|
Jacquard’s Web, by James Essinger, 2004, Oxford University Press.
Remember those funny cards with the little rectangular holes? Back in the fifties and sixties you would receive one sometimes in the mail with the instructions not to bend, fold or mutilate it. If you’re old enough to remember such cards, you may have wondered what the holes were for or how they were encoded. You probably knew that they were called “punch cards” and that they were for something called a “computer”. Ever wonder where the punch card came from? Who invented it, and why? Well, that is what this book is all about.
Jacquard’s Web has nothing to do with what we refer to today as the Web, but has everything to do with the information age, and how weaving fancy patterns in silk led to the invention of information processing, and hence to the the modern computer–machines that weave information. The major characters in this book are Joseph-Marie Jacquard, Charles Babbage, Lady Ada Lovelace, Herman Hollerith, Thomas Watson, and Howard Aiken. All of these people were geniuses in their own right. [Sorry, or perhaps as you might expect, Bill Gates is not even mentioned.] The book’s biggest weakness is that Essinger tries to cover the development of electronic computers in one chapter, and in his ardor to tie everything to the punch card, betrays a few unfortunate factual errors.
The book starts by taking a large step back to cover the invention of silk in China in about 2700 BC, and then how the secret of silk production made its way to France, where a silk weaving industry grew in the city of Lyons. European aristocracy created the market for not just the silk but silk brocade, woven with intricate patterns and even illustrations. The men who could weave these patterns on the hand looms of the day were highly skilled artisans.
Fast forward to the time of Napoleon. The industrial age is dawning, and the technology of machines and machinery is developing to the point where inventors are turning an eye to improving the weaving of fancy silk brocade. The hand loom has remained basically unchanged for literally hundreds of years. Enter Joseph-Marie Jacquard, the son of a master weaver in Lyons. Here in the core of the book, the author becomes almost lyrical in his descriptions of Jacquard, and again in the later chapters on Babbage and Ada Lovelace. It’s obvious he spent a lot of time reading the original documents, mostly letters, that detail the life and times of both Jacquard and Babbage. When you read this book, you feel you are there uncovering a wealth of history. After spending a moderate inheritance, Jacquard survived the French Revolution by switching sides at an opportune moment.
Now that he was poor, Jacquard needed something to make money in post-Revolutionary France. With no previous experience as an inventor, he set out to make a better loom and as the nineteenth century began, Jacquard took out his first patent on a loom—not THE loom, but an improvement on the current hand loom in 1799. Why was weaving still important? It seems that silk was still a major French export and Napoleon wanted to boost French science and industry. After several hundred years, weavers had figured out how to create lavish designs in silk brocade, but the process was excruciatingly slow. A skilled weaver could produce only one or two inches of a fancy design in fabric a day. So a tapestry eight feet tall would take months to complete, and a second would take just as long. Jacquard wanted a loom that was both faster and controlled in such a way that the design could be repeated at will.
In 1804, Jacquard patented his automatic, card-controlled loom. The problem in weaving brocade was that after each pass of the shuttle from one side of the loom to the other, the weaver had to manually adjust the warp threads up or down to control which colors would be visible on the front of the fabric on the next pass of the shuttle. It was this intricate manual process that slowed the process to an inch or two a day. Jacquard’s loom controlled the warp threads by rods that when pressed would lift a corresponding warp thread. Now the genius part—the rods were controlled by pressing a card with holes in it over the ends of the rods. Where there was a hole, the rod would pass thru the card and that warp thread would remain down. When a series of cards were tied together, they could be pressed against the rods in sequence as the shuttle made it’s trip back and forth. Jacquard’s loom did all of this and automatically advanced from one card to the next with each pass of the shuttle. There was no limit to the number of cards that could be used to encode a pattern, and thus a pattern need not be a simple repeat, but could be as complex as a painting!
As you might imagine, this invention made Jacquard a wealthy man. However, the fact that the cards contained the information of the pattern was lost upon the world for several decades. This next step fell to Charles Babbage, who not only realized the potential of Jacquard’s punched cards, but conceived the stored program computer in an age far ahead of the needed technology to build it.
The nineteenth century saw rapid progress in just about everything except computations. From the earliest invention of mathematics, the word ‘computer’ referred to a human—a person who performed computations. Skilled computers were employed to create tables of numbers that could then be used to speed other complex computations. Logarithmic tables are an excellent example. When you need to multiply two numbers, you look up the logarithm of each number and add the logarithms together. Then you find the number that has this logarithum (the anti-log) and that is your answer. The complex process of multiplication is replaced by the simpler process of addition. Logarithic tables were a great tool for the time, except that the tables themselves contained errors, and these errors would compound with other errors when solving the large engineering and navigation problems of the day. “If only these tables could be created by a machine run by steam!” was a frequent statement of Charles Babbage.
Babbage’s story begins quite differnetly from Jacquard’s. Babbage was the son of a wealthy banker and it was this fortune that was to enable him to devote his life to automating the problem of complex computations. Babbage enjoyed the good life, and his parties were famous in London. At these occasions, he enjoyed displaying what looked like a painting of a man standing at his writing desk holding a picture of a man standing at his writing desk - holding a picture... This portrait was not a painting at all, but a silk weaving showing Jacquard himself, produced on one of his looms! Babbage’s letters and journal tell often of his fascination with the Jacquard loom.
Babbage had two great projects in his lifetime. The first, the Difference Engine, was intended to calculate (and print!) complex mathematical tables with great accuracy. He did this with the technology of the day—stacks of interconnected cog wheels. The device would be constructed such that it would compute the entries in a mathematical table, while an operator turned a crank. The calulation of the specific table was controlled by the way the cog wheels interconnected, and thus each Difference Engine was to be a special-purpose device intended to compute one type of table. Unfortunately, the technology of 1830’s London could not produce enough cog wheels and the other needed parts in sufficient quantity and quality. As great a step as the Difference Engine would have been, midway through its construction, Babbage had his real genius, when he realized that by combining the punch cards of Jacquard’s loom with the computing of the Difference Engine, he could design an even greater device for general purpose computations. Babbage called this device the Analytical Engine.
The Analytical Engine was to be a real, programmable, stored program computer. In a day when electricity was barely understood, Babbage was still forced to work with the inadequate technology of his day, but his writings show that he knew what he had conceived. The Analytical Engine was to have two main parts, the ‘store’ where variables and intermediate results were stored, and a processor called ‘the mill,’ where computations were performed. The operations to be performed and the varables (data) were both loaded in the Engine via punched cards. Babbage did not hesitate to give credit to Jacquard for the idea of the punched cards. In the Analytical Engine, the cards were pressed against a frame, where they were ‘read’ by rods, just as in the loom.
Like the Difference Engine before, Babbage was unable to complete more than a few sections of the Analytical Engine during his lifetime. Babbage was limited by more than technology; he was a stubborn man, who had little to no people skills. The fact that he made as much progress as he did must be credited to Lady Ada Lovelace, the daughter of poet Lord Byron. Lady Lovelace was raised by a mother who was determined that her daughter have a real education, in a time when “education” for women consisted of sewing and manners. Ada was fascinated by mathematics, and thought that Charles Babbage might be a suitable tutor. What evolved was a life-long friendship that benefited both. Lady Lovelace (her married name) wrote several long papers that explained and expanded what Babbage had written about the Analytical Engine. Ada even wrote several subroutines for the Engine, making her, by many accounts, the world’s first programmer.
Babbage’s death in 1871, at nearly eighty years, ended all work on cog wheel computing. However, the punch card was to resume its work in just a few years, this time in the United States. Herman Hollerith was the son of German imagrants. In the 1880’s it was apparent that the country was headed for a crisis of information. It took more than four years to manually tabulate the 1880 census, and our population was growing faster each year. Some estimated that the 1890 census would require eleven years to complete!
Hollerith was hired by the head of the 1880 census to collect statistics on steam and water power used in iron- and steel-making. As he completed this task, he was given the job to create a table of life expectancies for different age groups. Hollerith quickly made a name for himself based on the quality of his work. This early work led him to believe that the “census problem” could be solved mechanically. According to the author, there is no evidence linking Hollerith’s used of the punched card back to Jacquard’s loom. In fact Hollerith’s first experiments used strips of paper rather than individual cards. Paper tape was used in the automatic telegraph, so this seemed a logical choice. However he soon realized that the tape made it difficult to find data for an individual person, and switched to using a card to carry information for an individual. To accurately encode the data on the card, he adapted the pantograph, a device of jointed parallel levers used to copy a drawing to another scale.
To read the cards, Hollerith used something new, called electricity, stored in batteries. He called the device a tabulator and the basic concept was to remain unchanged for the next few decades. In his initial design, the cards were positioned over a pool of mercury, and pins were lowered over the card. Wherever there was a hole, the pin passed through to contact the mercury. A counter for each pin advanced when that pin made contact with the mercury, recording the fact that there was a hole at that position. Hollerith’s system was used for the 1890 census with great success. Another innovation that would carry through was to cut off one corner of each card so it was easy to see that all of the cards in a stack were alligned correctly. This simple design reduced operator error by making it impossible to insert a card in the wrong direction.
As good as Hollerith was at engineering his inventions, and ensuring their accurate use in each census, he was not a particularly good businessman. The census was a feast or famine existance, yet he was actually reluctant to market his machines to businesses. He did find some success in Europe, but was overly focused on using his machines for processing census data. Since he developed the machines to count and derive statistics from census data, he could not see beyond this horizon.
Hollerith finally incorporated his business in 1896 as the Tabulating Machine Company (TCM) after winning the bid to process the 1900 census. By 1903 he found his company once again headed for bankruptcy. This time, he was forced to make a serious effort at marketing to the commercial sector. To his surprise, this market proved far greater than all the bureaus of the government. His technology was markedly superior to his competitiors, and the Tabulating Machine Company became a commercial success.
As the 1910 census was underway, Hollerith’s health was failing. At fifty he still wanted to be the inventor rather than a businessman. A millionaire investor financier, Charles Ranlegh Flint offered to buy TCM. Flint was known for his ability to put together successful mergers—sort of the Warren Buffett of his day. Hollerith’s machines now processed cards faster than the eye could see, and Flint saw this as the future of American business. After first turning him down, Holllerith, after some negotiation, accepted Flint’s offer of $1.2 million. This was quite a deal for the day especially considering that Hollerith’s business was still on the sharp slope of its growth curve.
Flint felt that TCM was too small to be truely effective, so he combined the company with three others: the Computing Scale Co. of Toledo, which made scales and cheese slicers, the International Time Recording Co. of Binghamton, which made employee time clocks, and the Bundy Manufacturing Co. of Endicott, NY, that also made time recording machines. All were manufacturers of precision machinery. The new company was called the Computing-Tabulating-Recording Company or C-T-R. Flint did not want to manage the day-to-day operations of his new company. For that he hired Thomas John Watson.
Thomas J. Watson was born in 1874 in upstate New York. He began his career as a store sales clerk, and then became a ‘commercial traveler,’ selling pianos and organs. In 1898, he joined the National Cash Register Co. of Dayton, OH. Watson was a born salesman, and thought that the job was an honorable one. He believed that salesmen should look the part, and had a zero-tolerance for drinking on the job. Inside of ten years, he was general sales manager, reporting to the company’s ultra-competitive president, John H. Patterson. Long before mass merchandising, a company’s salesmen were often the only means a company had to get its message to potential customers.
Patterson had developed a successful system of incentives to motivate his sales staff, and it was this system that Watson brought to C-T-R. Watson brought more that just a sales system; he had done his homework before accepting Flint’s offer. C-T-R was still much smaller than NCR but he had investigated Hollerith’s patents, and understood the real potiential of Hollerith’s punched cards. So certain was he of this potiential that he negotiated only a small basic salary, but with a commission of 5% of C-T-R’s annual profits. If C-T-R was successful, Watson would be too.
C-T-R was immediately reorganized. Hollerith’s low-key approach to sales was replaced by an NCR-style sales incentive program. Watson was not all sales; he understood that a quality product was needed, and he continued Hollerith’s obsession with product improvement. He also renamed the company to what we know today: International Business Machine Corporation Inc., or IBM. By 1928, IBM had grown to number four in business machines. The leaders show where the emphasis still lay in American offices: Remington Rand (typewriters) was number one, followed by NCR (cash registers), and the Burroughs Adding Machine Co. Hollerith passed away on November 17, 1929, after a leisurly retirement on his farm.
Remember, this book is the story of how Jacquard’s invention of the punch card-controlled automatic loom led the way for the information revolution. As such, there is one more critical step and that is the leap to the computer of today. In punch card history, the next player is Howard Hathaway Aiken. Aiken was born in 1900 in Hoboken, NJ, but grew up in Indianapolis, IN, in what we would now describe as a disfunctional family. He was fascinated by electricty, and studied electrical engineering at the Univerities of Wisconsin and Chicago, and eventually on the staff at Harvard. Along the way, he amassed conciderable practical experience as well.
Like Babbage before, Aiken saw the need for faster computing. The modern age required complex and repetitive computations, and the tools of the 1930’s were inadequate. While Hollerith and Watson had vastly improved the processing of information, tabulating machines were still limited to very simple arithmetic operations. Essentially, they could add, and that was about it. The science of complex computations had not progressed much from the time when Babbage began work on the Analytical Engine more than one hundred years earlier.
Aiken was known for work using the most sophisticated computing devices of the day, differential analysers and analog computers, and he was among the first to realize the potiential of electronic digital devices. In 1936, a ‘computer’ was still a human who performed computations. And those computations were still often done using tables of mathematical functions, just as in Babbage’s day. During the mid-thirties, Aiken was working on his thesis, and decided that many areas of science were at an impass because the calculations needed were just too complex and time consuming. Looking at the machines of the day, he decided that something new was needed, and developed a proposal for a machine that “provided quanto-matic computation of the four rules of arithmatic; facilitated storage and memory of installed or computed values; established sequence control that could automatically respond to computed results or symbols, and provided a printed record of all that transpired within the machine and a recording of all the computed results.”
Does this sound at all familiar? He presented his plan to the Monroe Calculating Company, the largest American manufacturer of calculators. Monroe’s director of research, George Chase, thought the machine would be one of the most remarkable innovations in the history of technology, and worked to gain approval from Monroe’s board of directors to build the enormously expensive machine. When Monroe rejected the idea, Chase introduced Aiken to the only other company with the expertise—and money—to attempt the project, IBM.
Of course things aren’t quite that direct. Where Chase actually took Aiken was to Harvard Business School, and Prof. Ted Brown, an applied mathematician, consultant to IBM and a friend of Thomas Watson. Like many great leaders, Watson was good at finding people smarter than himself and putting them in key positions at IBM. Brown was on the board of a laboratory Watson had created at Columbia University, and thus was already aware of the problems in scientific computing. Brown took Aiken to Watson.
In his proposal, Aiken discusses Babbage’s Analytical Engine as “pointing the way... to the punched card type of calculating machine... it was intended to use performated cards for it’s control, similar to those used in Jacquard’s loom.”
Aiken then discusses Hollerith’s invention of punched card “tabulating, counting, sorting and arithmetical machinery” even to emphasize the historical relation to Babbage. If Watson had a sense of history, this would be music to his ears. The result was a project at IBM with a group of IBM’s most senior engineer inventors to build what would be called the Harvard Mark I. In it’s own way, IBM called the device the “Automatic Sequence Controller Calculator.” And a magnificent device it was: fifty-one feet long, eight feet high, two feet wide and more than five tons. There were miles of wire, yet not one vacuum tube. It was capable of three additions per second, while a complex multiplication took six seconds. The logical device used was many, many relays. The Mark I was finally announced to the world in August of 1944. While it did use punched cards for input and output and Aiken considered it to be based on Babbage’s ideas, the Mark I actually lacked one important concept of Babbage’s Analytical Engine—the conditional branch. That is, the Mark I could not change the sequence of the running program based on an intermediate result. It could not do something like “IF A = 5 GO TO XYZ”. Even so, the Mark I was actually constructed and worked, so it gets credit as the first digital computer rather than the Analytical Engine of nearly a century earlier.
As we go past IBM and the Mark I, the book just falls apart. The next step in the story of Information Processing is electronics and the story of ENIAC and what became Univac, then Sperry-Rand. Unfortunately for the author, the first electronic computer does not use punched cards at all (one of the author’s mistakes) and Univac computers used metal tape as their input/output device. Because the punched card was so entrenched at that time, many Univac customers adapted IBM card readers; however, this is not the story the author wants to follow, and consequently, he devotes only one chapter to this part of the history.
As long as this review has been, there is much in this book that I have not included here. The author’s research gives great insight into the personalities of each of the major players. Along the way, we see innumerable places where history would have been altered but for those arcane twists of fate that make history so interesting. Despite the ending I still highly recommend this book for its in-depth coverage of the beginnings of our industry.
|
