'The Cathedral' at twenty-five: citations and impressions.

• Author: Krier, James E.
• Position: Property Rules, Liability Rules, and Inalienability: A Twenty-Five Year Retrospective

It was twenty-five years ago that Guido Calabresi and Douglas Melamed published their article on property rules, liability rules, and inalienability.(1) Calabresi, then a law professor, later a dean, is now a federal judge. Melamed, formerly a student of Calabresi's, is now a seasoned Washington attorney. Their article -- which, thanks to its subtitle, we shall call The Cathedral -- has had a remarkable influence on our own thinking, as we tried to show in a recent paper.(2)

This is not the place to rehash what we said then, but a summary might be in order. First, we demonstrated that the conventional wisdom about liability (damage) rules, that judges should use them when transaction costs are high, is incorrect, because the costs of assessing damages might in fact be higher still; if they are, property (injunction) rules are superior, at least from the standpoint of efficiency. Second, and relatedly, we identified problems of correlation and synergy that come into play as one tries to choose between damages and injunctive relief. Correlation problems arise because the same considerations that yield high transaction costs usually yield high assessment costs as well; synergy problems arise because the use of damage rules can inhibit the development of more effective bargaining practices. Third, we showed that Calabresi and Melamed's celebrated Rule 4 (reverse damages) contains a paradox, which we went on to resolve by inventing reverse-reverse damages (the "double reverse twist"). The trick of the double reverse twist relates to our fourth point, having to do with a "best-chooser axiom" which can be used to illuminate matters of institutional (not just judicial) design generally. Finally, we suggested in conclusion the relationship of much of the foregoing to relevant literature in other disciplines.

For now, we have nothing to add to these thoughts of ours, so we turn to the influence of Calabresi and Melamed on others. The Cathedral is undoubtedly a classic, as the occasion for this Essay suggests. But how so?

1. Citation Analysis

   Our inquiry relies in part on "citation analysis," a term of art. The purpose of this art is to determine the influence of particular scholars or scholarly products on the minds of others writing in a field; the practice of the art depends on the fact that researchers make reference in their publications to work by predecessors, thereby generating citations that can be collected, counted, and otherwise analyzed. What links practice to purpose (arguably a weak link, for reasons to be considered) is the notion that the material cited by the researchers "in their own papers represents a roughly valid indicator of influence on their work,"(3) such that citations can be taken as an indication of the "quality or impact" of scholarly contributions, among other things.(4) Frequency of citation (as opposed to, say, place of citation or nature of citation) is generally the key, and quality or impact is taken to vary directly with it, in both absolute and relative terms. Basically, it is good to be cited a lot, or at least a lot more than others.

   Citation analysis is most widely practiced in scientific communities, including both the hard and the social sciences. It is far less common as a measure of scholarly influence in the legal world, yet it is in connection with law (though not academic writing in law) that it originated about a century ago. We have learned this and much else from the work of Fred Shapiro, a librarian and lecturer at the Yale Law School and the only person who has demonstrated a sustained interest in modern citation studies of legal scholarship. His periodic reports, which continue to date,(5) provide a good primer on the subject, as does Richard Posner's recent study of Benjamin Cardozo.(6)

   As already suggested, the crucial assumption of citation analysis is "that the number of times a scholarly work is cited is a proxy for the influence or importance of the work."(7) Sometimes, however, it is not, as a few examples suggest. Survey articles are regularly cited but they are usually derivative, convenient as opposed to important. In contrast, very important works are vulnerable to "obliteration," their contributions becoming so much apart of the common understanding in a field that citations become superfluous and would even look silly.(8) Moreover, recent publications might rack up higher citation counts than older ones because the new work is more accessible or otherwise salient; on the other hand, youth is a disadvantage when total citations over the years are being tabulated. Famous scholars might be overcited, out of deference or because they and their work come most readily to mind. Some authors who publish a lot publish a lot of junk, yet they could enjoy high counts because their work is routinely criticized. Relatedly, prolific scholars given to citing themselves can become legends in their own footnotes.

   This is not a complete catalog of the vices of citation analysis, but it is suggestive, sufficient, and less damning than might at first appear. Regarding obliteration, for example, Shapiro rightly says, "[A]ny work so successful as to achieve this status would have already amassed a [sic] impressive citation total before becoming `obliterated,' and would still rank near the top of the list."(9) As to critical citations, we agree with Posner that "[n]egligible work is more likely to be ignored than to be criticized in print; and work that is heavily criticized, even work decisively shown to be erroneous, plays a vital role in the growth of knowledge."(10) And perhaps prolificacy should be its own reward, for it might be the only reward.

   "Citation rates are as good a measure for influence," says Arthur Jacobson, "as suicide rates are for anomie, though both citation and suicide have personal sides to them as well."(11) Citation analysis is understood to be an imperfect proxy, but usually there is no superior (or equally wieldy) alternative. Empirical studies demonstrate a high correlation between citation counts and peer judgments,(12) and our assessment of The Cathedral relies (eventually) on more than mere numbers in any event. Whatever the state of the art, our concern is the status of the article.

2. Citations to The Cathedral

   But we begin with mere numbers, some of them borrowed, with thanks, from Fred Shapiro's compilations of the most-cited law review articles, including his most recently published survey, an update in late 1996(13) of his original 1985 study.(14) The original looked back only to articles published since 1947, lacked data from some relevant publications, and limited its most-cited list to the top fifty. The update, on the other hand, has no chronological restrictions on articles surveyed, includes interdisciplinary journals that previously had been excluded, and aims to identify "the one hundred most-cited legal articles of all time, that is, most often cited within other articles."(15)

   Because Shapiro's full-blown tabulations are readily available, there is little point in our reproducing them in their entirety here. It is interesting, however, to look at how the top thirty articles on Shapiro's two lists (1985, 1996) changed over the eleven years between them. As can be seen in Appendix I, no general pattern emerges. Some very well-known works have suffered rather marked declines since 1985,(16) others have held pretty steady,(17) and several have moved up significantly on the list.(18) The Cathedral is among the latter, rising from its 1985 position in the middle of the pack to a place just shy of the top ten in the expanded 1996 list.

   Shapiro's 1996 rankings are based on data from the Social Science Citation Index (SSCI). Out of curiosity, and as a check, we searched another source of citation information, the Westlaw Journals and Law Review (JLR) database. The JLR database has several advantages for our purposes. First, it is more familiar to lawyers than SSCI, which makes it easier for skeptics to replicate our searches should they wish to do so.(19) Second, JLR is tailored to law, whereas SSCI is a huge international multidisciplinary citation index, designed to cover the most important social science journals worldwide (some 1500 periodicals covering everything from anthropology to urban planning), plus selected articles from another 2400 journals in the natural and physical sciences.(20) SSCI does...