Local law enforcement officials are no less convinced of the need for mandatory minimums, as a Utah sheriff informed the Senate Judiciary Committee: "Today I bring you a strong message from Sheriffs throughout this great nation, that we support tough mandatory minimum sentences for Federal crimes, and that we welcome the Federal help in curtailing the criminal street gang, and the pattern of criminal gang, activity that seems to transcend State borders." Gang Crime Bill: Hearings on S. 54 Before the Senate Judiciary Comm., 105th Cong., 1997 WL 10570055, at 8 (1997) (statement of Sheriff Aaron S. Kennard, President, Utah Sheriffs' Association, and future Vice-President, National Sheriffs' Association).

While cause and effect are always difficult to determine, there is reason to believe that the policy of stiff, determinate sentences adopted both by the federal government and many states is working: Violent crime has declined 7 years in a row, with robbery down 32% and murder 31% since 1991. See Fox Butterfield, Decline of Violent Crimes is Linked to Crack Market, N.Y. Times, Dec. 28, 1998, at A18. Other factors may also have contributed to this trend, but "[t]here is no question that the almost quadrupling of the number of people incarcerated since 1970, to 1.8 million, has incapacitated many criminals and prevented many crimes." Id.; see also William J. Bennett, John J. DiIulio, Jr., & John P. Walters, Body Count 115 (1996) (arguing that longer incarceration reduces crime); Developments in the Law-Alternatives to Incarceration, 111 Harv. L.Rev. 1863, 1880-84 (1998) (reporting that much of the increase in incarceration levels over the last two decades is attributable to mandatory minimum sentences and that statistical data on crime rates since 1960 support an inference that increased incarceration has helped control crime by "incapacitat[ing] dangerous people, who otherwise would do a great deal of harm to the general public").

Against this background, the panel's assertion that "a just system of punishment demands that some level of discretion be vested in sentencing judges to consider mitigating circumstances," 154 F.3d at 1085, is more of a polemic than an indisputable truth. As Judge Kleinfeld points out, judges are not very good at figuring out whether a convicted criminal has been rehabilitated: "We have no professional competence and obtain no useful evidence for looking into offenders' hearts to evaluate their remorse or for judging whether they have truly repented and will in the future maintain law-abiding lives." Andrew J. Kleinfeld, The Sentencing Guidelines Promote Truth and Justice, Fed. Probation, Dec. 1991, at 16, 17. The simple truth is that every time a violent criminal-and in particular a repeat offender-is released into society, law-abiding citizens are put at risk. Rational lawmakers could surely conclude that justice consists, not in a highly uncertain and enormously costly effort to rehabilitate those who have violated the rights of others again and again, but in protecting citizens from the predation of remorseless criminals.

Unlike some other sentencing schemes, where later offenses are punished more severely than earlier ones, see, e.g., Solem v. Helm, 463 U.S. 277, 103 S.Ct. 3001, 77 L.Ed.2d 637 (1983), the scheme here merely ensures that defendants don't get a wholesale discount when they commit a series of violent acts. Why the panel believes that a second, third, fourth and fifth armed robbery ought to be punished less severely than each of the earlier ones, is beyond me. After all, robberies conducted at the point of a gun are extremely dangerous; they can, and often do, result in shootings and hostage-takings. See, e.g., The Gissendaner Verdict, Other Gwinnett Defendants on Georgia's Death Row, Atlanta J. & Const., Nov. 21, 1998, at 12 (robber hijacked a car while trying to escape and shot and killed the driver); Tim Bryant, Second Gunman Gets Death in Robbery at Lindell Bank, St. Louis Post-Dispatch, July 24, 1998, at B2 (two bank robbers murdered guard during the robbery); Natalie Pompilio, Two Tellers Killed in Napoleonville Bank Robbery, New Orleans Times-Picayune, June 5, 1998, at A8 (three bank robbers with criminal records, including one who had served time after pleading guilty to manslaughter for the ambush and slaying of a sheriff, shot two bank tellers in the head during the robbery); Tony Hartzel & Todd Bensman, Woman Killed In Robbery Try; Guard, Bystander Catch Suspect; Another Sought, Dallas Morning News, Oct. 16, 1994, at 35A (woman shot, run over and killed during a botched robbery of a restaurant); Bystander Killed in Botched Robbery, Buffalo News, Apr. 17, 1993, at A12 (bystander crossing the street was killed in the crossfire during an attempted armored car robbery). Every new robbery carries more or less the same risks as each of the previous ones, putting the lives of a new group of innocent people in danger. How can it be immoral or unjust to require the defendants to pay exactly the same penalty for each of the crimes they committed? And since when does justice and morality require us to reduce the punishment for each successive crime because defendants chose to commit several violent crimes one after another? Many rational, moral people believe that those who commit five robberies ought to be punished more than five times as severely as people who commit only one.

As I said at the outset, these are legislative decisions. They call for the kind of judgment we are neither equipped nor empowered to make, and it strikes me as unseemly for our court to take sides on this policy issue. Were I a legislator, I'm not sure how I would vote as to mandatory minimum sentences. But I'm a judge, so I need only determine whether Congress acted rationally enough to pass constitutional muster. For the reasons explained above, it clearly did and that ends the matter for me. If my colleagues feel they have something useful to add to the debate, they can do so as individuals and not on behalf of the court. But since they have chosen to deliver their homily from the bench we share, I write to explain that they do not speak for us.