An example of the first came in the late 18th century with the Napoleonic reforms following the French Revolution. Archaic structures and practices were to be swept away and replaced with a ‘modern’ system of government. A new decimal system of measurement was introduced and cartographers employed to help map France and contribute to a new and more equitable property tax, the dominant source of revenue. This was a time of the kindling Industrial Revolution and early attempts at specialisation, the division of labour that Adam Smith had highlighted in his recently published Wealth of Nations. The computational work needed to produce logarithmic tables in decimal form was split into subcomponents and organised along factory lines. Unfortunately, after almost ten year’s work and with the tables nearly completed, the fiscal problems of the Napoleonic regime meant that publication and printing of the information never took place. The idea of segmenting the different tasks in calculation and using methods of differences to shortcut the laborious arithmetic work did, however, cross the English Channel to reach Britain.
The individual who promoted these ideas has become known as the inventor of the computer. Charles Babbage was born in Devonshire to a relatively wealthy family, where he grew up to develop an obsession with mathematics. In 1810 he entered Trinity College, Cambridge, to pursue this interest. During his time at Cambridge, he was to learn firsthand the gaps in existing knowledge. In checking calculation tables used by the Astronomical Society, Babbage quickly discovered that the tables were riddled with computational errors resulting from human mistakes in computation, copying or printing. The errors tended to be compounded over time as subsequent tables used the existing body of work as their starting point.
Babbage and his engines
Babbage believed a machine that could perform the calculations and print the results, thus removing inevitable human errors from the process, was what was needed. What Babbage proposed was a machine that could do mechanically what had been achieved in France through recourse to a battery of individuals. It was the replacement of laborious and error-prone manual calculation that lay behind Babbage’s idea for the ‘difference engine’. The engine was to mechanically recreate the known properties of numbers, in particular the relationship between numbers multiplied by themselves, or powers. These relationships enabled users of figures to perform complex calculations with a simple checking mechanism. The trick was to construct a machine capable of performing the calculations. To assist in obtaining funding, Babbage constructed a small working prototype that he marketed with the support of the Astronomical Society (which he had helped to found). With the backing of the society, Babbage appealed to the British Treasury for financial support. The appeal was based on a shrewd knowledge of the British government’s reliance on the merchant and Royal navies. Shipping required maps and navigational tables. Their accuracy in turn relied on calculations being correct. The government listened to Babbage’s overtures and the Exchequer awarded him the sum of £1,500 (roughly $1m) to develop a ‘difference engine’.
Interestingly, this was awarded around the same time that proposals to improve the system of communication between the Admiralty and Portsmouth through the use of electric signals (the telegraph) was being turned down by the same government. In a pattern that has proven to be endemic to commercial scientific efforts, Babbage soon found that his initial estimate of the cost of development was too low. What he needed were both designs for the working mechanisms and skilled operators who could produce the equipment to the necessary level of precision.
Babbage’s initial grant of £1,500 towards the total cost of £5,000 ($4m) to build the difference engine was not unqualified. Many eminent members of the scientific community publicly disparaged his efforts. What Babbage required was a publicity campaign to provide reassurance to those providing the capital to support his efforts, which, in this case, was the UK government. He did little to win the support of public or scientific opinion at large.
The adverse comment on his work began to undermine his financial support. Babbage was forced to fall back on his personal inheritance of some £100,000 (nearly $70m). After spending £6,000 of his personal funds, he persuaded the British prime minister, the Duke of Wellington, to advance a further £1,500. This was followed by further lobbying and two additional allocations of £3,000. But even after spending over £30,000 ($20m) of his and the government’s money, Babbage still had not produced a final working version of his machine. Eventually, after disagreements over money, his chief engineer walked out, taking with him some of the equipment and components of the engine.
What Babbage had in 1833, therefore, was a small working prototype and the expertise he had gained from his efforts over the years. He no longer had government support, nor in many ways did he have the desire to continue the quest to build the engine. The question of whether it could fulfil the tasks he had specified had been resolved by his efforts, and those of a Swedish scientist named Georg Scheutz, who had built a scaled-down version of the difference engine from the reports he had received from England. Once the machine had shown it could do the job, the challenge receded for Babbage. Work on constructing the difference engine effectively ended in 1834. The challenge of physically constructing the difference engine was replaced by that of the theoretical construction of a machine of even greater power, which Babbage named the ‘analytical engine’. Wellington declined the opportunity to provide public finance and Babbage was left to fund the development work on his own, supported by voluntary contributions. One of these came from the unlikely source of Augusta Ada, the young daughter of the poet Lord Byron. Ada proved to be an adept mathematician and was absolutely entranced with the work of Babbage. She was to provide Babbage with emotional and scientific support in his quest to design the analytical engine. When she married William King, the future Earl of Lovelace, her husband also supported Babbage.
Plans were drawn up for the construction of the engine. The engine would have various component parts to perform separate tasks. There was a ‘store’, which would contain the variable upon which operations were to be conducted, and the results of these operations. There would be the ‘mill’, into which the variable would be brought for the operations to be conducted, and finally sets of defined formulae for the computation of known tasks. The design thus predated modern computers by over a hundred years, these three functions equating to what we now know as the memory, central processing unit and algorithms respectively. Babbage also suggested the use of punch cards to control the general operation of the system and the definition of the variables.
The fate of the analytical engine was to mirror that of the difference engine. History has shown that both machines would have worked, but unfortunately neither was ever fully constructed. Both were interesting scientific curios, but failed to attract sufficient funds to be developed. The development of digital computation would not be resurrected for some considerable time after Babbage stopped his work. Why did it not get taken further? Perhaps Babbage’s personal wealth blinded him to the need to maintain the support of his principal backers. Perhaps also the lack of urgency stopped him producing a working prototype that the customer could actually use. Whatever the reason, his inability to maintain confidence in his work left Babbage viewed as a sad and ridiculed figure by some parts of the population, although still a respected if idiosyncratic scientist to others. It was only well after his death that the pioneering nature of his work was widely recognised. His obituary described his engines as examples of ‘noble failure’.
The cash register rings up
If Babbage’s machines were stillborn, the same could not be said of a successor machine, the cash register. The cash register was first developed by James Ritty in 1879. It was developed in response to Ritty’s frustration with thefts by staff in his restaurant and was given the less than catchy title: ‘Ritty’s Incorruptible Cashier’. Few of these new machines were sold, although one customer, a struggling coal merchant named John Patterson, was to play an important role in spre
ading its use. Patterson was sufficiently impressed with the potential economic value of the register business that he sold all his coal interests to fund his expansion into the new mechanical device which displayed and recorded sales totals.
What Ritty lacked, but Patterson did not, was a devotion to selling. In 1884, Patterson bought Ritty’s company and renamed it the National Cash Register Company. He quickly introduced a wide range of innovations and incentives for his sales force, copying and improving on ‘best’ current sales practice. Salesmen were heavily incentivised through high commissions; training was introduced, sales literature was prepared and direct marketing to potential customers employed to generate interest. In addition to a professional and aggressive sales force, NCR sought to provide customer support through the establishment of a retail-distribution chain. Patterson also realised that the technology could not remain static. He therefore set up a department charged with the task of constant improvement to the registers. For the first 40 years of its life, NCR enjoyed phenomenal success in manufacturing and selling cash registers. It ruthlessly protected its dominant market position, even setting up a subsidiary to protect it from the companies sprouting up to sell secondhand registers.
In 1903 a successful young salesman named Thomas Watson was called to NCR headquarters where he was informed that he was to run NCR’s secondhand operation. The principal goal of this operation was to undercut its competitors and put them out of business. Watson proved highly successful in this role, and was rapidly promoted to the point where by 1910 he was effectively Patterson’s number two. Unfortunately for Patterson and Watson, the tactics that had been used to protect their market became the subject of an investigation instigated by one of their competitors, the American Cash Register Company. In 1910, with antitrust feeling running at its strongest, Patterson and Watson were found guilty, fined $5,000 and sentenced to one year’s incarceration.⁷⁷ The verdict was overturned on appeal. In the meantime, public opinion had begun to swing behind the company as a result of its humanitarian efforts to counter a local natural disaster.
Relief at this reversal was short-lived for Watson – soon after, the axe fell on him, just as it had on so many other of Patterson’s protégés. Watson took his sales training and expertise to the Computing-Tabulating-Recording Company (CTR) and set about developing a business to rival that of NCR. Unlike NCR, whose client base had been in the commercial sector, the business that Watson joined had its genesis in the public sector, in the dull but lucrative business of census enumeration. NCR meanwhile continued to go from strength to strength, culminating in a massive IPO in 1924 which despite its size was five-times oversubscribed, attracting subscriptions of over $250m ($6bn). This made it the largest new issue in history at the time.
Big business in counting heads
The need for increased computational power that Babbage had sought to address was a real one. Whether it was nautical tables, the rising analytical demands created by the Industrial Revolution or the needs of government, the pressure to reduce costs and increase the speed of data processing was intense. In Britain, the business sector mainly sought to refine existing methods to increase efficiency. In America, the approach was different. Most companies had no historic baggage in their operations – in the terminology of today, they did not suffer the problems of ‘legacy systems’. In the event, the first large-scale use of mechanical aids in the computation and tabulation of figures came at the behest of the US government. Like Napoleon in France nearly a hundred years before, the root of need lay in the government’s appetite for revenue.
Fundamental to government was a knowledge of exactly who was being governed, which meant holding a regular census. Census taking was a massive data-gathering and processing exercise that took many years to complete, meaning that by the time usable information was available in tabular form, it was likely to be out of date. As the population of the industrialised world was growing rapidly, it was simply not feasible for the process to be sped up without recourse to mechanical aids. To give an idea of the growth, the 1840 census in America recorded a population of just over 17 million people. The census was conducted by 28 clerks. In 1860 the equivalent numbers were 31 million people and 184 clerks; ten years later, 38 million people and 438 clerks. By 1880 the population had reached 50 million, and the census took 1,495 clerks and seven years to produce.⁷⁸
It was against this background that the US Census Bureau under Robert Porter organised a competition to evaluate potential alternatives to the laborious manual tally process that had been employed in previous censuses. The trial took place in St Louis. It featured two similar systems, one that used a colour-coding system to speed up the manual process, and the second a punched-card system which tabulated mechanically. The speed with which results could be tabulated once the cards had been completed proved the decisive factor. The card-based system won the competition comprehensively. The Hollerith Electric Tabulating System was awarded the contract for the 1890 census, and the company subcontracted the production of equipment to the Bell subsidiary Western Electric. The 1890 census proved a success. It was estimated that the cost of producing the census, nearly $12m ($865m), was roughly a third lower thanks to the new machines. The time to publication, two and half years, was nearly two thirds less than the time taken to publish the results of the 1880 census.
Although the reduction in the time to completion was undeniable, the cost saving was later challenged. As the estimate was provided by Porter, it was not unbiased. The fact was that, for all the hypothetical saving, the cost of the 1890 census was still double that of its predecessor ten years earlier. Whatever the truth of the cost estimates, the Hollerith system certainly revolutionised large-scale data collection and tabulation. Henceforth data processing need not have its boundaries set by the limitations of manual tabulation. The success was recognised globally, with enquiries and contracts being received from as far afield as Scandinavia, Central and Eastern Europe.
Setting Up Shop in Georgetown
‘A menace to the health of the men…’
“Woe, especially, to the worker who slacked off on the job. Hearing of an employee who spent an excessive amount of time reading the newspaper in the lavatory, Hollerith devised his own solution to the problem. He drove nails upward through the toilet seat, filing them smooth where they emerged. From their heads protruding underneath he ran wires back to his office nearby, where they were attached to a magneto that sat on his desk. Watching through a peephole from his office for the malingerer to take up his reading habit, the inventor gave the magneto crank a sudden turn, sending a shock along the toilet seat.”
8.1 – More than one way to stimulate productivity
Source: G. D. Austrian, Herman Hollerith: Forgotten Giant of Information Processing, New York: Columbia University Press, 1982.
The race to find other uses
The immediate problem that Hollerith faced was that despite the success of his apparatus in the census, a business could not be sustained by waiting ten years for each new contract and having the equipment lie idle in between. Although orders were received from foreign governments, it was clear that alternative uses had to be found and the equipment modified if the company was to be commercially viable. The economic backdrop to this was not auspicious. The early 1890s were characterised by severe economic conditions and widespread business failure. It was against this backdrop that Hollerith had to persuade companies to invest potentially substantial sums in a relatively unproven set of equipment which would assist in tabulating information.
Hollerith concentrated on the dominant companies of the time: the railroads. The railroads had huge data management requirements relating to goods shipments, receipts and timetabling. The need for improvement in their ability to handle and analyse data was self-evident. What Hollerith had to do was persuade them that his equipment represented the solution. First, though, he had to modify his machines to make them suitable for the business community. The system he devised us
ed one machine to punch the cards, a second to process and tabulate results, and a final one to sort the cards. The machines were powered electrically but operated mechanically. They depended heavily on the card system employed. Major revisions were made to the card to make it intelligible to the operator and a logical storage medium that could be easily manually interpreted if necessary. This was achieved by the introduction of different data fields categorised by type.
With a prototype in hand, Hollerith then sought to persuade the railroads to adopt his equipment. Not surprisingly, given its unproven nature and the underlying business conditions, his early efforts were not successful. Eventually he managed to persuade the New York Central to take the equipment on a trial basis, but only after he offered it at no cost to the railroad. Unfortunately, teething troubles led to the railroad jettisoning the equipment after a relatively short space of time. Hollerith had been in Europe negotiating the sale of census equipment to the Russian government at the time, and he returned with a pressing need to modify his machinery and restore confidence.
Once he had dealt with the technical problems, Hollerith returned to the New York Central with a proposal to install and operate the machines for 12 months free of charge and with no obligation. Given that his financial position was precarious, with his company only sustained by the sale of assets and borrowing, this represented a final throw of the dice. Before long, however, Hollerith was approached by the Library Bureau in Boston, offering to act as his agent for his non-library clients. Libraries were well-versed in cataloguing techniques and more able to see the potential of his machine.
Engines That Move Markets (2nd Ed) Page 39