Punched Cards? Sheer Bloody Luxury

I remember being astonished when my dad told me how they handled punch card accuracy at his office: engineers would write out their programs on (special) paper forms (i.e. a box for each character on each line), and hand the forms off to the computer department, where they'd have two different keypunch operators key in the program. Then they'd feed the two stacks of cards into a special punched-card-comparator machine, which would tell whether the decks matched. If so, it was considered successfully transcribed (and one of the decks could be thrown away).

I also remember reading a story long ago about decks of cards containing scientific programs being sent by rail between cities in two different countries in Europe (IIRC), and they kept having problems with the programs not working. Finally, they sent a courier along with the box of cards at one point. And, as the train crossed the border, the customs inspector boarded the train and checked passports and such, and inspected goods that were being transported, and, as is customary, took a sample from many of the transported goods (as one might take a bit of grain, say)... and, yeah, they took a couple random cards out of the box. Problem identified, if not immediately solved.

I was really lucky back then - I was a high school sophomore who was given free rein of the local college's small computer lab. I got really good at keying in the IPL (initial program load) sequence on the front-panel keys, triggering a read of the IPL card from the card reader, which then would boot off the Winchester drive.

I got to show my daughter an IBM 029 card punch at the Seattle Living Computers Museum a few years ago. Don't think she was too impressed :-)

Plus, you would have to manually enter the sequence numbers of the new cards, waaaay over in column 73, which was really slow and tedious on a card punch. Especially when you were in a rush because you were granted a brief slice of time on the computer, which was not capable of multiprogramming.

You could always duplicate the deck and re-sequence from scratch, but that was done on a machine that did not print (interpret) the characters at the top, making it impossible to read and manually maintain the deck.

The solution was: just don't drop the cards. I wrote and maintained a 4,000-card assembler program on a Univac 9300 in the 70's, and I never dropped my cards. The program was stored in two 16" metal trays.

My high school was smarter. After I "hacked" the computers there - they put me in charge of maintaining them - I'd get called out of class to run around campus fixing computer things up. I didn't realize it at the time, but that was a forward thinking administrative office.

After a couple of such assignments, they turned us loose on the terminals. Those who were quick/lucky got one of the few VT100s or the DECwriter III, but any terminal was better than the mark-sense cards. I think that was a weeding-out exercise.

At my first job, my boss told me the story of a colleague who was taking a box of cards out to the car. You had to drive your program to The Computer, of course, because this was back when there was a The Computer. It was out by the airport.

They took one step out the door, and just at that moment a foot of snow decided to let go and slide off the roof. Oof.

Clearly this author worked on one of the gigantic, luxurious MVS boxes. I, on the other hand, toiled away relentlessly on a small VSE Box. He had it easy!