Yeah, well, I’m thankful punchcards were almost completely obsolete by the time I was old enough to start learning computers. I don’t think I’d have had the patience for that.
1 Like
I agree, it was quite a tedious task to cut out all these tiny holes to exacting standards with very fine pointed scissors, many new words were learned at the time 
6 Likes
Back in the day when you handed a deck of punched cards to an operator and went back the next day to see if it ran successfully, the worst error message you could get was “DECK DROPPED”. To make it worse, not all cards had printing across the top.
Interactive systems with remote terminals for entering input started in the late 60s. Caltech introduced one called CITRAN when I was there, and as a EE major I tried to use it a few times. Despite the brief description Caltech published at the time https://calteches.library.caltech.edu/263/1/research.pdf it wasn’t reliable or user friendly. It was not used in any of the EE or computer science courses I took, and I could never get it to work well. The first usable terminals I saw were on a CDC 6600 at the Air Force Cambridge Research Laboratory in 1972. When I worked at Honeywell Information Systems in 1973-74, they had Multics terminals, but they weren’t used in producing things like manuals. We wrote manuals on manual typewriters; the only decent electric typewriters were made by IBM, and no IBM equipment was allowed.
1 Like
But you could easily fix typos with scotch tape and a leftover cutout, without having to repunch an entire card. Those little things were all over the computer labs, and you were constantly cleaning them off yourself, like cat hair.
1 Like
I started with punch cards in 1954 (my Freshman year at Stanford). I got 75 cents an hour for feeding the 407 accounting machine and sorting cards and so on. When I finished college and got drafted the Army had me doing the same things. Except I designed systems using punched cards. Line 30,000 cards would be sent to Ft. Monroe and my system would produce all kinds of reports. It was sort of like programming in that you had to wire these huge control panels. After the Army I started as a programmer on an IBM 1401. Programs were in autocoder using punch cards. I later went to CDC and still used punch cards for my programs (3600s and 6600s). Finally we were able to use actual terminals. No more punch cards. We would use marking pens to put big stripes across the decks to help get them back in order if they got dropped. Another problem was a card jam that mangled a few. Usually easy to fix them. Those were the days.
4 Likes
Yes! This whole thread I was trying to remember how we used to keep decks in sequence. I know we didn’t use punched sequence numbers (for whatever reason), but I hadn’t remembered the diagonal stripe trick until you mentioned it.
Our first computing class in College too. Dropping off stacks to the technicians and waiting for printer output, days later. Not an exercise I can get my students to repeat. Though given the amazement shooting on large format film provokes ‘all this for one picture?’ It might be a good idea.
1 Like
Same here. However there’s one part of what we had to do that would serve them well even today:
Bench check (review) code before compiling.
We did it out of necessity. Today’s IDEs and editors will help you avoid the syntax errors. But like spell checkers, it doesn’t mean that what you write actually will do what you wanted. Reviewing your code may take time, but as my carpentry friends say “measure twice, cut once”.
1 Like
The reason we didn’t use sequence numbers because it was too much trouble. Essentially laziness. If there was a six digit space some used the first four at the beginning and when fixes and changes were made you could fill in the extra two. Required some guess work. For a deck that was changed a lot we would add another stripe across in a different color. The tops of some ended up looking like abstract art. It showed no class to just duplicate the deck to get a clean one.
We did that at the Air Force Lab for the same reason you did; the weather modeling program was long and the turnaround was a day, so you wanted to check the few cards at the start of the deck that identified whose deck it was and who would get it back.
The weather lab had a HP desktop calculator that could store something like 2000 instructions, and the scientists had me write a simple version of the mainframe card deck. It ran much faster so they could test ideas, and they were happy with it. This was in 1972, and shortly before I left the HP sales guy came by trying to sell another desktoop but the scientists were more fascinated by the HP pocket calculator.
1 Like
In retrospect, it was tedious but at the time it was magical that there was a machine that could do that at all (read the cards, do the calculations, output the results). We had no idea that so much more convenient technologies would come along later. Watching the card deck go through the reader was mesmerizing. As a Cornell undergrad in the early 1970’s, I still remember hiking up an icy Libe Slope in winter, final com sci project in a box of cards under my arm when a slip and fall just ruined my day! Kind of amazing that I’ve continued programming to this day.
1 Like
Punch cards, then later tapes, one mylar tape, your program, then a second mylar tape!
And punched paper tape. The first computer in our lab was a DEC PDP-8i that did data acquisition from one of the lab machines. It had a teletype machine connected to it that I believe punched the output on tape. We also had to punch tapes manually (don’t recall why). One mistake and you had to toss the tape and start all over.
We used to draw a diagonal line or two across the top of the deck to assist in putting it back together quickly. When a program consisted of ten to twenty BOXES of cards it paid to be careful.
The problem with the teletype tape punch is that it would sometimes miss tape frames. I wrote a modified binary punch program on the PDP-8s that fed the tape back into the reader (forming a loop between the reader and punch of arbitrary size) and detected missed frames. Written in machine language using the switches on the front panel. It worked well and saved several data sets from our experiments (it would ring a bell if there was a dropped frame).
Fun times!
The Hollerith card. I remember having to add a deck of those to my bookstore order at UConn in 1983. (This while I was doing a semester long internship implementing computer technology in a church office — that’s how I got my hands on a Lisa.)
And in the 1970s, my first job out of high school was with the Bell System as a long distance toll operator. It was a wonder of 1950s (and earlier) technology, but it also included a 1960s innovation: “Mark Sense” cards that were used to time and bill long distance calls. The customer would tell us what type of call they wanted, the number they were calling and the number they were calling from, and we would mark bubbles on the card at the speed of speech.
Every night, a driver would collect cases packed with the day’s cards and truck them to a central location, where they would be read and key-punched. The cards would then be redistributed to the local business offices, where specialists would decipher our handwritten notations and decode the time stamps on the card backs.
That one card had to carry a lot of information, in comparison with standard Hollerith cards that as Michael illustrates would carry a single line of computer code.
2 Likes
Oh yeah. Our introductory classes were to learn PL/1.
I was on a satellite campus, so in a way we had an advantage. Our tech room had two keypunch machines, a card reader/modem combo, and a line printer/modem. We could key our program, load it into the card reader, wait a couple of minutes, then watch as the line printer scrolled out pages with all our errors.
It did mean that we had a much tighter feedback loop and a decent chance to get two or three runs before our session time was up for the day.
1 Like
I thought these paper tapes were worse than the punch cards. There was a way (if I remember correctly) to copy the paper tape containing a mistake up to the point which had to be corrected, then correct, then copy again.
Wait, where’s the JCL cards? 
Well, they’re not physical cards anymore, but data files where each line of text is processed as if it was a card (and they’re still referred to as cards by many applications), but JCL is quite alive and well on even the newest z/OS mainframes from IBM.
1 Like