When Computers Were Human

by David Alan Grier

  6. Difference engine constructed from the original plans of Charles Babbage

  The Laputian philosopher that Babbage wished to avoid was not the computing tailor but an inventor who lived on the flying island. This inventor claimed to have a computing machine whereby “the most ignorant person, at a reasonable charge, and with a little bodily labour, might write books in philosophy, poetry, politics, laws, mathematics, and theology, without the least assistance from genius or study.”63 The machine was a silly device, a box of shafts and gears that spun through every possible combination of words, but it was a symbol of Jonathan Swift’s mockery of the Royal Society. Babbage, though he shared some of Swift’s reservations about the society, desired to avoid any comparison to the mythical device. He carefully described his invention in the context of de Prony’s computing organization. He explained how the computers prepared their tables and then analyzed “what portion of this labour might be dispensed with.” By his count, de Prony had employed ninety-six individuals to produce seven hundred computations a day. Babbage claimed that his machine could replace all of the human computers and most of the mathematicians. Of the original staff, all that would remain would be the ten planners and one mathematician, “or at the utmost two,” to direct the work.64

  Though the Royal Society was in no hurry to pass judgment upon Babbage’s proposal, the Astronomical Society rushed to give the idea uncritical praise. Based upon what they observed of the crude prototype, they offered Babbage a gold medal for his contribution to astronomy.65 “The labour of computing equations with the pen would be immense, and liable to innumerable errors,” wrote one member of the Astronomical Society, “but with the assistance of [Babbage’s] machine, they are all deduced with equal facility and safety.” He made a special effort to emphasize the general use of the machine, arguing that “astronomical tables of every kind are reducible to the same general mode of computation” and that the machine could even be applied to commercial tasks, such as the preparation of “Interest, Annuities, &c, &c, all of which are reducible to the same general principles.”66

  The Royal Society eventually endorsed the Difference Engine, and the English government offered to finance the construction of the machine. Eager to devote his entire energies to the project, Babbage resigned his office with the Astronomical Society in 1823. The work progressed more slowly than he would have wished, as both he and his mechanic needed to refine his design and improve their metalworking skills. The project was interrupted by the death of Babbage’s wife, Georgiana, an event that disrupted his life in a way that nothing else could have. The loss “left Babbage a changed man,” observed biographer Anthony Hyman. There was “an ‘inner emptiness’ to the man, who had only recently seen so much potential in his life.” In “his public controversies, there was a new note of bitterness of which there was no trace while Georgiana was alive.” Babbage left England for a Continental tour in 1827, leaving the engine unfinished.67

  In later years, Babbage recognized that he had been naive in his attempt to build the Difference Engine. Without referring to himself directly, he confessed that a novice engineer could be “dazzled with the beauty of some, perhaps, really original contrivance,” and would rush into its construction “with as little suspicion that previous instruction, that thought and painful labour, are necessary to its successful exercise.”68 Babbage worked on the Difference Engine for ten years. During this time, he was engaged in other projects, such as computing life insurance tables, forming two new scientific societies, and writing a book about manufacturing. Even accounting for these other projects, the work on the Difference Engine took longer than Babbage had anticipated, and it encountered unforeseen problems. The English government eventually grew impatient with Babbage’s progress. Concluding that they would see no return on their investment, they withdrew their financial support in 1834, forcing Babbage to terminate the project.

  Undeterred by his failure to complete the Difference Engine, Babbage moved to design a second, more ambitious device. He never even attempted to construct this machine, which he called the Analytical Engine. Modern writers have generally viewed the machine as an important intellectual step toward the stored-program electronic computer. One historian has gone so far as to claim that it was “a general purpose computer, very nearly in the modern sense.”69 The drawings of this machine show how Babbage anticipated the features of a modern computer, though his design used gears and levers rather than chips and circuit boards. The Analytical Engine had a means of storing numbers, a central processor, and an elementary programming mechanism. Unlike the Difference Engine, this machine was not restricted to a single mathematical method, such as the method of finite differences. The programming mechanism, which read instructions from a string of punched cards, controlled the order of operations. One observer, the daughter of the poet Lord Byron, Ada Lovelace (1815–1852), called the Analytical Engine the “material and mechanical representative of analysis,” a triumph of the division of mathematical labor. Lovelace herself illustrated the nature of the machine by writing a sample program for it.70

  Babbage would spend almost fifteen years designing the Analytical Engine. He left nearly three hundred detailed engineering drawings of his proposed machine.71 As he worked over these drawings, he recognized that the Europe of the early nineteenth century might not be able to support large computing organizations or computing machines based upon the division of labor. The “most perfect system of the division of labour is to be observed,” he wrote, “only in countries which have attained a high degree of civilization, and in articles in which there is a great competition amongst the producers.”72 As the early nineteenth century saw little competition for scientific computation, it offered little opportunity for the sophisticated division of labor espoused by Babbage and de Prony.

  CHAPTER THREE

  The Celestial Factory: Halley’s Comet 1835

  Saw the heavens fill with commerce, argosies of magic sails, Pilots of the purple twilight, dropping down with costly bales.

  Alfred, Lord Tennyson, Locksley Hall (1842)

  IN HIS NOVEL Hard Times, Charles Dickens described an astronomical observatory “made without any windows” and an astronomer who “should arrange the starry universe solely by pen, ink, and paper.” He used this description, which sounded more like the computing room of the Nautical Almanac than the staff of an observatory, as a metaphor for a factory. In this factory, the director “had no need to cast an eye upon the teeming myriads of human beings around him,” just as the director of almanac computations had no need to watch the stars each night, “but could settle all their destinies on a slate, and wipe out all their tears with one dirty little bit of sponge.” When Dickens used this metaphor, both the Royal Observatory at Greenwich and the British Nautical Almanac Office had adopted the basic elements of factory production, elements that included a central facility and a standard schedule of operations or, as Dickens described them, “a stern room, with a deadly statistical clock in it, which measured every second with a beat like a rap upon a coffinlid.”1

  Both the almanac and the observatory consciously accepted such methods around the time of the 1835 return of Halley’s comet. Neither organization had seen much innovation since the days of Nevil Maskelyne. Through the middle part of his career, Maskelyne had been an active leader of both organizations, acquiring new equipment for the observatory and developing new methods for the almanac. At some point in the 1780s or 1790s, he had settled into a comfortable routine and had watched innovations occur elsewhere. The major astronomical discovery of the late eighteenth century, the planet Uranus, had been accomplished by an independent observer, William Herschel, not by the observatory staff. The radical division of labor came from Gaspard de Prony. A second periodic comet, the first to be discovered after Halley’s, was identified by at least four individuals, none of whom was associated with Greenwich. This comet was ultimately named for a German astronomer, Johann Franz Encke (1761–1865), who calculated the
object’s orbit.2

  Following the death of Nevil Maskelyne in 1811, the British Admiralty made only a few improvements to the almanac and the observatory. The new Astronomer Royal apparently spent little time at the observatory and acquired a reputation for hiring unimaginative assistants, “indefatigable, hard-working, and, above all, obedient drudges.” A critic of the observatory argued that “Men who had the spirit of ‘drudges,’ to whom observation was a mere ‘mechanical act,’ and calculation a ‘dull process,’ were not likely to maintain the honour of the Observatory.”3 Such assistants had not sustained the honor of the Nautical Almanac. In 1818 the Admiralty removed the publication from the observatory and appointed an independent supervisor.4

  Of the two organizations, the Nautical Almanac Office was the first to undergo a serious reform and adopt factory methods. Its workers were more familiar with the demands of production, as they had labored under fixed deadlines since the founding of the almanac in 1767. The pressure to reform the almanac came not from the almanac staff or even from the Admiralty but from the Royal Astronomical Society, the group that had started its life in a London tavern as the simple Astronomical Society. “The most prominent subject of public interest,” reported the society president, “was the proposing of an amended form of the Nautical Almanac.”5 He argued that the almanac devoted too many pages to the lunar distance method of navigation and not enough to tables that would assist contemporary navigators and astronomers. The society claimed, with no contradiction from the British Admiralty, that most ship officers had abandoned Maskelyne’s technique for calculating the time. In its place, chronometers, the high-precision clocks, could “be found in perhaps every ship which relies upon astronomical means for her guidance.”6

  With the consent of the Admiralty, the Royal Astronomical Society formed a review board for the almanac. This committee met in the offices of the society and consisted of a broad selection of almanac users, including naval officers and astronomers, shipowners and insurance men, financiers and merchants, scientists and clergy, the Astronomer Royal and Charles Babbage.7 The committee published its recommendations in 1830 and 1831. In spite of their objections to the lunar distance method, they refused to cut any of Maskelyne’s lunar distance tables and suggested that such tables be expanded and redesigned. They also requested several substantial changes: new tables of the planets, values that would improve astronomical observations made on board ships, the “mean time of high water at London Bridge,” dates of Islamic holidays (“which may be occasionally useful to officers cruising in the neighbourhood of Mohammedan states”), and the expansion of many other tables. The proposed new additions increased the size of the periodical by 50 percent, but the committee felt that the calculations could be distributed “amongst the several computers as will afford them constant employment.” “With due economy,” they concluded, “the whole of the additional computations may, in a short time, be obtained without much (if any) additional cost to the nation.”8

  After receiving the report of the Royal Astronomical Society, the British Admiralty appointed one of the review board members, Lieutenant William Samuel Stratford (1791–1853), to oversee the almanac and implement the needed changes. From the start, Stratford concluded that the computers would be most efficient and be most constantly employed if they worked in a central office. He gave the old staff “due notice that their services would no longer be required after the completion of the Almanac for 1833,” leased rooms for an almanac office in central London, and hired new computers. Only a few of the old staff moved to London and joined the new office. Most of the old computers wished to remain with their homes and families in the country. Stratford’s staff would be drawn from city dwellers and would calculate in the almanac computing room, follow a daily schedule, and be under the watchful eye of the superintendent.9

  Stratford opened the new almanac office shortly before the 1835 return of Halley’s comet. There was less anxiety over this return than over its 1758 appearance. No one, at least none of the major astronomers, questioned the basic principles of Newton’s theory, no one argued that the comet was anything more than a celestial object, and no one decried the “spirit of calculation.” Stratford identified five major attempts to compute the comet’s orbit and reviewed all five in the first volume of the almanac that was prepared under his superintendence. He reported that all of them identified roughly the same orbit and that the “principal variation appears to be in the time of the perihelion passage.”10 He felt that the most complete calculation was done by the French astronomer Philippe Gustave Le Doulcet, Comte de Pontécoulant (1795–1874). Pontécoulant had spent five years computing the comet’s path and adjusting his figures. He started with an idealized orbit, a perfect ellipse around the Sun. Step by step he added the major influences on the comet: the gravity of Jupiter, Saturn, and Uranus. He even adjusted his equation to account for the position of the Earth during the 1758 passage. His first calculations identified November 7 as the date of the perihelion. His second effort moved the date to November 13. The third retreated it to November 10. The last advanced the perihelion to the evening of November 12.11

  Under the direction of Stratford, the almanac computers produced an adjustable ephemeris for Halley’s comet. In comparison with Pontécoulant’s calculation, it was a mundane activity, a practical contribution to astronomy rather than a grand test of Newton’s theory. Once the comet had been spotted, this ephemeris could be used to plot the comet’s position in the night sky and to predict the date of perihelion. Stratford used the ephemeris to engage about one hundred astronomers, both professionals and amateurs, to search the sky for the comet and to record its position. He instructed the observers to send their records of the comet to the almanac office. From the data, the almanac computers revised their equations for the comet and filed their results for the next generation of astronomers to use. The Royal Astronomical Society praised this work as a “most useful and arduous task.”12 Taken as a whole, the computations for the 1835 return halved the error of the 1758 calculation. The actual date of the perihelion, November 16, fell within sixteen days of all the major predictions. One of Pontécoulant’s predictions came within three days and a few hours of the true perihelion.13

  7. Halley’s comet in 1835 (fifth from right) with other nineteenth-century comets

  The reform of the Greenwich Observatory began just as the comet swung past the sun and began retreating from view. In the eyes of contemporary astronomers, the observatory required many changes. It needed new equipment, a stronger staff, and revised methods of operation. Like the Nautical Almanac Office, it had become a center of production, but this production was considered a burden on the staff. The British Admiralty had given the observatory the responsibility of caring for the navy’s stock of chronometers. Observatory personnel cleaned, tested, and corrected the time of every chronometer before it departed England on an ocean voyage. These tasks occupied an entire room of the observatory and the labor of several observatory personnel. The director of the observatory at Cambridge, George Airy (1801–1892), looked at the activities of the Greenwich staff and complained that chronometer work degraded the Royal Observatory “into a mere bureau of clerks” and added that “it is difficult even for the director to resist the contagion” of such tasks.14

  Airy had made his reputation as a reformer. Under his direction, the Cambridge Observatory had been transformed from a small, uninteresting teaching facility into “one of the pace-setting scientific centers of England.”15 This transformation was the crowning achievement of his career at Cambridge. He had enrolled at the university as a sizar, a scholarship student too poor to pay the tuition, and he had graduated at the top of his class. He held the title of “Senior Wrangler,” a title bestowed on the student with the top score on the Tripos exam, the honors exam for students in astronomy, physics, and mathematics. For a time, Airy had also served as the Lucasian Professor of Mathematics, the position that had once been occupied by Isaac Newton and, m
ore recently, by Charles Babbage.16

  In 1835, the Admiralty appointed George Airy as the Astronomer Royal. The first test of his leadership was a collection of unprocessed astronomical data that had been accumulated over twenty-five years. These numbers lay unused in observatory logs, some written with neat and refined digits, some scribbled in haste, some recorded with hands that had grown stiff and cramped from the cold night air. This backlog of data was “a lump of ore,” Airy observed, which is “without value till it has been smelted.”17 The smelting process is called “reduction.” When computers reduce data, they take the raw values from the telescopes and convert them into a form that astronomers can use for research. For each observation of a celestial object, an astronomer generally records four numbers: the height of the object above the horizon (altitude), the direction of the object in the sky (azimuth), the time of day, and the date of the year. These four values change as the Earth rotates. The process of reduction collapses, or reduces, these four values into a pair of fixed numbers that represent the position of the object on the celestial sphere.

  The celestial sphere is an imaginary hollow globe that surrounds the Earth. Astronomers identify positions on the celestial sphere with astronomical longitude (called the right ascension) and astronomical latitude (declination). The right ascension is measured from the celestial equator, which lies above the Earth’s equator. The declination is measured from the celestial prime meridian, which runs north and south through the constellation Aries, the first constellation of spring, the traditional herald of the new year. The computations for data reduction require about ten steps, depending on how precise one wishes to be, and require a firm knowledge of trigonometry.18