In March 1988 Cori Smith, then four years old, picked up a cigarette lighter and set a small fire in her parents’ bedroom. She was admonished not to play with lighters or matches — of which the household had many, because all four adults smoked cigarettes. About a week later Cori found a BIC lighter on a table in the living room. Cori used the lighter to set ablaze some papers in the bedroom where Tiffany Todd, then 23 months old, was sleeping. Tiffany died in the conflagration; the adults and the five other children in the household survived. Tiffany’s father Rodney, as special administrator of her estate, filed this diversity action seeking damages from the fighter’s manufacturer. The parties agree that Illinois, where the fire occurred, supplies the governing law.
Plaintiff has two principal theories: that BIC should have warned customers about the risks fighters pose to households with children, and that BIC should have designed its lighters to resist children’s efforts to use them. Failure to choose the right design or give proper warnings made the fighter defective or unreasonably dangerous, in either case leading to liability in tort. BIC moved for summary judgment. It conceded that misuse of lighters by children is foreseeable and that it is possible to make child-resistant fighters at some cost in both money and inconvenience to adult users. In 1992 BIC began selling a child-resistant lighter, and the Consumer Product Safety Commission has directed all other manufacturers to follow suit. 16 C.F.R. Part 1210 (effective July 1, 1994), 58 Fed.Reg. 37557, 37584-91 (July 12, 1993). BIC contended, however, that it gave sufficient warnings and that Illinois does not require manufacturers to make their products child-resistant when parents may take effective precautions. The district court granted BIC’s motion and dismissed the suit. 1992 WL 4971 1992 U.S.Dist. Lexis 88 (N.D.I11.). A divided panel of this court reversed, concluding that a trial is necessary on each of plaintiffs theories. 991 F.2d at 1334 (1993). Concern about the implications of the panel’s reasoning led the full court to vacate the panel’s decision and set the case for hearing in banc. 991 F.2d 1344 (1993).
I
That fire attracts youngsters — and that cigarette fighters in the hands of children can lead to calamity — no one doubts. According to the CPSC, “for the period 1988-90, these fires [set by children under 5 playing with fighters] caused an annual average of 150 deaths, approximately 1,100 injuries, and nearly $70 million in property damages.” 58 Fed.Reg. at 37564. BIC recognized this danger, and its fighters were emblazoned: “KEEP OUT OF REACH OF CHILDREN”. Plaintiff deems this warning insufficient because BIC did not tell parents that children between three and five are attracted to flame, able and eager to open closets, cabinets, and purses in order to inspect their contents, and unable to follow instructions not to fiddle with what they find there.
Parents who followed BIC’s advice to keep their fighters out of the reach of children did not need separate admonition about youngsters’ inability to follow directions. Households are full of potentially dangerous items, including knives, matches, and drugs, that are bound to cause injury in inexperienced hands. BIC told owners to make access physically impossible, not to issue instructions to the children. Had the adults in this household followed the warning, there would not have been a fire.
Manufacturers could of course provide secondary warnings about the consequences of not following primary warnings. BIC could have written something like: “Keep this fighter out of the reach of children, and be aware that children not only are resourceful in finding things but also are apt to disobey your instructions not to play with lighters they can get their hands on.” It could have amplified this longer warning with data about the number of fires children set with fighters, in order to impress on parents the importance of following the primary warning to make the lighters inaccessible (or, perhaps, to induce the parents to quit smoking). Extended warnings present several difficulties, first among them that, the more text must be squeezed onto the product, the smaller the type, and the less likely is the consumer to *1219read or remember any of it. Only pithy and bold warnings can be effective. Long passages in capital letters are next to illegible, and long passages in lower case letters are treated as boilerplate. Plaintiff wants a warning in such detail that a magnifying glass would be necessary to read it. Many consumers cannot follow simple instructions (including pictures) describing how to program their video cassette recorders. To be more than a scare tactic, the warning could not stop with the number of fires and deaths. It would have to include the number of fighters sold to households with small children (so that the buyer could determine the risk per lighter) and the number of fires that children set with matches (so that the buyer could evaluate whether it is safer to switch). These numbers are abstract. For a parent determined to smoke, the right question is what to do. BIC provided that information.
There is a further practical inquiry: will consumers who disdain a bold and (if followed) effective warning be influenced by smaller and more subtle points? If parents leave fighters on living room tables despite “KEEP OUT OF REACH OF CHILDREN”, and despite knowing that lighters cause fire (which is why the adults bought them), is a recitation of the CPSC’s data likely to alter their conduct? These adults did not heed the Surgeon General’s dire warnings, prominent on every package of cigarettes, about the hazards of smoking. They exposed their children to tobacco smoke, which causes more harm than does playing with fighters. Environmental Protection Agency, Respiratory Health Effects of Passive Smoking (1992). Would a screed on their fighters based on the CPSC’s data about fires succeed where concise advice failed? Extended advice, reducing the likelihood that parents will notice and follow the principal warning, is particularly ill-advised when the additional information is already well known. What parent is unaware that children between the ages of three and five explore and test their surroundings even when told not to? Can it be that only a warning by BIC Corporation will alert parents that kids are at least as successful as cats in getting into cabinets and other hiding places, and that children have minds of their own? Illinois does not require manufacturers to warn consumers about facts they already know, and it does not require manufacturers to dilute the principal warnings with distracting information. Genaust v. Illinois Power Co., 62 Ill.2d 456, 466, 343 N.E.2d 465, 471 (1976); Riordan v. International Armament Corp., 132 Ill.App.3d 642, 648, 87 Ill.Dec. 765, 769, 477 N.E.2d 1293, 1297 (1st Dist.1985).
Plaintiffs warning theory encounters an additional problem: causation. Rodney Todd and the three other adults in the household had received the most vivid warning imaginable: Cori had started a fire with a cigarette lighter only a week earlier. What words could be more potent than this incident? The household knew about both the risk of fire and about Cori’s proclivity and ability. All four adults conceded this during their depositions, adding that they appreciated the hazard even before Cori set her first fire. It is inconceivable that adults who left a cigarette lighter within the grasp of a child who had already started a fire would have been influenced by any written warning. Cf. Murphy v. Corey Pump & Supply Co., 47 Ill.App.2d 382, 197 N.E.2d 849 (1st Dist. 1964). BIC accordingly is entitled to summary judgment on the warning branch of the case.
II
Since 1965 Illinois has held manufacturers strictly liable for injuries caused by defective products, following the approach of Restatement (2d) of Torts § 402A (1965). See Suvada v. White Motor Co., 32 Ill.2d 612, 210 N.E.2d 182 (1965). “Strict liability” does not imply that manufacturers insure consumers against injuries associated with the use (or misuse) of their products. Some products are dangerous even when properly designed, and it is both easier and cheaper for consumers to obtain their own insurance against these risks than to supply compensation case-by-case through the judicial system. “Virtually any product is capable of producing injury when put to certain uses or misuses .... Injuries are not compensable in products liability if they derive merely from *1220those inherent properties of a product which are obvious to all who come in contact with the product. The injuries must derive from a distinct defect in the product, a defect which subjects those exposed to the product to an unreasonable risk of harm.” Hunt v. Blasius, 74 Ill.2d 203, 211-12, 28 Ill.Dec. 574, 578, 384 N.E.2d 368, 372 (1978).
What, then, does Illinois define as a “defect”? Until recently a state court would have dispatched a case such as ours with the observation that a “dangerous product which bears a warning, and which is safe if the warning is followed, is neither defective nor unreasonably dangerous.” Dugan v. Sears, Roebuck & Co., 113 Ill.App.3d 740, 741, 69 Ill.Dec. 620, 623, 447 N.E.2d 1055, 1058 (1st Dist.1983) (citing Restatement § 402A comments j and k). The BIC lighter would have been safe had the adults kept it out of Cori’s reach, as the warning stated. Jurisdictions that have applied § 402A to cigarette lighters have concluded on this reasoning that the absence of child-resistant features does not make the lighters defective or unreasonably dangerous. E.g., Adams v. Perry Furniture Co., 198 Mich.App. 1, 497 N.W.2d 514 (1993); Griggs v. BIC Corp., 981 F.2d 1429, 1431-34 (3d Cir.1992) (Pennsylvania law); Byler v. Scripto-Tokai Corp., 944 F.2d 904 (6th Cir.1991) (Kentucky law). But cf. Bean v. BIC Corp., 597 So.2d 1350 (Alabama 1992) (declining to apply any legal approach to cigarette lighters in advance of discovery).
A case decided in 1990 draws this straightforward approach into question. Lamkin v. Towner, 138 Ill.2d 510, 528, 150 Ill.Dec. 562, 570, 563 N.E.2d 449, 457 (1990), explains:
A plaintiff may demonstrate that a product is defective in design, so as to subject a ... manufacturer to strict liability for resulting injury, in one of two ways: (1) by introducing evidence that the product failed to perform as safely as an ordinary consumer would expect when used in an intended or reasonably foreseeable manner or (2) by introducing evidence that the product’s design proximately caused his injury and the defendant fails to prove that on balance the benefits of the challenged design outweigh the risk of danger inherent in such designs.
See also Doser v. Savage Manufacturing & Sales, Inc., 142 Ill.2d 176, 197-98, 154 Ill.Dec. 593, 603, 568 N.E.2d 814, 824 (1990). Plaintiff invokes both of these methods. It is foreseeable that some lighters will fall into the hands of children, who do not appreciate their risks. Thus plaintiff contends that BIC is liable under Lamkin’s first approach, the “consumer contemplation test.” As for the second or “risk-utility test,” plaintiff contends that the costs and benefits of adding child-resistant features to lighters are debatable questions that must be submitted to a trier of fact. BIC replies that the consumer contemplation test asks whether the purchaser can appreciate the risks and use the product safely — so that the customer may make an intelligent decision whether the benefits of the item outweigh the risks — rather than whether the manufacturer can foresee that non-purchasers will acquire and misuse the product. On any other approach, BIC observes, kitchen knives are defective products because children may cut themselves, aspirin is defective because children may take too many tablets, and so on. According to BIC, the risk-utility test applies only to complex products or the unexpected behavior of simple products. A lighter is not complex, people can understand its actions (and risks), and this lighter performed just as it was supposed to.
Each side has support for its position. Start with consumer contemplation. None of the cases expressly addresses the question whether the “consumer” for this purpose is the purchaser or includes any user (or misuser) the manufacturer can foresee. Plaintiff invokes what he characterizes as the plain language of Lamkin. BIC replies that, although the persons injured in that case were children who fell through a screen window, the court asked whether their parents could appreciate the risk rather than whether manufacturers could foresee that children could fall through if not restrained by their parents. Lamkin spoke of the ordinary consumer, not the foreseeable user. Doser says that the “liability of a manufacturer extends to those individuals to whom injury from an unreasonably dangerous product may be reasonably foreseen” (142 I11.2d at 197-98, 154 *1221IU.Dec. at 603, 568 N.E.2d at 824) but does not tell us whether foreseeable risks to non-purchasers are themselves enough to make the product “unreasonably dangerous.”
On the risk-utility front things are no clearer. Once again plaintiff emphasizes the language of the passage in Lamkin we have quoted, but judicial opinions should not be confused with statutes. Qualifications often are implied or developed as the judges grapple with additional circumstances. Scoby v. Vulcan-Hart Corp., 211 Ill.App.3d 106, 155 Ill.Dec. 536, 569 N.E.2d 1147 (4th Dist.1991), holds that when the product is simple or the risks easy to appreciate, the consumer contemplation test is the sole measure, emphasizing a portion of Lamkin that reiterated the observation in Hunt that strict liability is limited to attributes of the product that are hidden or otherwise faU outside consumers’ usual expectations. Scoby concluded: “We do not deem that Lamkin or other cases ... intend that all manufacturers of products ... should be subject to UabiUty depending upon a trier of fact’s balancing under that test ... Considering not only the obvious nature of any danger here but, also, the simple nature of the mechanism involved, we conclude that the circuit court properly apphed only the consumer-user contemplation test.” 211 Ill. App.3d at 112, 155 Ill.Dec. at 540, 569 N.E.2d at 1151. See also Mason v. Ashland Exploration, Inc.,. 965 F.2d 1421, 1428 n. 8 (7th Cir.1992) (treating Scoby as an authoritative gloss on Lamkin); Carroll v. Otis Elevator Co., 896 F.2d 210, 215-18 (7th Cir.1990) (concurring opinion) (doubting whether a universal cost-benefit approach to design questions in tort litigation would improve product quality at tolerable cost). Estimating the costs and benefits of a child-resistant design could be quite difficult, with corresponding costs and risks of error.†' For his part, plaintiff emphasizes Harnischfeger Corp. v. Gleason Crane Rentals, Inc., 223 Ill.App.3d 444, 165 Ill.Dec. 770, 585 N.E.2d 166 (5th Dist.1991), which applied the risk-utility test to the design of a crane, even though the danger in question — electrocution from bringing the crane into contact with overhead power Unes — was obvious to everyone. Hamisch-feger concluded that a cost-benefit analysis of the product’s design is essential no matter how transparent the risk. BIC insists that Hamischfeger is limited to complex products such as cranes, and a few cases suggest such a limitation even whüe applying the risk-utihty test to the product at hand. Besse v. Deere & Co., 237 Ill.App.3d 497, 500-01, 178 Ill.Dec. 475, 478, 604 N.E.2d 998, 1001 (3d Dist.1992) (citing Scoby favorably).
When acting under the diversity jurisdiction, a federal court must attempt to decide the case as the highest court of the state supplying the law would do. The disagreement among the appellate courts of IUi-*1222nois shows the difficulty of doing this. Certification is an alternative to prognostication. See Lehman Brothers v. Schein, 416 U.S. 386, 390-91, 94 S.Ct. 1741, 1743-44, 40 L.Ed.2d 215 (1974); Charles Alan Wright, Arthur R. Miller & Edward H. Cooper, 17A Federal Practice and Procedure § 4248 (2d ed. 1988). Circuit Rule 52 permits us to certify a question of state law when the answer “will control the outcome” of the case. Rule 20 of the Rules of the Supreme Court of Illinois authorizes that court to answer a certified question when the answer “may be determinative of the said cause”.
Most debatable issues of state law that arise in diversity litigation must be resolved in federal court if only because certification of all disputes would overwhelm the capacity of the state’s highest court. The federal court may essay an answer, resolving the case without subjecting either the parties or the state to the need for an additional hearing, yet without hampering the state’s ability to craft its own law. Sometimes, however, the federal court’s answer may have effects independent of its persuasive power. In a case such as ours, for example, any substantial divergence between the federal court’s estimate of state law and the state’s view of its own law will funnel all similar litigation to federal court. None of the manufacturers or principal importers of cigarette lighters is a citizen of Illinois (indeed, only BIC manufactures any disposable cigarette lighters in the United States, 58 Fed.Reg. at 37565), and the damages in these cases exceed $50,000, so the diversity jurisdiction is always available. If the federal court treats the plaintiff more favorably than the state tribunal would, then the plaintiff always files in federal court; similarly any departure in the manufacturer’s favor leads the defendant to remove any suit filed in state court. In either case, the state loses the ability to develop or restate the principles that it believes should govern the category of cases. Certification then ensures that the law we apply is genuinely state law. Covalt v. Carey Canada Inc., 860 F.2d 1434, 1440-41 (7th Cir.1988). The panel’s analysis had substantial propensity to attract all future cases of this kind into federal court; an error in either direction could do so. Little would be served by substituting the guess of eleven judges for that of three; far better to pose the questions to the only judges who can give definitive answers.
Now that the warning question has been resolved, the consumer contemplation and risk-utility tests have become potentially dispositive, making certification appropriate under both state and federal rules. We respectfully request the Supreme Court of Illinois to answer the following questions:
1. Whether the “consumer” for purposes of the consumer contemplation test includes non-purchasers who the manufacturer should foresee will use or misuse the product.
2. Whether the risk-utility test applies to consumer products whose risks can be appreciated by their intended users.
It may be that when revisiting the state’s law of products liability the Justices of the Supreme Court of Illinois will conclude that these are not the right questions. In that event, the Justices should feel free to reformulate the questions, just as they would when dealing with the issues posed in a petition for leave to appeal. It is not our purpose to constrain the state court. The Clerk will transmit the record and briefs in this case to the Supreme Court of Illinois.
Both costs and benefits are elusive. The CPSC estimates that child-resistant lighters would cost approximately 15$ to 20$ more per unit in the marketplace. 58 Fed.Reg. at 37566. About 678 million lighters are sold in the United States each year. Id. at 37563. The CPSC could not estimate total costs because it did not know what would happen to sales at the higher prices (particularly for "specialty” lighters, which sell in smaller quantities and therefore would incur much higher per-unit costs of compliance). Users of the lighters will suffer some inconvenience, which also counts as a cost — especially to older users whose fingers may not be strong or nimble enough. The benefits look clear enough: by the CPSC's estimate, the total cost of fires set by children under five playing with lighters is $385 million per year. Id. at 37564. At first glance, the benefits easily exceed the costs. But child-resistant lighters will not eliminate the fires. Child-resistant is not child -proof; the CPSC estimates that 15% of children will be able to use the lighters notwithstanding the safeguards and will set 30% of the former number of fires. Id. at 37564. The cost and inconvenience will lead some adults to switch from lighters to matches, posing fire hazards of their own (which CPSC estimates at one-third of the risk of lighters), or to refillable lighters, which are not covered by the rule. And all child-resistant designs present a subtle risk. Parents who (mistakenly) believe that the products are child-proof are more likely to leave them within youngsters' reach. Instead of having zero ability to play with lighters, formerly out of their grasp, these children now have a 15% chance of being able to set a fire. A false sense of security could be fatal. The introduction of child-resistant medicine bottles actually led to an increase in certain kinds of poisonings as parents relaxed their vigilance. W. Kip Viscusi, Consumer Behavior and the Safety Effects of Product Safety Regulation, 28 J.L. & Econ. 527, 537-48 (1985); see also Cass R. Sun-stein, Paradoxes of the Regulatory State, 57 U.Chi. L.Rev. 407, 417-19 (1990). It is hard to know whether child-resistant lighters will produce the same effect.