Kremen v. Cohen , 325 F.3d 1035 ( 2003 )

KOZINSKI, Circuit Judge,

dissenting.

When a federal court certifies a case to a state supreme court, it draws from a limited reservoir of comity. Certifying the case shifts the difficult work of deciding it to the state court, which is often so busy keeping its own house in order that it scarcely has time for our overflow laundry. Certification also burdens litigants, forcing them to reargue the case in a different forum — a process that is costly and full of delay. None of the parties or amici in our case has so much as hinted that we should certify; Kremen explicitly urged us not to, citing the many years already spent in litigation.

I am aware of the prevailing infatuation with this procedural device — the “sacred cow in our modern judicial barnyard.” Bruce M. Selya, Certified Madness: Ask a Silly Question ..., 29 Suffolk U.L.Rev. 677, 678 (1995). But we have a duty to use it sparingly and sensibly; that a case raises difficult legal questions is not enough. See L. Cohen & Co. v. Dun & Bradstreet, Inc., 629 F.Supp. 1419, 1423 (D.Conn.1986) (Cabranes, J.). Certification is justified only when the state supreme court has provided no authoritative guidance, other courts are in serious disarray and the question cries out for a definitive ruling.

These circumstances are not present here. We are perfectly capable of answering both questions ourselves, and there is no indication that courts are overrun with lawsuits raising the issue. Cyberspace will not implode if the supreme court confronts the majority’s questions at some point in the future rather than today; the issues may well be sharpened by common law development in the meantime.¹

The Certified Questions

By long common law tradition, those who give away the property of others do so at their peril.² This case is not about “regulation of the Internet under state law,” as the majority believes, see Order at 1040. It’s about general principles of tort law that happen to apply to the Internet because that’s the type of property that NSI gave away. Kremen asks only that we afford him the same remedial rights that California law gives other property holders.³

The majority poses two questions. They are of middling difficulty at best; neither merits certification. The California Supreme Court long ago answered the first in Kremen’s favor; that precedent alone resolves the dispute before us. The answer to the second question — which we don’t even need to reach — is equally obvious.

I. Conversion of Intangibles

The majority’s first question is whether, “[f]or the tort of conversion to apply to intangible property, [it is] necessary that the intangible property be merged with a document or other tangible medium.” Order at 1038. The quaintness of the question, couched in language more reminiscent of postillions than POP servers, gives a pretty good clue that the majority is disinterring legal arcana long since laid to rest. Conversion originated in the fifteenth century as a remedy against one who found a plaintiffs lost goods and put them to his own use. See Prosser and Keeton on the Law of Torts § 15, at 89 (W. Page Keeton ed., 5th ed.1984); J.B. Ames, The History of Trover, 11 Harv. L.Rev. 277, 277 (1897). Because of this pedigree, the tort “became encrusted ... with legal rules that assumed that the property taken was tangible” — the sort of thing that one could “find in a field and later sell at a market.” Val D. Ricks, The Conversion of Intangible Property: Bursting the Ancient Trover Bottle with New Wine, 1991 B.Y.U. L.Rev. 1681, 1685. This limitation was harmless enough when people’s worldly goods consisted of livestock and farm tools, but today it’s a relic.

Our first question, then, is whether California still clings to the dated distinction between tangibles and intangibles. Some states do, albeit with significant ad hoc exceptions to accommodate commercial reality. The Restatement extends conversion only to intangible rights “merged” in a document. Restatement (Second) of Torts § 242;⁴ see also cases cited in Ricks, supra, at 1689 n. 26. Many courts, however, have rejected this distinction altogether and extend conversion to intangibles without regard to the Restatement’s test.⁵

The California Supreme Court made quick work of the question more than a century ago in Payne v. Elliot, 54 Cal. 339, 1880 WL 1907 (1880), and has not seen fit to revisit it. Defendants in Payne were trustees accused of appropriating plaintiffs securities. They argued for dismissal on the ground that plaintiff had alleged only conversion of the shares, not conversion of the share certificates. The court rejected the argument: “[T]he action no longer exists as it did at common law, but has been developed into a remedy for the conversion of every species of personal property.” Id. at 341 (emphasis added). One can almost hear Justice McKee chortle at counsel’s hapless argument as he elegantly disposes of this common law anachronism.

The majority today quotes this passage from Payne and ventures that “[t]he question then arises as to the scope of coverage for intangible property.” Order at 1041. I would have thought this was precisely the question that Payne answered, and that its response was “every species.” “Personal property,” after all, is “[a]ny movable or intangible thing that is subject to ownership and not classified as real property.” Black’s Law Dictionary 1233 (7th ed.1999) (emphasis added). Sex.com, we all agree, is property subject to ownership — namely, Kremen’s. And, obviously, it's not real estate.⁶ Payne therefore squarely controls.

Courts applying California law have followed in Payne’s tracks, recognizing conversion of intangible property without inquiring whether it was merged in a document. A & M Records, Inc. v. Heilman, 75 Cal.App.3d 554, 142 Cal.Rptr. 390 (1977), applied conversion to a defendant who sold unauthorized recordings, holding broadly that “such misappropriation and sale of the intangible property of another without authority from the owner is conversion.” Id. at 570, 142 Cal.Rptr. 390. In G.S. Rasmussen & Associates, Inc. v. Kalitta Flying Service, Inc., 958 F.2d 896 (9th Cir.1992), we relied on A & M Records and held that unauthorized use of someone else’s regulatory approval is conversion — again without asking whether the right in question was merged in a document. Id. at 906-07. In FMC Corp. v. Capital Cities/ABC, Inc., 915 F.2d 300 (7th Cir.1990), the Seventh Circuit, applying California law, explicitly rejected the tangible-intangible distinction. Id. at 302, 304-05. It quoted with approval Prosser and Keeton’s observation that “ ‘there is perhaps no very valid and essential reason why there might not be conversion’ of intangible property.” Id. at 305 (quoting Prosser and Keeton on the Law of Torts, supra, § 15, at 92).

Most telling, however, is our own recent decision in Bancroft & Masters, Inc. v. Augusta National Inc., 223 F.3d 1082 (9th Cir.2000). Not only did we recognize conversion of intangibles; we recognized conversion of the very intangible at issue here' — a domain name. A majority of the panel held that conversion of a domain name is “tortious conduct” under California law. Id. at 1089 (Sneed & Trott, JJ., concurring). And, yet again, we made no reference to any requirement that the property be merged in a document. If this is an issue worth pestering the California Supreme Court about today, why didn’t we certify it two years ago? Nothing has changed in California conversion law since then.

As the majority notes, Payne has not been universally followed. In Olschewski v. Hudson, 87 Cal.App. 282, 262 P. 43 (1927), California’s intermediate appellate court declared (with remarkable audacity) that Payne didn’t really mean what it said: that the “every species” language — the very ratio decidendi of the case — was “too broad a statement as to the application of the doctrine of conversion.” Id. at 288, 262 P. 43. It adopted instead the Restatement-like rationale that stock is subject to conversion only because it is “represented by” tangible documents. Id. Another case, Adkins v. Model Laundry Co., 92 Cal.App. 575, 268 P. 939 (1928), followed Olschewski on similar facts.

If Olschewski and Adkins had pointed to intervening cases where the California Supreme Court had retreated from Payne, they might give us pause. But they did nothing of the sort; they simply refused to apply controlling precedent by incorrectly labeling it dicta. We are bound by the pronouncement of the state’s highest court unless there are convincing reasons to believe that it would no longer adhere to its earlier rationale. Olschewski and Adkins — like the majority in our case — offer nothing to suggest it would not. “Certification is inappropriate when ... the supreme court of a state has already ruled and its decision is unambiguous.” United States v. Pend Oreille Pub. Util. Dist. No. 1, 926 F.2d 1502, 1506 n. 3 (9th Cir.1991); see also Estate of Madsen v. Comm’r, 659 F.2d 897, 901 (9th Cir.1981) (Norris, J., dissenting from certification order).

Searching for another reason to doubt Payne meant what it said, the majority invokes the Restatement. Our order implies that, unless a state explicitly disclaims the Restatement’s rules, we’ll assume that any deviations from that canonical text must have been an oversight. Poor Justice Brandéis, whose fifty state laboratories have been amalgamated into a single research park run by the American Law Institute.⁷

The majority identifies no convincing reason to believe that the California Supreme Court would overrule Payne. Speculation that a state supreme court might revisit a 123-year-old precedent and mail its tort jurisprudence back to the dark ages is not a ground for certification. If it were, we would certify nearly every diversity case we hear.

II. The “Merged in a Document” Test

None of this matters anyhow, because Kremen wins even under the Restatement. The majority’s analysis on this point is lacking. It cites a handful of state decisions, observes that none involves a domain name and proclaims our interpretive faculties exhausted. This is not a frugal use of the privilege the California Supreme Court affords us. Certification is for resolving true uncertainties in state law; it presupposes that we’ve made a diligent effort to apply the traditional judicial tools of analogical reasoning.

Kremen can sue for conversion under the Restatement because his domain name is in fact merged in a document, and NSI frustrated his use of it. See Restatement (Second) of Torts § 242(2) (“One who effectively prevents the exercise of intangible rights of the kind customarily merged in a document is subject to liability....”). His intangible property is (among other things) the right to have people who type “www.sex.com” into their web browsers sent to his website. It is, in standard Geek, the right to have the second-level .com domain “sex” associated with Ms IP address in NSI’s .com registry.⁸ The majority doesn’t dispute, nor could it, that NSI prevented Kremen’s use of his property when it handed sex.com over to Cohen. The only question under the Restatement is whether Kremen’s property is merged in NSI’s .com registry.⁹

It most certainly is. NSI’s registry is the master list that associates .com domains like “sex” with particular IP addresses. It’s essentially a ledger with domains in one column and IP addresses in another. See Thomas, 176 F.3d at 505. The fact that the ledger is electronic rather than ink-and-paper doesn’t make it any less a document (this dissent is still a document even if you’re reading it online). See Thrifty-Tel, 46 Cal.App.4th at 1565, 54 Cal.Rptr.2d 468; cf. eBay, Inc. v. Bidder’s Edge, Inc., 100 F.Supp.2d 1058, 1069 (N.D.Cal.2000). Web browsers determine what server is associated with a particular .com domain from the information in NSI’s registry.¹⁰ Modify an IP address in the registry, and you change the server with which a domain is associated; someone expecting to find my web page will be sent to yours instead. It’s hard to imagine an intangible right that’s more closely merged in a document.

There are several analogues to NSI’s .com registry in the case law, but the closest is a corporate share register. A share register qualifies as a document in which shares are merged. See Phansalkar v. Andersen Weinroth & Co., 175 F.Supp.2d 635, 641-42 (S.D.N.Y.2001) (deriving this point from Payne); ef. Payne, 54 Cal. at 342 (“[T]he certifícate is only evidence of the property; and it is not the only evidence, for a transfer on the books of the corporation, without the issuance of a certificate, vests title in the shareholder: the certificate is, therefore, but additional evidence of title.... ”).¹¹ The relationship between a share and a share register is quite similar to that between a .com domain and NSI’s .com registry. Both documents are databases whose entries identify who gets the benefits of a particular intangible right — which shareholder gets dividends; whose computer gets Internet traffic.¹²

California courts have long held that a shareholder can sue a corporation for conversion if it wrongfully refuses to transfer title to shares on its books. Ralston v. Bank of Cal., 112 Cal. 208, 213, 44 P. 476 (1896). A corporation that actually gives away a shareholder’s stock by wrongfully amending its share register is similarly liable. See 5 B.E. Witkin, Summary of California Law § 621, at 716 (9th ed.1988). An owner of a domain registry who wrongfully gives away a registrant’s domain name is in precisely the same position.

Cases where a defendant is held liable for converting a document are also instructive. Plunkett-Jarrell Grocery Co. v. Terry, 222 Ark. 784, 263 S.W.2d 229 (1953), for example, cited with approval in the Restatement, see Restatement (Second) of Torts § 242 cmt. b & reporter’s notes, involved a defendant who took the plaintiffs account book, preventing him from collecting his receivables. Plunkett-Jarrell Grocery Co., 263 S.W.2d at 233-34. The court allowed damages for the lost accounts, not merely for the paper they were recorded on. Id. at 234; see also Pioneer Commercial Funding Corp. v. United Airlines, Inc., 122 B.R. 871, 884-85 (S.D.N.Y.1991). Under the Restatement, an account receivable is thus merged in an account book. Kremen’s domain name and NSI’s .corn registry are at least as closely related as an account receivable and an account book. In both cases, the document is merely a list of intangibles that’s instrumental to enjoyment of the plaintiffs rights.¹³

One can imagine arguments against recognizing conversion of all intangibles, but none applies to domain names. Some intangibles are vaguely defined and may not give the defendant fair notice of the plaintiffs property right. See Olschewski, 87 Cal.App. at 288, 262 P. 43 (alluding to the need for “evidence of a definite interest”). But domain names, like corporate stock, are clear and discrete property rights. One who alters title to a registered domain name is fairly on notice that he may be affecting someone else’s property-

A second difficulty arises when the property is a “nonexclusive” intangible like a trade secret, the theft of which does not actually prevent the plaintiffs use. See Ricks, supra, at 1705-07 & n. 100. But domain names are exclusive intangibles (again, just like corporate stock). A defendant who wrongfully takes a domain name deprives the plaintiff of its use entirely. Domain names are much more like corporate stock or accounts receivable than they are like customer goodwill or trade secrets — intangibles that many courts applying the Restatement have declined to protect, see Restatement (Second) of Torts § 242 cmt. f & reporter’s notes.

These considerations merely reinforce a conclusion that the case law compels. We don’t need the California Supreme Court to spell out the inevitable consequences of the state’s jurisprudence every time a new species of property emerges. California law, even narrowly construed, recognizes conversion of property that shares all the relevant features of domain names. That’s all we need to know to decide the case.

Comity and Responsibility

Although the great majority of states— including all those in our circuit — now have certification procedures, California came to the process late. It adopted its rule less than five years ago, and only after endless carping from the bar. See, e.g., Jerome I. Braun, A Certification Rule for California, 36 Santa Clara L.Rev. 935 (1996). Even then, California adopted a rule much narrower than those of other states in our circuit. Other states permit certification from any federal court, but California accepts only questions certified by a court of appeals or the United States Supreme Court. Compare Cal. Rules of Court 29.5(a) with, e.g., Ariz. Sup.Ct. R. 27(a)(1). The California Supreme Court has also been more parsimonious in accepting certified questions. It’s practically unheard of for a supreme court of another state to reject a certified question of our court.¹⁴ California, though, has rejected one-third of the cases we’ve certified to it since the rule went into effect. See Appendix tbl.l.

The California Supreme Court’s evident ambivalence toward the certification process reflects the brutal realities of supervising the judiciary of the most populous state in the nation. Congestion in the California Supreme Court has been a fixture ever since the state was admitted to the Union. See Karl Manheim, The Business of the California Supreme Court: A Comparative Study, 26 Loy. L.A. L.Rev. 1085, 1092 (1993). The court delivers about 100 written opinions per year— twenty-five percent more than the United States Supreme Court, with two fewer justices. Compare Judicial Council of Cal., 2002 Annual Report: Court Statistics Report 9 tbl.6 [hereinafter Court Statistics] (103 opinions for 2000-2001), with Admin. Office of the U.S. Courts, 2001 Judicial Business 73 tbl.A-1 [hereinafter Judicial Business] (83 opinions for 2000-2001). Overall, the court disposes of some 9000 cases per year, up more than sixty percent from ten years ago, and once again more than the corresponding figure for the United States Supreme Court. Compare Court Statistics, supra, at 4 tbl.l (9047 dispositions for 2000-2001, compared to 5466 for 1991-1992), with Judicial Business, supra, at 73 tbl.A-1 (7762 dispositions for 2000-2001).

The seven justices of the California Supreme Court sit atop a judiciary with approximately 100 justices in the courts of appeal and 2000 judicial officers in the superior courts. Court Statistics, supra, at 18 tbl.l (96.8 justice full time equivalents); id. at 39 tbl.l (1998.0 judge, commissioner and referee FTEs). By comparison, Washington, the next most populous state in our circuit, has a nine-member supreme court but only thirty judicial officers in the courts of appeals and only about 400 in the superior, district and municipal courts combined. See Washington Courts, at http://www.courts.wa.gov/courts/ (last visited Oct. 15, 2002) (court of appeals directories listing 22 judges and 8 commissioners); Superior Court 2001 Annual Caseload Report 31 tbl., at http://www.courts.wa.gov/caseload/ (175 judges plus 49 commissioner FTEs); Courts of Limited Jurisdiction 2001 Annual Caseload Report 43 tbl., at http://www.courts.wa.gov/caseload/ (151 judge FTEs plus 31 commissioner FTEs).

The California Supreme Court is further hamstrung by its mandatory death penalty jurisdiction; it reviews an automatic direct appeal from every case where a death sentence is imposed. See Court Statistics, supra, at 4 tbl.l. This is a daunting prospect, with California’s death row now numbering 600 and still growing. See Gerald F. Uelmen, Courtly Manners, Cal. Law., July 2001, at 37, 74.¹⁵ As we know from our own experience, capital cases — often raising dozens of issues — are far more burdensome than most. The supreme court’s death penalty docket has at times strained its ability to act as the “architect of California case law.” Gerald F. Uelmen, The Lucas Court: A First Year Report Card, Cal. Law., June 1988, at 30, 31; see also Stephen R. Barnett, California Justice, 78 Cal. L.Rev. 247, 251 (1990) (book review).

Even if the California Supreme Court turned down our certification request, we would still have taken up a disproportionate amount of the court’s time and attention. Our requests are doubtless given far closer scrutiny than the run-of-the-mill petition of an ordinary litigant. The supreme court has turned down a significant number of our certified cases, but this cannot have been an easy or pleasant task, and it would surely grant review if at all possible. We should avoid forcing the court to make the awkward choice between agreeing to answer a question that really doesn’t deserve its attention and telling us we’re out to lunch. In some instances, all reasonable minds will agree that the court’s intervention is needed. But here, no party or amicus has suggested certification, and even our own panel is divided. We surely have a poor candidate for imposing on the California Supreme Court’s goodwill.¹⁶

The crowded California docket also means that certification is a less efficient mechanism for ascertaining state law. The cases we send to the California Supreme Court are beset by the same delays that plague the rest of its caseload. The average length of time from certification order to decision is well over a year and a half. See Appendix tbl.l; cf. Stephen R. Barnett, Un-Rocket Docket, Cal. Law., May 2002, at 17, 17 (average time on docket for all non-death penalty cases of 543 days). This is approximately triple the national average computed by one comprehensive study from the 1980s. See Carroll Serón, Certifying Questions of State Law: Experience of Federal Judges 39 tbl.7 (Fed. Judicial Ctr., FJC Staff Paper 83-1, 1983) (six to seven months).¹⁷ One case was gone for more than two and a half years. See Vu v. Prudential Prop. & Cas. Ins. Co., 172 F.3d 725 (9th Cir.1999), certified question answered, 26 Cal.4th 1142, 113 Cal.Rptr.2d 70, 33 P.3d 487 (2001), conformed to answer, 291 F.3d 603 (9th Cir.2002). None has taken less than 500 days. See Appendix tbl.l. Even to deny a certification request takes on average close to three months. See id. Given the unique pressures facing the California Supreme Court, these statistics are easy to understand. Delays are an unavoidable consequence of this cross-jurisdictional procedure, but they are far longer when the state court is already overburdened with its own work.

Certification burdens litigants, who foot the bill while their lawyers reargue the controversy in a different forum. The parties will now file briefs in the California Supreme Court, explaining why it should or should not accept the certification request. Cal. Rules of Court 29.5(e)(1). Next, they will reply to each other’s briefs. Id. 29.5(e)(4). If the court accepts the request, the parties will file more briefs and replies, arguing the case on the merits. Id. 29.5(h)(1). Once the state supreme court sends the case back to us, the parties will no doubt want to argue some more over how we should interpret its response. These are the sorts of things that make lawyers rich but litigants understandably frustrated. The prospect is particularly troublesome in this case — Kre-men has already spent the past four years in litigation trying to get compensation for the profits he lost because of Cohen’s theft and NSI’s alleged bungling.

This is a cpmplex case, but not in a way that justifies certification. Whether NSI’s .com registry is a document in which intangible property rights are merged is a hard technical question, not a hard legal one. It’s a matter of coming up with the right analogy, and- that has more to do with understanding how the Internet works than with state property law. The relevant facts are not genuinely disputed, but it takes a close reading to reconcile the competing characterizations, and a familiarity with the underlying technology doesn’t hurt. We’re certifying this case to the California Supreme Court, not to the ghost of Jonathan Postel;¹⁸ as far as I know, the former has no unique expertise in the field of Internet architecture.

This case happens to be in federal court because the parties are from different states, but there is nothing inevitable about this party alignment. If the issue is as far-reaching as the majority believes, it will come up in state court soon enough. But I doubt the case really is that far-reaching. The facts date back to the Wild West days of domain name registration, when NSI had no written contracts with registrants. NSI changed that policy long ago. Domain name contracts are relevant because they provide for significant limits on liability and because they may affect the scope of the property right conferred. Future cases involving conversion of domain names will raise different questions. This decision will no doubt be relevant, but it won’t be dispositive.

The California Supreme Court is always free to overrule any decision we render on the subject. It may even benefit from the insights we are able to offer, just as it benefits from prior consideration by state court judges. In this sense, we are just like another state court of appeal. We do California no favors by asking its supreme court to solve our problems while we stand mutely by.

Appendix

Table 2. Cases Certified to the California Supreme Court

[[Image here]]

1 . The majority draws encouragement from Arizonans for Official English v. Arizona, 520 U.S. 43, 117 S.Ct. 1055, 137 L.Ed.2d 170 (1997). But Arizonans was a very different case from ours. The state law there had just been passed by popular initiative and had never been interpreted by the state appellate courts; indeed, the plaintiffs had rushed into federal court in an obvious effort to avoid that possibility. Id. at 49, 63 n. 18, 117 S.Ct. 1055. The state attorney general had urged certification, which would have afforded the state courts an opportunity to address a sensitive issue of state policy and potentially could have avoided a federal constitutional question. Id. at 75-80, 117 S.Ct. 1055. Here, there is no federal constitutional question lurking in the background, and not even the parties themselves — much less the state attorney general — have urged certification.

2 . Conversion is a strict liability tort, so NSI's negligence is not an issue. Nonetheless, NSI’s claim that it had no reason to question the authenticity of Cohen’s forged letter is too much to bear. Kremen originally registered sex.com to his d/b/a, Online Classifieds, Inc. (OCI), listing himself as the contact. The forged letter stated that OCI had fired Kre-men and that its board of directors had "decided to abandon the domain name sex.com” to Cohen, giving no explanation whatsoever for this singular generosity. NSI received the letter not from OCI but as an enclosure sent by Cohen; the letter explained, "Because [OCI] do[es] not have a direct connection to the internet, we request that [Cohen] notify the internet registration [sic] on our behalf....” That a company called "Online Classifieds” would have no Internet connection is beyond implausible. Yet NSI made no effort to contact Kremen before giving away the domain name. It’s a bit as if Judge Reinhardt sent a letter to the DMV saying, "Judge Kozinski wants you to transfer title to his Lamborghini to me — he’d write to you himself, but he’s out of stamps.”

3 .The majority's citation to Moore v. Regents of the University of California, 51 Cal.3d 120, 271 Cal.Rptr. 146, 793 P.2d 479 (1990), thus misses the mark. Moore warned against "creating] new tort duties,” id. at 146, 271 Cal.Rptr. 146, 793 P.2d 479, but only after rejecting the plaintiff's claim under settled law, id. at 136, 271 Cal.Rptr. 146, 793 P.2d 479. Moreover, the issue in Moore was whether the plaintiff had a property right at all, not whether he could sue for conversion to enforce a property right he concededly held. Id. at 136-37, 271 Cal.Rptr. 146, 793 P.2d 479.

4 . The Restatement provides:

(1) Where there is conversion of a document in which intangible rights are merged, the damages include the value of such rights.

(2) One who effectively prevents the exercise of intangible rights of the kind customarily merged in a document is subject to a liability similar to that for conversion, even though the document is not itself converted.

Restatement (Second) of Torts % 242.

5 . See, e.g., Grynberg Prod. Corp. v. British Gas, p.l.c., 817 F.Supp. 1338, 1348 (E.D.Tex.1993) (intangible contract rights); Quincy Cablesystems, Inc. v. Sully’s Bar, Inc., 650 F.Supp. 838, 848 (D.Mass.1986) (satellite TV signals); Charter Hosp. of Mobile, Inc. v. Weinberg, 558 So.2d 909, 910-12 (Ala.1990) (drug abuse treatment programs); Nat’l Sur. Corp. v. Applied Sys., Inc., 418 So.2d 847, 850 (Ala.1982) (software programs); In re Estate of Corbin, 391 So.2d 731, 732-33 & n. 1 (Fla.Dist.Ct. App.1980) (interests in a business venture, including goodwill); Northeast Bank of Lewiston & Auburn v. Murphy, 512 A.2d 344, 348 (Me.1986) (future rights to receive proceeds); Foremost Ins. Co. v. Allstate Ins. Co., 439 Mich. 378, 486 N.W.2d 600, 610 n. 3 (1992) (intangible lien interests); Miracle Boot Puller Co. v. Plastray Corp., 57 Mich.App. 443, 225 N.W.2d 800, 804 (1975) (patent rights), rev’d on other grounds after remand, 84 Mich.App. 118, 269 N.W.2d 496 (1978); Schnucks Twenty-Five, Inc. v. Bettendorf, 595 S.W.2d 279, 284-85 (Mo.Ct.App.1979) (trade names); Brown v. Meyer, 580 S.W.2d 533, 534 (Mo.Ct. App.1979) (exclusive newspaper distribution areas); Benaquista v. Hardesty & Assocs., 20 Pa. D. & C.2d 227, 229, 1960 WL 8370 (1959) (intellectual property); Evans v. Am. Stores Co., 3 Pa. D. & C.2d 160, 161, 1955 WL 5263 (1955) (intangible ideas); see also United States v. Drebin, 557 F.2d 1316, 1332 (9th Cir.1977) (holding that "conversion” under 18 U.S.C. § 2314 applies to intangibles).

6 . Reports to the contrary notwithstanding. See, e.g., Shannon Lafferty, Legal Battle for Sex.com Continues in Calif., Legal Intelligencer, Feb. 1, 2001, at 4 ("the most valuable piece of real estate on the Internet”).

7 . Thrifty-Tel, Inc. v. Bezenek, 46 Cal.App.4th 1559, 54 Cal.Rptr.2d 468 (1996), did not "favorably view[]" the Restatement’s approach, cf. Order at 1042; it explicitly left the issue unresolved. Thrifty-Tel, 46 Cal.App.4th at 1565-66, 54 Cal.Rptr.2d 468.

8 . The majority's order describes the basic function of the Domain Name System and the fact that it associates domain names with IP addresses — sets of numbers that uniquely identify each computer connected to the Internet. See Order at 1038-39. But it omits any discussion of how the DNS actually works beyond the observation that it is a "decentralized, albeit hierarchical, process.” Id. at 20 (quoting Am. Online, Inc. v. Huang, 106 F.Supp.2d 848, 851 (E.D.Va.2000)). The majority’s failure to analyze the DNS and the role that NSI’s .com registry plays within it helps explain why it can't detect a document in this case.

NSI’s .com registry, also known as the ".com zone file,” associates particular second-level .com domains like "sex” with particular IP addresses. If a browser wants to find the website "www.sex.com,” it goes through the following steps: It first looks in a "root” registry to find out who has the list of .com addresses — and that root registry says "NSI.” It then looks in NSI’s .com registry to find out who has the list of sex.com addresses — that registry (now) says "Gary Kremen.” Finally, it looks in Gary Kremen's sex.com registry to find out where the website www. sex.com is located. (These registries don’t literally say "NSI” and "Gary Kremen”; they list the IP addresses of their computers. But it's the same idea.) Thus, while the DNS as a whole is a “decentralized, albeit hierarchical, process,” NSI’s .com registry is not — it’s just a list of second-level .com domains and corresponding IP addresses; a document in any relevant sense of the word. See generally Brief of Amicus Curiae Electronic Frontier Foundation at 6-8; Thomas v. Network Solutions, Inc., 176 F.3d 500, 503-04 (D.C.Cir.1999); Improvement of Technical Management of Internet Names and Addresses, 63 Fed.Reg. 8826, 8826 (Feb. 20, 1998); Milton L. Mueller, Ruling the Root: Internet Governance and the Taming of Cyberspace 41-43, 46 fig.3.5, 194-96 (2002); InterNIC FAQs: The Domain Name System (Mar. 25, 2002), at http://www.internic.net/faqs/authoritative-dns. html.

9 . The majority bases almost its entire discussion on the mistaken assumption that the relevant document is the decentralized DNS as a whole. Order at 1042-43. The only part of the DNS at issue here is NSI's registry of .com domains — the .com zone file. NSI’s claim that "there is no master directory of domain names and IP addresses to which a computer refers” is true but irrelevant. Other registries list the second-level domains in .gov, .mil, .edu and the like; still others list the third-level domains within each second-level .com domain. But NSI’s registry of second-level .com domains is the only piece of the DNS that matters in this case because Kremen’s entry in this particular registry is what NSI gave away. NSI’s .com registry, not the DNS as a whole, is the document in which Kremen’s property is merged.

10 . NSI’s .com registry isn't actually consulted directly on every query, because other servers copy and store its information in order to speed up response times. If a browser wants to know an IP address, it may get it from a nearby server that previously copied the information from NSI. See Thomas, 176 F.3d at 503-04; Mueller, supra, at 48; InterNIC FAQs, supra. The fact that other servers copy information in NSI’s registry for ease of reference, however, doesn’t change the fact that NSI’s registry is the authoritative listing of .com domains. Thomas, 176 F.3d at 505; 63 Fed.Reg. at 8828; Mueller, supra, at 196. When. NSI changes an entry in its registry, other servers all over the world copy the updated information, typically within 24-48 hours. See Modifying Your Domain Name Record (2002), at http://www.net-sol.com/en_US/help/modify-dnr-06.jhtml.

11 . Payne did not so hold, of course; it imposed no merger requirement at all. It held that share registers and share certificates have the same evidentiary function. See Payne, 54 Cal. at 342. But the corollary for states that do follow the merger requirement is that these equivalent evidentiary functions imply an equivalent ability to satisfy the merger requirement — which is precisely what Phansalkar held, 175 F.Supp.2d at 641-42. Courts routinely recognize conversion of un-certificated shares. See, e.g., Haskell v. Middle States Petroleum Corp., 165 A. 562, 563 (Del.Super.Ct.1933); Connelly v. Estate of Dooley, 96 Ill.App.3d 1077, 52 Ill.Dec. 462, 422 N.E.2d 143, 147 (1981); Mahoney v. Walsh, 16 A.D. 601, 605-06, 44 N.Y.S. 969 (N.Y.App.Div.1897). Uncertificated shares by definition are not "customarily merged in” share certificates; the merging document must be something else — for example, the share register, see Phansalkar, 175 F.Supp.2d at 642.

12 . A share register identifies the person who owns the shares, while the .com zone file only identifies the address of the owner’s computer. Even if this difference mattered — which seems hard to believe — domain names are explicitly linked to their owners in another document, the "WHOIS database” maintained by the registrar (also NSI in this case). Kremen's WHOIS record for sex.com, for example, can be retrieved by typing "sex.com” into the web interface of NSI’s WHOIS server, currently located at http://www.net-sol.com/cgi-bin/whois/whois. NSI's WHOIS database seems to be yet another document in which Kremen's property is merged.

13 . In our case, the document itself wasn’t converted; the .com zone file remained in NSI’s hands throughout. But that means only that Kremen’s claim sounds in section 242(2) of the Restatement rather than section 242(1). Compare Restatement (Second) of Torts § 242(2) ("One who effectively prevents the exercise of intangible rights of the kind customarily merged in a document is subject to liability ...."), with id. § 242(1) ("Where there is conversion of a document in which intangible rights are merged, the damages include the value of such rights.”). "Merged” surely can't mean one thing in one section and something else in the other.

14 . Even when it's painfully obvious that we asked the wrong question. See, e.g., Scheehle v. Justices of the Supreme Court, 203 Ariz. 520, 57 P.3d 379 (2002).

15 . The only reason that the court's death penalty docket is at all manageable, apparently, is a parallel shortage of appointed counsel. Uelmen, Courtly Manners, supra, at 74.

16 . The majority declines to consider the burden we impose on the state supreme court when we force it to rule on our certification requests, opining that "it is not our role to pass advance judgment on the Court's priorities.” Order at 1038. By that logic, we would certify every case on our diversity docket — and every Federal Tort Claims Act case too — so that the state supreme court could have first crack at the state law issues presented. Surely, exercising some judgment about the productive use of the state court’s time and resources, not to mention those of the parties, is part of our responsibility in deciding whether to certify.

California's rule, moreover, requires that the certification order be signed by the presiding judge, making no allowance for a situation, like ours, where the presiding judge is in dissent. Cal. Rules of Court 29.5(d). This has led to the odd situation where I find myself dissenting from an order that bears my signature. Cf. Crocker Nat’l Bank v. Clark Equip. Credit Corp., 724 F.2d 696, 700 n. 3 (8th Cir.1984) (Lay, C.J.) ("As author of this opinion, I dissent.... ”). The drafters of California’s certification rule obviously did not anticipate that the presiding judge would dissent, which means they contemplated cases would only be certified when all panel members agree.

17 . Turnaround times for the next two largest states in our circuit are shorter. Washington typically returns cases to us in about nine months; Arizona usually takes about a year.

18 . Nor to A1 Gore, for that matter.