cards

Nowadays, programmers have countless resources when developing their creations. It was even before the revolution of AI and vibe coding. “Click code” is complex, but at least it is relatively comfortable thanks to modern integrated development environments (IDE) that facilitate programming in all types of languages. Not only that: programming is free, and any relatively modest PC can do it, although AI assistants have increased costs. Half a century ago things were very different, and those who dedicated themselves to programming did so with significant obstacles. There were no personal computers, access to mainframes and servers was only for the privileged and there were not even monitors on which to see how you programmed. Everything was much more artisanal and uncomfortable, and punched cards are the legacy of an era that shows that any past time was not always better. Who needs a screen? I explained it in a Foone Twitter threada technology collector and historian who recounted how programmers got by in 1962. To begin with, those programmers had a very different image than the young people who today create giant companies from scratch with flip-flops in their college dorm room or a garage. These programmers tended to be adults who also dressed in a jacket and tie: the ways were different because to access this world one had to work for large companies, the only ones where you could have access to a mainframe of the time. The example that this technological historian gave was that of IBM 7090one of the first computers based on transistors and not on vacuum tubes, like its predecessor, the IBM 709. That was a revolution in power, because the performance of the previous one was multiplied by six and the IBM 7090 managed to execute 100,000 floating point operations per second. But as we said, to program that computer there was no interface like the current one: you did not write while seeing the code on the screen. They were also not multi-user or multi-threaded systems, so only one person could use “all” that power at a time. That made these machines very precious and very expensive assets that IBM actually rented. In 1962 he rented one of these computers for a month It cost $63,500.which with inflation would be equivalent to $421,000 today. If we do a simple division (a month has about 44,000 minutes), each minute of use of that computer would cost about 10 current dollars. In a couple of hours one had spent the same amount that a good PC or laptop costs today, for example. This imposed clear restrictions when using these machines, because time was money in them. That’s where punched cards came into play, which had a capacity of 80 characters each, the maximum size of a line, although curiously the normal thing was to use only the first 72 characters and not go beyond there. The IBM template allowed you to program on paper without going overboard. To punch the cards, a special machine was used, which for example was manufactured by IBM itself and which could be mechanical or, if they were more modern, electromechanical. The idea was simple: the characters that someone typed on that machine were “translated” on the punched card, where perforations were made according to the characters on each line. To program, you didn’t sit down at that electromechanical machine and start typing commands without stopping. Instead the program was written by hand or typed. IBM had prepared templates that made it possible not to get lost and to avoid exceeding the number of characters per line. Wait, it took a while to run your program This meant that a program with all its lines ended up occupying a stack or deck of punched cards on which were all the instructions of the program, which also had to be perfectly ordered in the appropriate sequence. That deck of punched cards was given to the computer operators, who inserted them along with a task control card that told the system how and for how long it had to be executed, for example. Other programs could be in run queue (remember, it was one job at a time, and other programmers also used the same system), so it wasn’t just arriving and executing. This is what a computer program looked like in the 60s. That program could take a long time to complete its execution, so the programmer did not wait for the result to appear, but rather the operator left both the deck and the printed result in a small cubicle where the programmer could then access to pick it up. The problem, of course, is that the program could be wrong, not work or give an unexpected output. In that case, the error had to be detected, the punched card or cards that caused the error corrected, and the program run again. There were striking advances at that time such as being able to convert punched cards into stored programs on magnetic cassette tapessomething that made the reading of those punched cards faster. That was basically the process that programmers followed in their daily lives, who usually used FORTRAN or COBOL in their programs. These machines were used, for example, for the development of projects such as CTSS (Compatible Time-Sharing System), one of the first operating systems that was programmed by the MIT Computing Center. They were also used by NASA for the Mercury and Gemini space missions, and in fact an IBM 7904 was also used to run the flight planning software on the Apollo missionsbecause it had not yet been programmed for the new System/360 that had been acquired for NASA. There were also more curious applications that are still being explored today: in 1962, mathematicians Daniel Shanks and John Wrench were pioneers in using these computers for mathematical calculations and calculated the first 100,000 decimals of π. A year earlier, another mathematician, Alexander Hurwitz, used an IBM 7090 to discover the two largest prime numbers … Read more