It Began with Babbage Page 5
3. L. F. Menabrea. (1842). “Sketch of the Analytical Engine”, Bibliothéque Universelle de Genève [On-line], October, no. 42. Available: http://www.fourmilab.ch/babbage/sketch.html
4. Letter from Charles Babbage to the Earl of Rosse, President, Royal Society, June 8, 1852 (pp. 77–81). In C. Babbage. (1994). Passages from the life of a philosopher (p. 79). Piscataway, NJ: IEEE Press (original work published 1864).
5. C. Boyer. (1991). A history of mathematics (2nd ed., Rev., pp. 443–445). New York: Wiley.
6. Menabrea, op cit.
7. A.. A. Lovelace. (1842). Notes. In Menabrea, op cit.
8. Menabrea, op cit.
9. M.. Campbell-Kelly. (1994). Introduction (pp. 7–36). In Babbage, op cit., p. 23.
10. Lovelace, op cit.
11. M. B. Hesse. (1966). Models and analogies in science. London: Sheed & Ward; J. Holland, K. J. Holyoak, R. E. Nisbett, & P. R. Thagard. (1986). Induction (Chapter 10). Cambridge, MA: MIT Press.
12. S. Dasgupta. (1994). Creativity in invention and design (pp. 27–33). New York: Cambridge University Press.
13. R. L. Hills. (1990). Textiles and clothing. In I. McNeil (Ed.), An encyclopaedia of the history of technology (pp. 803–854). London: Routledge (see especially pp. 822–823); J. Essinger. (2004). Jacquard’s web. Oxford: Oxford University Press.
14. Essinger, op cit. D. S. L. Cardwell. (1994). The Fontana history of technology. London: Fontana Press.
15. Lovelace, op cit.
16. Ibid.
17. Ibid.
18. C.. Babbage. (1837). On the mathematical power of the calculating engine. Unpublished manuscript, December 26, Oxford University, Buxton MS7, Museum of the History of Science; printed in B. Randell. (1975). The origins of the digital computer (2nd ed., pp. 17–52). New York: Springer-Verlag (see especially p. 21).
19. R. Moreau. (1984). The computer comes of age (p. 15). Cambridge, MA: MIT Press.
20. A. G. Bromley. (1982). Charles Babbage’s Analytical Engine, 1838. Annals of the History of Computing, 4, 196–217 (see especially p. 196).
21. Ibid., p. 197.
22. Bromley (op cit.) has several drawings that show some of the mechanisms designed by Babbage.
23. M. V. Wilkes. (1981). The design of a control unit: Reflections on reading Babbage’s notebooks. Annals of the History of Computing, 3, 116–120.
24. Bromley, op cit., pp. 197–198.
25. M. V. Wilkes. (1971). Babbage as a computer pioneer. Historia Mathematica, 4, 415–440.
26. Wilkes, 1981, op cit.
27. See chapter 1.
28. Campbell-Kelly, op cit., p. 24.
29. Babbage, 1994, op cit., p. 79.
30. Ibid.
31. The Age of Romanticism, circa 1770 to 1835, was a time when not only poetry, fiction, and art were imbued with the spirit of wonder about nature and ourselves, but also science was touched by the same spirit. See R. Holmes. (2008). The age of wonder. New York: Viking Books.
32. G. Sturt. (1923). The wheelwright’s craft. Cambridge, UK: Cambridge University Press.
33. J. C. Jones. (1980). Design methods: Seeds of human future (2nd ed.). New York: Wiley; C. Alexander. (1964). Notes on the synthesis of form. Cambridge, MA: Harvard University Press.
34. S. Dasgupta. (1999). Design theory and computer science (pp. 368–379). Cambridge, UK: Cambridge University Press (original work published 1991).
35. L. Pyenson & S. Sheets-Pyenson (1999). Servants of nature (p. 336). New York: W. W. Norton.
36. Nor is the history of art any better. See W. Chadwick. (2007). Women, art and society (4th ed.). London: Thames & Hudson.
37. Pyenson & Sheets-Pyenson, op cit., pp. 342–344.
38. Holmes, op cit.
39. S. Wood (2010). Mary Fairfax Somerville [On-line] (original work published 1995). Available: http://www.agnesscott.edu/lriddle/women/somer.htm
40. Pyenson & Sheets-Pyenson, op cit., pp. 347–348.
41. B. Toole (2011). Ada Byron, Lady Lovelace [On-line]. Available: http://www.agnesscott.edu/lriddle/women/love.htm
42. Campbell-Kelly, op cit., p. 27.
43. Babbage, 1994, op cit., p. 102.
44. Lovelace, op cit., Note G.
45. Babbage, 1994, op cit., p. 102.
46. Ibid.
47. Lovelace, op cit., Note G.
3
Missing Links
I
IN CHAPTER 2, I suggested that Babbage’s place in the history of computing was twofold: first, because his Analytical Engine represented, for the first time, the idea of automatic universal computing and how this idea might be implemented, and second, because some of his design ideas—the store, mill, control, user interface via punched cards—anticipated some fundamental principles of the electronic universal computer that would be created some 75 years after his death. There is a modernity to his idea that makes us pause. Indeed, it led Babbage scholar Allan Bromley to admit that he was “bothered” by the architectural similarity of the Analytical Engine to the modern computer, and he wondered whether there is an inevitability to this architecture: Is this the only way a computer could be organized internally?1
Thus, Babbage’s creativity lay not only in conceiving a machine that had no antecedent, but also it lay in his envisioning an idea of universal computing that disappeared and then reappeared many decades later, and came to be the dominant architectural principle in computing. This observation is, of course, present-centered; we might be perilously close to what Herbert Butterfield had called the “Whig interpretation of history” (see Prologue, section VII), for we seem to be extolling Babbage’s achievement because of its resonance with the achievements of our own time. But were there any direct consequences of his idea? What happened after Babbage? Did he have any influence on those who came after? And, if not, what took place in the development of what we have come to call computer science?
II
In fact, there is a view that between Babbage’s mechanical world of computing and the electronic age, nothing really happened—that the time in between represented the Dark Ages in the history of computing. This is, of course, as misguided a view as another held by historians at one time that Europe, between the end of the Roman Empire (circa fifth century) and the Renaissance (the 15th–16th centuries)—the Middle Ages—was in a state of intellectual and creative backwardness. Which is why the Middle Ages was also once called the Dark Ages.
Just as modern historical scholarship revealed that the Middle Ages was anything but Dark,2 so also must we discount vigorously the idea that the period between Babbage and the age of the electronic computer was a Dark Age in the history of computing. In fact, when we examine what transpired after Babbage, we find that it was a period when two very different and lively views of computing took shape. I will call these the abstract and the concrete views, corresponding—broadly—to the conception of computational artifacts that are abstract and physical, respectively.
Let us dwell, for the moment, on the concrete view. What we find is that this was a period (roughly between 1880 and 1939) during which several species of material computational artifacts were created. These species were linked in that they were invented with a common purpose: to automate computing as much as possible. Yet, the linkages differed in that, although some machines aspired to universal computing, others had more modest aspirations.
I use the word species metaphorically, of course. Biology offers us yet another metaphor. Ever since the 18th century (well before Darwin, Wallace, and the advent of their particular theory of evolution), physical anthropologists and natural historians3 interested in the connection between man and ape have quested for the “missing link” between the two.4 After Darwin, the very credibility of Darwinian natural selection rested, at least in part, on the discovery of missing links in the fossil record.5 I am tempted to term collectively the period between Babbage’s design of the Analytical Engine (circa 1840s) and the advent of the electronic computer (circa 1940s) the Age of Missing Link
s, for they constituted designs and inventions that paved pathways of ideas from Babbage’s vision of a universal automatic computing engine to the practical realization of that vision.
III
The punched card, used by Jacquard for his loom and then Babbage’s key to the universality of the Analytical Engine, developed an identity and a universality of its own. It became a repository of information or symbol structures—a memory device, in fact. And a whole new genus of machines, now powered by electricity, came into existence just to manipulate and process the contents of these punched cards. “Computing” not only signified esoteric mathematical computation, but also came to mean an activity called data processing—a genre of computing involving data pertaining to human beings; to human society, commerce, health, welfare, economy, and institutions. “Tables” meant not only tables of logarithms or trigonometric functions, but also printed tables of such data on the nitty-gritty of individual life—name, place and date of birth, age, gender, religion, occupation, ethnicity, educational level, date of death, and so on.
And the United States entered the history of computing.
IV
In 1890, Herman Hollerith (1860–1929), American-born son of German immigrants, inventor and statistician, submitted a dissertation bearing the title In Connection with the Electric Tabulation System which Has Been Adopted by U.S. Government for the Work of the Census Bureau to the Columbia University School of Mines, for which he received a PhD. This must surely be the first doctoral degree awarded in the field of computing, the first recognition of computing as an academically respectable discipline.
Hollerith obtained an engineering degree from the Columbia School of Mines in 1879; his academic record was such that, after graduating, one of his professors, William P. Trowbridge (1828–1892), appointed him his assistant. It was a fateful appointment. When Trowbridge became chief special agent in the U.S. Bureau of Census, Hollerith moved with him as a statistician. At the Bureau, Hollerith met John Shaw Billings (1839–1913), a surgeon in the U.S. Army who had been assigned to the Bureau earlier to help with statistical work related to census data. At the time Hollerith joined the agency, Billings was in charge of collecting and tabulating data for the 1880 U.S. census.6
The history of computing repeatedly tells a story of dissatisfaction with the use of human mental labor for tasks of a mechanical nature. Thus it was with Babbage (as we saw); thus it was with Hollerith. Although accounts differ,7 the essence of the story is the same: Billings, remarking in Hollerith’s presence, that there ought to be a mechanical way of tabulating census statistics. Thus was planted a seed in Hollerith’s mind. In 1882, he spent some time teaching in the mechanical engineering department at the Massachusetts Institute of Technology (MIT)—between Babbage and Hollerith we find the first of many appearances of the two Cambridges in the history of computing—and during his brief tenure there (he left in 1884 to take a post in the U.S. Patent Office), he worked on the problem of converting information punched as configurations of holes in cards into electrical impulses that, in turn, would drive mechanical equipment. This was the beginning of electromechanical computing.
V
Analyzing the way census and other similar kinds of demographic data had been gathered before, Hollerith identified some basic data processing operations involved in the process: sorting data in some order, counting, and tallying such data.8 The operations seem simple enough, involving mental activity that any literate, numerically competent person can carry out—clerical operations, in other words. The problem is that of volume. The amount of data that may have to be sorted and tallied may be unmanageably vast. In the case of Census data—and this was the original need posed to Hollerith—it may entail records on the population of a city, a state, a whole country. Rather than process such massive volumes of data manually, perhaps this work could be done mechanically, as far as possible.9 We are, once more, reminded of Leibniz’s remark: “excellent men” should not have to waste time in the drudgery of calculation that could be “delegated” to machines.10
Leibniz’s theme, as I called it, may have been meant for mathematical computation; his “excellent men” referred to astronomers and mathematicians. His theme, however, echoed by Babbage, also became Hollerith’s theme: to automate computation—not the kind of computation Leibniz and Babbage had in mind, involving transmutations of complex mathematical functions into sequences of machine operations, but computation nonetheless: sorting, counting, tallying, and organizing large volumes of data. To Hollerith, automatic computation meant automatic data processing. We might say that if Babbage initiated the realm of automatic scientific computing, Hollerith commenced the realm of automatic commercial or business computing, and the latter responded to a particular need. Hollerith’s major intellectual achievement was to invent “an electric tabulating system,” which also gave him a doctoral degree from Columbia University.
Tabulation entails printing a summary obtained by counting. Electromechanical machines that did this came to be called tabulators or, sometimes, accounting machines.11 A well-known example was the IBM 407 Accounting Machine. In fact, Hollerith invented not just a single machine but a system comprising machines and a process that would be carried out by the system. His electric tabulating system involved a device called a keyboard punch, a tabulator, and a device that he called a sorting box (a distant forerunner to the later automatic sorter). Curiously, the printing of tabulated data was done by hand in Hollerith’s system—a situation that would be improved in later developments by others—to include an automatic printer as part of the system.12 What was processed was data transcribed onto punched cards.
In Hollerith’s hands, the original Jacquard card (or its Babbageian version) became something entirely different. His cards were 3 inches by 5.5 inches,13 and each card held relevant information about an individual. Hollerith described, as an example, the format of the card used by the U.S. Surgeon-General for compiling U.S. Army health statistics.14
Each card represented an individual soldier. The card’s surface was partitioned into several sections (or “fields”, to use a later term), each devoted to particular, relevant aspects. Each field, in turn, consisted of an array of subfields, with numbers or abbreviated codes assigned to the subfields. One field was given, for instance, to the division to which the soldier was assigned, and the subfields corresponded to the various divisions. A hole punched in one of the subfields would indicate the soldier’s division.
Thus, a single Hollerith card became a memory device that held data that, in a manner of speaking, defined the “identity” of that individual (from the Army’s perspective, that is). It held a record about a person, such as rank, the branch of the armed services, age, race, nationality, and length of service.15 To paraphrase a comment by a character in Don DeLillo’s novel White Noise (put in an altogether different context), one is the sum total of one’s data.16
FIGURE 3.1 A Hollerith Data Processing System.
The system was completely general—universal in the same sense that the Jacquard card (in the realm of weaving) and the Babbage card (in the realm of mathematical computing) were universal. By changing the format design a different card would obtain, as was used by New York City’s Board of Health to compile mortality statistics17 and, of course, a different format design would be used to compile statistics in the 11th U.S. Census of 1890, in which Hollerith’s system played a central role.
His system can be depicted by a system diagram, as shown in Figure 3.1. From the census returns documents, a keyboard punch transcribes the information for each individual onto a card. The cards are then placed, one by one, on a part of the tabulating machine Hollerith called “the press”, which has sensors that read the data by sensing the holes, and send electrical signals to the other main component of the tabulator comprising mechanical counters in the form of dials. Specific counters are wired through circuits to the sensors that read specific fields in the card. For example, one counter could be connected t
o the sensors in the press so that it counted the number of males in a district, whereas another, the number of females. This is done manually by establishing the electrical connections between sensors and counters. If some combination of fields are to be compiled, for instance the number of white males and females, respectively, additional circuits are established manually that connect sensors for the fields “male” and “white” to a separate counter.
The cards may need to be sorted in some order or separated into stacks according to some particular field or combination of fields. In the case of census processing, for example, they may have to be separated and/or ordered by district so that statistics for each district can be compiled. The sorting box is used for this purpose. It consists of a box divided into compartments, each of which is closed by a lid that springs open when an electromagnetic circuit is closed. The lids of the sorting box are connected to the press component of the tabulator in the same way as the counters. When a particular district number is read by the appropriate sensor in the press, the lid wired to that sensor springs open. The card is then deposited manually in the opened compartment, the lid is closed manually, and the next card is read. If a different district number is now read, a different lid opens, and so on.
In present-centered language, Hollerith’s tabulating system was a human–machine system. Human intervention was necessary to transcribe the data onto punched cards; to feed the cards one by one into the press part of the tabulator; to set up the connections between sensors and counters, and between sensors and the compartment lids in the sorting box; to transfer cards from the tabulator to the sorting box; and to print out the values held in the counters. However, the time-consuming, tedious, and error-prone task of actually tallying and compiling information was automated.
VI
Hollerith was not content in being “just” an inventor. Like his great contemporary and compatriot Thomas Alva Edison (1847–1931), he was also an entrepreneur, and in this capacity he enters into the corporate history of the computer.