Conclusion

Abstract

Holmes, in tune with the “scientism” prevalent in his time, speculated that science might reveal rules about society and human relations, and that these rules might come to replace the legal system—that society might be regulated using a rules-based algorithmic formalism. He had a long standing interest in statistics (“the man of the future is the man of statistics” he wrote), and perhaps he saw the recently-developed Hollerith Tabulator, the forerunner of automated data processing, as a step towards an age of scientific governance led by statistics. It is ironic, then, that computing is now taking a path that has more in common with Holmes’s description of the law as it was, than with where he thought it might go.

In every department of knowledge, what wonderful things have been done to solicit the interest and to stir the hopes of new inquirers! Every thing is interesting when you understand it, when you see its connection with other things, and, in the end, with all other things.

Oliver Wendell Holmes, Jr., Remarks at the Harvard Commencement Dinner (June 22, 1880)¹

This book has posited a connection between two seemingly remote things: between a late nineteenth century revolution in law and an early twenty-first century revolution in computing. The jurist on whose work we’ve drawn, if he’d been transported to the present day, we think would have been open to the connection. Holmes’s early milieu had been one of science, medicine, and letters, these being fields in which his father had held a prominent place and in which their city, in Holmes’s youth, in America had held the preeminent place.² Leading lights of nineteenth century philosophy and science numbered among Holmes’s friends and interlocutors at home and abroad in the years immediately after the Civil War. Holmes continued throughout his life to engage with people whom today we would call technologists. His interest in statistics and in the natural sciences was broad and deep and visible in Holmes’s vast output as a scholar and a judge. Lawyering and judging, to Holmes, were jobs but also objects to be searched for deeper understanding.

We hope that in the preceding chapters, by considering “its connection with other things,” we have contributed to a deeper understanding of where machine learning belongs in the wider currents of modern thought. The common current that has shaped both legal thought and computer science is probability . It is a strikingly modern concept. As we have recalled, its origins are not less recent than the mid-seventeenth century . Its impact has been felt in one field after another, though hardly all at once. Law was present at its origins, though it took over two centuries before a new jurisprudence would take shape under its influence. Computing, too, did not begin as an operation in probability and statistics, but now probability and statistics are the indispensable core of machine learning . Thus both law and computing have undergone a shift from their earlier grounding in deductive logic: they have taken an inductive turn , based on pattern finding and prediction.

But, to conclude, let us turn away from intellectual history and look instead to the future.

10.1 Holmes as Futurist

Holmes, notwithstanding the strains of fatalism evident in his words, was fascinated by the potential for change, in particular change as driven by science and technology. Speaking in 1895 in honor of C. C. Langdell, that leading expositor of legal formalism, Holmes stated with moderate confidence that a march was on toward a scientific basis for law and that it would continue: “The Italians have begun to work upon the notion that the foundations of the law ought to be scientific, and, if our civilization does not collapse, I feel pretty sure that the regiment or division that follows us will carry that flag.”³ With the reference to “[t]he Italians” Holmes seems to have had in mind the positivism that was prevalent in legal theory in late nineteenth century Italy⁴; to the possibility of civilizational collapse, the pessimism prevalent generally in European philosophy at the time. In 1897, in Law in Science and Science in Law , Holmes hedged his prediction, but he continued to see contemporary advances in science and technology as pertinent to the organization of public life in the widest sense: “Very likely it may be that with all the help that statistics and every modern appliance can bring us there never will be a commonwealth in which science is everywhere supreme.”⁵ To entertain the possibility of a technological supremacy arising over law , even if to doubt that a scientific revolution in law would ever be complete, was still to place the matter in high relief.

Technological change had affected society at large for generations by the time Holmes wrote Law in Science and Science in Law . However, advances were accelerating and, moreover, in specific technical domains technology was interweaving itself with public order in unprecedented ways. This was the decade in which the U.S. Census Bureau first used a punch card machine with electric circuits to process data. The machine, known as the Hollerith Tabulator after its inventor, Herman Hollerith,⁶ was the forerunner of modern data processing. Hollerith’s company, the Tabulating Machine Company, was one of several later amalgamated to form the company that was eventually re-named IBM. The basic concept of the machine remained the cornerstone of data processing until the 1950s.⁷ By 1911 (when Hollerith sold the Tabulating Machine Company), Hollerith Tabulators already had been used to process census data in the United States, United Kingdom, Norway, Denmark, Canada, and the Austrian and Russian Empires. Railroads, insurance companies, department stores, pharmaceutical companies, and manufacturers employed Hollerith machines as well.⁸ The Hollerith Tabulator lowered the cost of handling large quantities of data and accelerated the work; the Scientific American, which ran an article on the machine in its August 30, 1890 edition, attributed the “early completion of the [census] count… to the improved appliances by which it was executed.”⁹

The Hollerith machines did more than increase the efficiency of the performance of existing tasks, however. Because they enabled users to interrogate datasets in ways that earlier were prohibitively time-consuming—for example, asking how many people in the year 1900 in Cincinnati were male blacksmiths born in Italy—the Hollerith machines opened the door to new uses for data, not just more efficient head counts. The Scientific American referred to the “elasticity of function” that the machines enabled.¹⁰ Hollerith himself was referred to as the first “statistical engineer.”¹¹

Holmes was not excited about putting his hands on the various innovations that technologists were bringing to the market; he doubted that his house would have had electricity or a telephone if his wife and not had them installed.¹² It would be surprising, however, if Holmes had not known of the Hollerith machine.¹³ The edition of Scientific American containing the article about Hollerith and the census featured illustrations of the tabulator at work on its cover. The same periodical had run an article four years earlier on Holmes’s father.¹⁴ Holmes was a paid subscriber.¹⁵ He also encountered technological innovations in the course of his principal employment: Holmes authored a number of the Supreme Court’s decisions in patent matters¹⁶ (none, it seems, concerning Hollerith, though the “statistical engineer” was no stranger to intellectual property disputes¹⁷). Curiously enough, Hollerith’s headquarters and workshop were in a building in the Georgetown part of Washington, DC not many blocks from where Holmes lived after moving to the capital,¹⁸ and the building in which the Census employed a large array of the machines was a short block off the most direct route (2.2 miles) between the Capitol (which then housed the Supreme Court) and Holmes’s house at 1720 I Street, NW. Contemporaries remarked on the distinctive chimes that bells on the machines made, a noise which rose to a clamor in the building and which could be heard on the street below.¹⁹ Holmes was a keen rambler whose peregrinations in Boston, Washington, and elsewhere took him in pursuit of interesting things.²⁰ Whether or not the Hollerith Tabulator was the appliance Holmes had in mind in Law in Science and Science in Law , technology was in the air. The emergence of modern bureaucracy in the early nineteenth century had been associated with an ambition to put public governance on a scientific basis;²¹ the emergence of machines in the late nineteenth century that process data inspired new confidence that such an ambition was achievable.²² To associate the commonwealth and its governance with statistics and “modern appliance,” as Holmes did, was very much of a piece with the age.

Holmes’s interest in technology induced him to maintain wide-ranging contacts, some of them rather idiosyncratic. A fringe figure named Franklin Ford²³ for a number of years corresponded with Holmes about the former’s theories regarding news media and credit institutions. Ford imagined a centralized clearing mechanism that would give universal access to all news and credit information, an idea today weirdly evocative of the world wide web; and he said that this mechanism would supplant the state and its legal institutions, a prediction likewise evocative of futurists today who say, e.g., the blockchain will bring about the end of currencies issued under government fiat. Holmes continued the correspondence for years, telling Ford at one point that he (Ford) was “engaged with the large problems of the sociologist, by whom all social forces are equally to be considered and who, of course, may find and will find forces and necessities more potent than the theoretical omnipotence of the technical lawgiver.”²⁴ Holmes evidently continued to speculate that law might in time give way to the experience embodied in “all social forces” which, in the context of that correspondence, suggested “social forces” mediated in some way by technology. In his correspondence with Franklin Ford, whose schemes aimed at the dissemination and use of data, Holmes seemed to intuit that, if machines came to martial data in even “more potent” ways, civilization-changing effects might follow.

In a flight of fancy, in a 1913 speech, Holmes went so far as to speculate about the evolution of the species:

I think it not improbable that man, like the grub that prepares a chamber for the winged thing it never has seen but is to be—that man may have cosmic destinies that he does not understand… I was walking homeward on Pennsylvania Avenue near the Treasury, and as I looked beyond Sherman’s Statue to the west the sky was aflame with scarlet and crimson from the setting sun. But, like the note of downfall in Wagner’s opera, below the sky line there came from little globes the pallid discord of the electric lights. And I thought to myself the Götterdämmerung will end, and from those globes clustered like evil eggs will come the new masters of the sky. It is like the time in which we live. But then I remembered the faith that I partly have expressed, faith in a universe not measured by our fears, a universe that has thought and more than thought inside of it, and as I gazed, after the sunset and above the electric lights there shone the stars.²⁵

Holmes in this passage holds his own with the most imaginative—and the most foreboding—twenty-first century transhumanists. The operatic reference, with a little stretch, is even more evocative of change wrought by science than first appears. True, the characters in Wagner’s opera don’t use electric circuits for data processing.²⁶ But it is not too foreign to Holmes’s speculations about the world-changing potential of statistics—or to the conceptual foundations of the machine learning age—that, in the Prologue to that last of the Ring Cycle operas, the Fates, whose vocation is to give prophecies, are weaving: and the rope with which they weave is made of the knowledge of all things past, present, and yet to come. The rope breaks, and thus the stage is set for the end of one world and the start of another.²⁷ Mythological data scientists foretelling the epochal changes their science will soon effect!

In less fanciful tenor, in Law in Science Holmes suggested that science might aid law and possibly replace it:

I have had in mind an ultimate dependence upon science because it is finally for science to determine, so far as it can, the relative worth of our different social ends, and, as I have tried to hint, it is our estimate of the proportion between these, now often blind and unconscious, that leads us to insist upon and to enlarge the sphere of one [legal] principle and to allow another gradually to dwindle into atrophy.²⁸

Science, in Holmes’s view, would be put in harness to law; or it would replace law by taking over the social functions that law for the time being serves. Richard Posner, Chief Judge of the U.S. Court of Appeals for the Seventh Circuit at the time, on the 100th anniversary of The Path of the Law read Holmes to contemplate that law would be “succeeded at some time in the future by forms of social control that perform the essential functions of law but are not law in a recognizable sense.”²⁹ The most accomplished scholar of Holmes to have served on an American court in the present century, Posner also thought the passage about Götterdämmerung and “cosmic destinies” noteworthy.³⁰ Whatever the precise role Holmes contemplated for science, and wherever he thought science would take us, it is evident that Holmes’s philosophy did not equate with narrow presentism. Holmes was keenly interested in the future, including the future impact of science on law.

Writers have cautioned against “scientism,”³¹ the unjustified confidence in the potential for science to solve society’s problems. Our focus here has not been to repeat well-known critiques of unexamined enthusiasm for technological change. The acknowledgement of correlation between new technologies and risk has tempered scientistic impulses³²; admonitions have been sounded in regard to Holmes’s ideas about science.³³ It nevertheless is timely to alert practitioners of computer science that they ignore sanguinary lessons of the history of ideas if they place blind faith in the power of their craft. Holmes, perhaps, can be read for cautionary notes in that regard.

But our chief purpose in this book has been to use the analogy from Holmes’s jurisprudence to cast light on machine learning. Let us ask, then, what, if any, lessons for the future of computer science might be found in Holmes’s speculations about the future of law.

10.2 Where Did Holmes Think Law Was Going, and Might Computer Science Follow?

Holmes, in thinking about law, found interest in wider currents that law both is borne upon and drives. Holmes considered the possibility that science will replace law—more precisely, that scientific method and technological advances will reveal rules and principles that law will adopt and thus give law a more reliable foundation. A curious irony would be if it went the other way around. A self-referential system—prediction as the system’s output and its input as well—which is to say Holmes’s concept of law as he thought law actually is—is what computer scientists, we speculate, might seek to make machine learning into. That has not been what machine learning is. True, machine learning has moved beyond logic and so is now an inductive process of finding patterns in input data to attain an output . So far however that is the end of the road. If machine learning goes further, if it comes to embody self-referential mechanisms such as Gödel and Turing devised in mathematics and computation, then machine learning will come to look even more like Holmes’s law—an inductive system of prediction-making and self-referential prediction-shaping. The law, as Holmes understood it, would then have foreshadowed the future of computer science.

This is not how Holmes seems to have imagined things would go. We discern in his futurist and scientistic vein that Holmes thought that law, as he understood it to be, would give way to something else. What he thought law as prediction would give way to is not clear, but, as Judge Posner suggested, Holmes seems to have contemplated that science and technology would end society’s reliance on law and bring about new mechanisms of control. The new mechanisms would be based on logic, rather than experience, and thus, in Holmes’s apparent vision, would come full circle back to a sort of formalism —not a formalism based on arbitrary doctrines and rules, but based, instead, on propositions derived from what nineteenth century thinkers conceived of as science.

Holmes’s speculations about science replacing law would seem to have a genealogy back to Leibniz, though we are not aware to what extent, if at all, Holmes was thinking about that antecedent when he wrote about a “scientific” future for law. Leibniz wrote about the possible use of mathematical models to describe law and philosophy in sufficient detail and scope that (in Leibniz’s words), “if controversies were to arise, there would be no more need of disputation between two philosophers than between two accountants. For it would suffice for them to take their pencils in their hands and to sit down at the abacus and say to each other (with a friend if they wish): Let us calculate.”³⁴ It is indeed this branch of Leibniz’s thought that interests people, like Michael Livermore, who are considering how to put state of the art computing to work on legal problems.³⁵ As we have suggested, however, it is Leibniz’s thinking about probability , not his speculation that fixed rules might one day answer legal questions, that has special resonance in a machine learning age. Leibniz thus, arguably, presaged Holmes , both in the application of probability theory to law and in the speculation that such application ultimately might be set aside in favor of a universal body of rules. What is more, he may have presaged Holmes, too, in thinking past that part of his thinking that has real salience to machine learning.

The irony, then, would be if computing followed Holmes’s description of law as he thought law is—not his speculations about where he thought law was going. Machine learning today finds itself on the path that Holmes understood law actually to traverse in his own day. Machine learning has shifted computer science from logical deduction to an inductive process of finding patterns in large bodies of data. Holmes’s realist conception of law shifted the law from rules-based formalism to a search for patterns in the collected experience of society. Reading Holmes as he understood the law to be, not his speculations about where law might go, we discern a path of the law that very much resembles that taken by machine learning so far.

Along that path, Holmes supplied a complete description of law. He described law as prophecy—meaning that all instances of law, all its expressions, are prophecy, and each successive prediction , whatever its formal source, in turn shapes, to a greater or to a lesser degree, all the prophecies to come. There is thus a self-referential character in law’s inputs and outputs. In such self-reference, the law perhaps even anticipates a way ahead for machine learning : experience supplies the input from which present decisions are reached; and, in turn, those outputs become the inputs for future decisions. In short, though Holmes might have been waiting for technology to inform law, it could turn out that it is law that informs technology. The lawyers might have something to teach the computer scientists.

10.3 Lessons for Lawyers and Other Laypeople

Through a reading of Gödel, Turing, and Holmes in Chapter 9, we’ve identified a self-referential path that machine learning might follow. Regardless of where the technology goes from here, however, it is already too important for laypeople, including lawyers, to ignore. Thus we recall our initial task: to explain machine learning in terms that convey its essentials to the non-specialist.

We have aimed in the chapters above to convey the essentials. We have done so with the aid and within the limits of an analogy between two revolutions—one in jurisprudence, one in computing. While a much wider audience needs to come to grips with machine learning, lawyers, in view of their function in society, find themselves involved in distinctive ways in the questions it presents. Coming at machine learning through a legal analogy hopefully has established some new connections which will help non-specialists in general. The connections likely will have particular salience for the lawyers.

Lawyers whether by inclination or by habit are conservative. Law has much to do with authority, and legal argument seldom wins praise for conspicuous innovation. Legal minds, moreover, are skeptical; the enthusiasm for new machines that enlivens a technologist is not prevalent among lawyers. And, yet, lawyers from time to time have been involved in revolutions. As we noted at the start of this exploration of the conceptual foundations of machine learning , probability theory —the common current on which the two revolutions addressed in the chapters above have been carried—owes much to thinkers who were educated in law. So, too, long after, influential ideas in law have come from lawyers whom science and technology have interested.

H.L.A. Hart, having dedicated his Inaugural Lecture in 1952 as Professor of Jurisprudence at Oxford to Definition and Theory in Jurisprudence,³⁶ a few years later addressed his Holmes Lecture at Harvard to the challenge that arises when the legal system is called on to classify the sorts of “wonderful things” that solicited Holmes’s interest time and again through his career. “Human invention and natural processes,” Hart wrote…

continually throw up such variants on the familiar, and if we are to say that these ranges of facts do or do not fall under existing rules, then the classifier must make a decision which is not dictated to him, for the facts and phenomena to which we fit our words and apply our rules are as it were dumb… Fact situations do not await us neatly labeled, creased, and folded, nor is their legal classification written on them to be simply read off by the judge.³⁷

The impact of machine learning, realized and anticipated, identifies it as a phenomenon that Hart would have recognized as requiring legal classification. Lawyers and judges are called upon to address it with what rules they already have to hand. New legislation has attempted to address it in fresh terms. Explainability, an objective that we considered above, has motivated a range of new legislation, such as the GDPR, which entered into force in 2018 in the EU.³⁸ Enactments elsewhere pursue a similar objective, such as the California Consumer Privacy Act (CCPA) which enters into force in 2020,³⁹ as do a great many more.⁴⁰ It is both too early and beyond the scope of the present short book to take stock of the legislative output . It is not too early to observe the need for a wider, and more intuitive, understanding of what the legislator is being called upon to address. Law makers and practitioners need to understand the shift from algorithms to machine learning if they are to make good law and to practice it effectively. To attempt to label, crease, and fold machine learning into a familiar, algorithmic form is a mistake that Hart would have cautioned us to avoid.

Interestingly enough, Turing, so influential a figure in the line of human ingenuity that has interested us in the preceding chapters, seems to have been not too far removed from Holmes. The proximity was via Hart. Held by some the foremost legal philosopher since Holmes, Hart called Holmes a “heroic figure in jurisprudence.”⁴¹ Hart addressed the earlier jurist’s idea of the law in detail, partly in riposte to Holmes’s critics.⁴² Hart did not write about Turing, but they were contemporaries—and linked. During World War II, which was before Hart embarked on a career as a legal academic, he was assigned to MI5, the British domestic intelligence agency. Hart’s responsibility was to lead the liaison unit between MI5 and project ULTRA, the latter having been under the jurisdiction of MI6, the external intelligence agency. It was under project ULTRA, at a country house at Bletchley Park in England, that Turing did his codebreaking and developed the computational strategies that provided the point of departure for modern computing. Turing’s work at Bletchley Park enabled MI6 to decipher encrypted German communications. So closely, however, did MI6 guard ULTRA that it was not clear at the start that the liaison unit for which Hart was responsible would serve any purpose. It appears that Hart’s personal relations with key people in ULTRA played a role in getting the liaison to function—and, thus, in helping assure that Turing’s technical achievements would add practical value to the war effort.⁴³ An eminent former student and colleague, John Finnis, notes that Hart never divulged further details about his wartime duties .⁴⁴ Years after the war—but still some time before Turing’s rise to general renown—Hart did mention Turing: he mentioned to family that he admired him very much.⁴⁵

That Turing’s renown now extends well beyond computer science⁴⁶ evinces the wider recognition of computing’s importance to modern society. Machine learning, as the branch of computing that now so influences the field, requires a commensurate breadth of understanding. We have written here about jurisprudence and the path to AI. Machine learning’s impact, however, extends well beyond the legal profession. Every walk of life is likely to feel its impact in the years to come. Existing rules might help with some of the problems to which the new technology will give rise, but lawyers and judges will not find all the answers ready to “read off” the existing rules. We hope that having presented the ideas and ways of thinking behind machine learning through an analogy with jurisprudence will help lawyers to fold the new technology into the law—and will help laypeople fold it into the wider human experience across which machine learning’s impact now is felt.

At the very least, we hope that lawmakers and people at large will stop using the word “algorithm” to describe machine learning, and that they will ask for “the story behind the training data” rather than “the logic behind the decision.”

Notes

• 1. The Occasional Speeches Of Justice Oliver Wendell Holmes, compiled by Mark De Wolfe Howe, Cambridge, MA: The Belknap Press of Harvard University Press, Copyright © 1962 by the President and Fellows of Harvard College. Copyright © renewed 1990 by Molly M. Adams.

• 2. It was Holmes’s father who referred to Boston as the “hub of the universe,” in part to mock its inhabitants’ self-importance, but also to observe the real importance of the city for arts and sciences at that time: Budiansky 23–27.

• 3. Learning and Science, Speech at a Dinner of the Harvard Law School Association in Honor of Professor C.C. Langdell (June 25, 1895), De Wolfe Howe (ed.) (1962) 84, 85.

• 4. See for an overview Faralli, Legal Philosophy in Italy in the Twentieth Century in Pattaro & Roversi (eds.), A Treatise of Legal Philosophy and General Jurisprudence (2016) 369 ff.

• 5. 12 Harv. L. Rev. at 462.

• 6. See https://www.census.gov/history/www/innovations/technology/the_hollerith_tabulator.html.

• 7. See https://www.ibm.com/ibm/history/ibm100/us/en/icons/tabulator/.

• 8. See https://www.ibm.com/ibm/history/ibm100/us/en/icons/tabulator/. The potential uses of the machines in different industries was reflected in the patents that Hollerith filed for them: “my invention is not limited to such a system [for the census] but may be applied in effecting compilations of any desired series or system of items representing characteristics of persons, subjects, or objects.” Quoted at Geoffrey D. Austrian, Herman Hollerith: Forgotten Giant of Information Processing (1982) 83. A commercial breakthrough for Hollerith came when managers recognized that the machine vastly improved cost accounting in factories: id. at 200–203; and another when department stores started using it to analyze sales data: id. at 203–205.

• 9. The Census of the United States, 63(9) Scientific American 132 col. 2 (Aug. 30, 1890).

• 10. Id. The machine’s versatility was recognized from the start. Robert P. Porter, Superintendent of the U.S. Census for 1890, reported to the Secretary of the Interior that the machine allowed “the facts [to] be presented in a greater variety of ways” than heretofore practical: Porter to the Secretary of the Interior, July 3, 1889, as quoted by Austrian at 49. Cf. id. at 64–65, 69. Emphasizing the qualitative change that the Hollerith machine brought about, see Norberg (1990). As to the sheer speed of the machine, this was demonstrated in Census Bureau tests in which it handily beat Hollerith’s two best contemporary rivals: Austrian at p. 51.

• 11. Austrian at 124.

• 12. Holmes to Pollock (June 10, 1923), reprinted De Wolfe Howe (ed.) (1942) 118.

• 13. Holmes’s later critic, Lon Fuller, who understood legal realism to fail as a philosophy of law because of its tendency to identify a normative force in facts, picked out the Hollerith machine as an emblem of the realists:

  The facts most relevant to legal study will generally be found to be what may be called moral facts. They lie not in behavior patterns, but in attitudes and conceptions of rightness, in the obscure taboos and hidden reciprocities which permeate business and social relations. They are facts of a type which will not pass readily through a Hollerith statistical sorting machine…

  Lon Fuller, Lecture II, Julius Rosenthal Lectures, Northwestern University: The Law in Quest of Itself (1940) 45, 65.

• 14. Dr. Oliver Wendell Holmes at Cambridge University, 22(551) Scientific American 8806 col. 2 (July 24, 1886).

• 15. The Holmes materials in the Harvard Law School Library Digital Suite include a receipt for Holmes’s 1927 subscription to the Scientific American; and it appears from the inventory in his estate that he was clipping articles from the journal in 1922 and receiving it in his library in 1913. See 3:HLS.Libr:7678129 seq. 151; 3:HLS.Libr:8582493 seq. 37; HLS.Libr:8268117 seq. 39.

• 16. As to which, see Smith Rinehart, Holmes on Patents, 98 J. Pat. Trademark Off. Soc’y 896 (2016).

• 17. Consider for example the dispute involving the Census Office itself, Hollerith’s former customer: Austrian at 264.

• 18. Hollerith’s Tabulating Machine Company had its headquarters and workshops from 1892 to 1911 at 1054 31st Street, NW: Austrian, pp. 97–99; and see photograph id. between pp. 182 and 183. On first arriving in Washington in December 1902 to begin service as Associate Justice, Holmes and his wife lived at 10 Lafayette Square: Catherine Drinker Brown, Yankee from Olympus: Justice Holmes and His Family (1945) 353. They later moved to 1720 I Street, which Holmes acquired and refurbished in 1902: Budiansky 286. Senator Henry Cabot Lodge, Jr., writing after Holmes’s death to Edward J. Holmes, Holmes’s nephew, thought the house on I Street, which Holmes had bequeathed to the government, could be used as a “shrine to [Holmes’s] memory” (letter dated June 6, 1939): 3:HLS.Libr:8582488 seq. 53. Edward said the idea would not have pleased Holmes: letter to Lodge dated June 16, 1939, HLS.Libr:8582488 seq. 55. Nothing came of plans to preserve the house. A nondescript office building now occupies the site. The exterior walls of the former warehouse in which the Tabulating Machine Company did its work still exist. IBM placed a plaque there in 1984 to note the connection.

• 19. Austrian, pp. 60–62.

• 20. Characteristic was his excitement, expressed in a letter to Lady Pollock, about an outing to see a troupe of jugglers: Holmes to Lady Pollock (May 13, 1898), reprinted De Wolfe Howe (ed.) (1942) 87. And he walked to and from work (at least as late as his 70s, which is to say he was still commuting by foot when the 1910 census was counted). See, e.g., Holmes’s reference to walking home past the Treasury: Holmes, Law and the Court, Speech at a Dinner of the Harvard Law School Association of New York (Feb. 15, 1913), in Posner (ed.) (1992) 145, 148. It also appears that Holmes and his wife at least on one occasion had an outing along the C & O Canal, the waterway on which the Hollerith building is located: Budiansky 281.

• 21. See example, regarding science and bureaucracy in mid-nineteenth century France, Fox, The Savant and the State: Science and Cultural Politics in Nineteenth-Century France (2012) 94–137; regarding the systematized survey of its new borders after the partitions of Poland, see Olesko, The Information Order of the Prussian Frontier, 1772–1806 (Max-Planck-Institut für Wissenschaftsgeschichte, 2019).

• 22. Hollerith naturally comes up in connection with the growth and intensification of public bureaucracy. See, e.g., Biography of an Ideal: A History of the Federal Civil Service (U.S. Civil Service Commission, Office of Public Affairs, 1973) 176. See also Beniger, The Control Revolution: Technological and Economic Origins of the Information Society (1986) 399–400 and passim.

• 23. Not to be confused with eminent historian of modern Europe, Franklin L. Ford (1920–2003).

• 24. Holmes to Ford (Apr. 26, 1907), quoted in Burton (1980) 204.

• 25. Holmes, Law and the Court, Speech at a Dinner of the Harvard Law School Association of New York (Feb. 15, 1913), in Posner (ed.) (1992) 145, 148. Elsewhere, Holmes suggested he was no fan of Wagner: see Budiansky 194–95, 424.

• 26. At least in any performance of the opera as staged to date.

• 27. The weaving characters at the start of Götterdämmerung were called Norns, Old Norse for Fates. See Richard Wagner, Götterdämmerung (Twilight of the Gods) (Stewart Robb, trans.) (London: Scribner, 1960) 1–2.

• 28. 12 Harv. L. Rev. at 462–63.

• 29. Posner, 110 Harv. L. Rev. 1039, 1040 (1997).

• 30. Judge Posner quotes the passage here: 110 Harv. L. Rev. 1040 n 3; and (1997) 63 Brook. L. Rev. 7, 14–15 (1997).

• 31. See for example Voegelin, The New Science of Politics (originally the 1951 Walgreen Foundation Lectures), republished in Modernity Without Restraint, Collected Works, vol. 5 (2000); The Origins of Scientism, 15 Soc. Res. 473–476 (1948).

• 32. See for example Bostrom, Existential Risks: Analyzing Human Extinction Scenarios and Related Hazards, 9 J. Evol. Tech. (2002).

• 33. For an arch-critic, see Alschuler (1997). An admirer who nevertheless critiques Holmes on this score is Posner: see 110 Harv. L. Rev. at 1042 (1997).

• 34. Gottfried Wilhelm Leibniz, Dissertatio de Arte Combinaoria (1666), quoted by Livermore in Rule by Rules, chapter in Whalen (ed.) (2019) 3.

• 35. Livermore (2019).

• 36. Reprinted in H.L.A. Hart, Essays in Jurisprudence and Philosophy (1983) 22.

• 37. Hart, Positivism and the Separation of Law and Morals, 71 Harv. L. Rev. 593, 607 (1958).

• 38. Adopted April 27, 2016, applicable from May 25, 2018.

• 39. AB375, Title 1.81.5 (signed into law June 28, 2018, to be operative from Jan. 1, 2020).

• 40. For a survey, see U.S. Library of Congress, Regulation of Artificial Intelligence in Selected Jurisdictions (January 2019).

• 41. 71 Harv. L. Rev. 593.

• 42. Lon Fuller was one of the main critics. See Hart’s 1957 Holmes Lecture, Positivism and the Separation of Law and Morals, 71 Harv. L. Rev. 593–629 (1958). See also Sebok (2009).

• 43. Nicola Lacey, in her A Life of H.L.A. Hart: The Nightmare and the Noble Dream (2006), notes the close involvement of Hart with Bletchley Park during World War II, but she does not note any particular personal ties between Hart and Turing: Lacey 90–93. A personal connection is suggested, in passing, in Gavaghan, Defending the Genetic Supermarket: The Law and Ethics of Selecting the Next Generation (London: Routledge, 2007) 37. Lacey, who worked closely with Hart’s personal papers, is not aware of any closer tie between Hart and Turing apart from the circumstantial one of Hart’s role as MI5’s ULTRA liaison: correspondence from Professor Lacey to T.D. Grant (Aug. 9, 2019).

• 44. Finnis notes that Hart “patriotically maintained the mandated secrecy about these activities [for MI5 during the war] down to the end of his life.” John Finnis, H.L.A. Hart: A Twentieth Century Oxford Political Philosopher: Reflections by a Former Student and Colleague, 54 Am. J. Jurisprud. 161 (2009).

• 45. Correspondence from Adam Hart to T.D. Grant (Aug. 15, 2019).

• 46. Recall Chapter 9, p. 111, n. 17.