Feb 5, 2020

History of the punch card Welcome to the History of Computing Podcast, where we explore the history of information technology. Because understanding the past prepares us to innovate (and sometimes cope with) the future! Today we’re going to cover the history of punch cards. A punch card is a piece of paper, or card stock, or card, that holds data. They look like two index cards next to each other with a bunch of holes in them. The data they hold is in those holes. It’s boolean, with a true or false represented by a hole in a predefined location, or the absence of a hole - simple as that. The logic is then interpreted by a language, often one that was specific to the machines that each ran. My grandma used to configure punch cards and I remember seeing some of these when I was a kid and being awestruck. So they’ve held a fascination for me since what seems like the beginning of time. But those punched cards didn’t start out used for processing data. Or did they? The weaver Basile Bouchon then built a loom that could be controlled by holes punched into a paper tape in 1725. He was storing the positions for colors and patters on the loom with cards. And so saving time for humans by using the positions of those holes. You can call this computational memory. The holes controlled how rods could move and the positions were stored in the cards. And so the first memory came in the form of cards of paper that stored data. Much as there was already data stored on paper, in books. And before that tablets and papyrus. The design was improved by his assistant Jean-Baptiste Falcon and by Jacques Vaucanson. And ultimately isn’t programming just putting data into storage. So let’s say the first programmers hacked language by putting data into temporary storage called our brains. And then written languages. But now we were putting data into storage using a machine. Not just moving gears to calculate. The merger of calculation and memory would some day prove to be a pretty fantastic combination. But we weren’t there yet. Although these improvements controlled the patterns woven, they still required a human to operate the loom. In 1804 Joseph Marie Jacquard took that next step from stored memory to adding a mechanism capable of automating the loom operation. And so the Jacquard Loom was born. A bunch of 9 inches by 1.25 inches by 1/16 inches of punched cards in stacks. They were linked into a chain and read to build patterns. Each card held the instructions for shedding, which is moving the warp up and down) and setting the shuttle for each pass. You recorded a set of instructions onto a card and the loom performed them. Then comes the computer. The originals had gears we set to run calculations but as they became capable of more and more complex tasks, we needed a better mechanism for bringing data into and getting data out of a computer. We could write programs and these cards became the way we input the data into the computers. We loaded a set of commands and the device printed the output. And then Semyon Korsakov comes along and brings punched cards to machines in 1832. Charles Babbage expanded on the ideas of Pascal and Leibniz and added to mechanical computing, making the difference engine, the inspiration of many a steampunk. Babbage had multiple engineers building components for the engine and after he scrapped his first, he moved on to the analytical engine, adding conditional branching, loops, and memory - and further complicating the machine. The engine borrowed the punchcard tech from the Jacquard loom and applied that same logic to math, possibly with a little inspiration from Korsakov. He called them “Number Cards.” Carl Engel adds to the concept around 1860. Come 1881, Jules Carpentier brought us the harmonium using punch cards, converting those little grooves to sound. And then. Well, then comes Herman Hollerith and the 1890 US census project. He’d tried out a number of ways to bring in all the data from the census to a tabulating machine and I think he knew that he was on to something. He then went on to work with the New York Central, Hudson, and the Pennsylvania Railroads to help automate their data processing needs using these punched cards. At this point, those cards were 12 rows by 36 punch positions and 7 3/8 inches wide by 3 1/4 inches high by .007 inches thick. He chose the size based on the standard size of banknotes in the US, so that he could store his cards in boxes that had been made for the Treasury Department Based on these early successes, he was able to successfully read the data on those cards into his tabulating machine and so he founded the Tabulating Machine Company in 1896. The most important aspect of Hollerith’s contributions was actually bringing us machinery that could process the data stored on those cards. That company merged with a few other similar companies to join forces bringing in Thomas Watson to run the company, and in 1924 they became International Business Machine Company, or IBM. And so the era of unit recording machines begun. G. W. Baehne published Practical Applications of the Punched Card Method in Colleges and Universities in 1935 showed plenty of programming techniques and went through a variety of applications for use on the cards. As with any real industry there was competition. Remington Rand also began building punch cards and readers, along with others in the industry. And by 1937 IBM was running 32 presses at their Endicott plant where they print and sorted 10 million cards a day. By World War II, English cryptanalysts at Bletchley Park they ended up with over 2 million cards used to store decrypted messages, including those tabbed out of the Enigma. We were trying to automate as much as possible. Contracts, checks, bonds, orders out of the Sears catalog, airline ticket entry. And suddenly loading computers with punch card data for further processing was becoming critical to the upcoming need to automate the world. Punch cards had become a standard by 1950. Those IBM cards had said "Do not fold, spindle or mutilate” and many a bill would come with a card and potentially be used for processing when returning the bill with a check. Now, back in the 30s, Remington Rand and IBM had gotten in trouble for anti-trust by forcing their cards in their machines and by 1955, IBM was owning the market and you know, the innovation and automation of the country couldn’t be left to just one company. So Thomas Watson Jr was forced to sign a deal that IBM would drop to not any more than half of the manufacturing capacity of punched cards in the US. But we were already to go past the punch card. And computers couldn’t be programmed using jumper cables forever, so we started using punch cards. there were a lot of file formats and other conventions that were set in that era, that still trace their origins to the 80 column of text. And the programmers of those cards began to ask for cards to be printed that could support functions, to make their jobs easier. These were used for the GE 600 and other vendors, and Univac had a format, and with languages like FORTRAN and COBOL having come along, generic punched cards became popular. And the UNITYPER came along, giving us magnetic tape in the 50s. Then in the 60s we got an easy magnetic type encoder and it wouldn’t be long until we got computer terminals, light pens, and minicomputers. By then it would take years for older tech to be unnecessary. The dimensions would be set and standardized for the RS-292 punched card but the uses would be less and less and less and less. And so punch cards had survived the transistorization of computers. But not newer and better forms of input and output. Tape ribbons would sit in drawers in places like MIT and Stanford. In fact some of the first traffic to run over the Internet precursor DARPANet would be using those tape ribbons to write output. The last bastion of the punched card was electronic voting, which had begun in the 60s. But then, the State of Iowa basically banned punched cards in 1984. Their use had been shrinking over time but at this point it was time you could say that the punched cards are obsolete. That doesn’t mean they weren’t being used, more that they just weren’t being used to build much new tech. I suppose that’s how I ended up getting to play with some that my grandma brought home. I don’t think I had a clue what they were actually for at this point. The punch card then gave way to programming with paper where you filled in bubbles with a pencil, but that was a stop-gap for dealing with an era when computers were starting to become common, but there weren’t enough for entire classes to learn programming. So the punch cards gave us what we needed to get input and output to these early computing devices in a time before truly interactive computing. And the were useful for a time. But once keyboards became common-place, they just… weren’t needed as much. And it’s good because otherwise we might never have gotten object oriented programming. And loading large programs with cards was never very fun. But they had their uses and got us to a time when we didn’t need them any more. And so we owe them our thanks. Just as I owe you a thank you, listeners, for joining me on this episode of the history of computing podcast, to chat alllllll about punch cards. I hope you Have a great day!