Speed as a Measure of Change in the History of Printing and Information Processing

These days with data traveling through fiber optic cables at the speed of light we may take processing speed for granted because current processing speeds are usually far beyond our perception, and probably also beyond our imagination. Instead they are statistics, to be compared with other statistics. Today we would typically not equate speed of data processing with speed of printing because they are different processes, but mainly because even when the fastest digital inkjet presses are employed, such as those used to print ballots in the 2020 presidential election, the speed of printing is so much slower than the information processing driving the press.

Printing, invented by Johann Gutenberg in 1455, should be considered one of the earliest technologies of information processing, and one of the first to accelerate the speed of book production. As far as I have been able to determine, the first report of the average speed of handpress printing was reported by Joseph Moxon in the second volume of his Mechanick Exercises (1684). This was the first true manual on printing in any language. In that work Moxon wrote of the "token", a measure of printing speed by which pressmen were then paid: 250 sheets per hour, printed on one side, by two pressmen. This speed, which in practicality may have been closer to 200 impressions per hour because of fatigue over a long workday, had not changed since Gutenberg had invented printing by movable type in the mid-15th century, and it did not change until Friedrich Koenig developed the first steam-powered platen press in 1810. Pressmen using Koenig's first mechanized press achieved a printing speed of 800 sheets per hour, more than three times the speed of handpress printing. This was the first significant advance in printing speed in over three centuries. We should also take note that the development of mechanized printing throughout the 19th century, with its steady significant increases in production speeds, did not replace handpress printing for the remainder of the 19th century. Hand presses made of iron continued to manufactured, and printers continued to use them. Larger printers had rooms where they operated machine presses or printing machines (machine rooms) and other rooms where they continued to operate handpresses (press rooms) for short editions.

How the speed of handpress printing may have compared to manuscript copying by hand in the 15th century is not a topic about which I have noticed any data. We may assume that the speed of handpress printing was well understood early on because it was used as a basis for compensation of pressmen. Far earlier than the 15th century scribes were paid by the line or by the page, and according to the quality of writing. For this we have documentation as early as 303 CE from the Edict on Prices by the Emperor Diocletian. "For one hundred lines in 'scriptura optima', twenty-five denarii; for somewhat lesser script, twenty denarii, and for functional script ('scriptura libelii bel tabularum'), ten denarii." The unit of valuation was the normal length of line in a verse of Virgil [Vergil]. Thus compensation for writing on papyrus rolls was correlated with writing speed since it would have taken more time to write scriptura optima than scriptura libelii, though actual speed would also have varied with the writing speed of individual scribes. Bearing writing speed in mind, I think it is reasonable to assume that handpress printing may have been around one hundred times the speed of quality hand copying in manuscript book production in codex form. This may be about as accurate as we can estimate the differential between quality manuscript copying and handpress printing. Considering how slow handcraft processes were in the Middle Ages, an increase of speed by around one hundred fold must have been perceived as truly dramatic at the time. How much the speed increment mattered to book producers is unclear, especially if they were paying by the "token," rather than by the hour, but we do have records of the appreciation of cost savings in printed books over manuscript copies. One of the earliest records of this occurred in 1468 when humanist Giovanni Andrea Bussi (Joannes Andreae de Bussis), bishop of Aléria, and the chief editor for the printing house of Conrad Sweynheym and Arnold Pannartz after it moved from Subiaco to Rome, wrote to Pope Paul II: "In our time God gave Christendom a gift which enables even the pauper to acquire books. Prices of books have decreased by eighty percent" (Hirsch, Printing, Selling and Reading 1450-1550 [1967] 1).

I have not taken the time to study and compile statistics on the speed of mechanical calculating devices or abacus (soroban) versus the speed of manual or mental computation in the centuries prior to electronic computing; however, I presume that some statistics may exist. Once again, the speed of computation using any mechanical device would have depended both on the advantages that the specific device may haved provided and the speed of the human using the device. There was also a famous competition in 1946 between a Japanese soroban operator and an American electric calculator operator in which the soroban operator won. This showed that the right operator of an ancient calculating device could beat a skilled operator of a newer calculating device, but we could not generalize from a single competition.

Among the earliest statistics about calculating speed in the history of electronic computing is the comparison between the electromechanical Harvard Mark 1 machine that used relays as switches, and the ENIAC that used 18,000 vacuum tubes as switches. The Harvard Mark 1, which became operational in 1944, performed calculations at one hundred times the speed of a human computer, while the electronic ENIAC, which became operational in 1945, performed calculations at 10,000 times the speed of a human computer. This thousand times increase in speed using electronic rather than electromechanical switches immediately made clear to all involved that the future of computing would be electronic, and demonstrated that electronic computing far exceeded the capacity of the fastest human computers. Within a short time the fear that the new machines built by man would replace people came to the fore, and people like neurosurgeon Sir Geoffrey Jefferson wrote reassuringly about the differences between computers and the human brain.

Even though the earliest electronic computers, such as the ENIAC or UNIVAC, were very significantly faster than people, their printed output was very slow compared to the speed of the information processing by electronic computer. The first electronic computers output data on punched cards. Or, if they used a printer, it was through a terminal like a teletype which output between 60 and 100 words per minute. Assuming 250 words per page, this speed would, ironically, have been equivalent to handpress printing, but quality of teletype was totally inferior printing on a handpress. Over time the speed of computer printers and their print quality gradually increased to speeds like 20,000 "impressions" or copies per hour on an industrial size digital inkjet press, but their high quality output remained far slower than the information processing that controled them. At 20,000 impressions per hour, roughly 1000 times the spreed of handpress printing, we can still watch the copies moving through the digital press while the electronic information processing done by the computers driving the press remains so far beyond our sense perception.

Jeremy M. Norman
November 30, 2020