I programmed the 7094 remotely long before I sat at its console pushing buttons. But late one new year's eve I was hanging around the computing center when John Steele encouraged me to give it a try.
The IBM 7094 console with its piano sized switch register and motor driven reset.
With a book of instruction codes in front of me I started punching them in through the switch register. This was a multi-step process: key the instruction, transfer keys to AC, key a store instruction, execute instruction from the keys.
One developed a tempo striking instructions like chords on a piano. The left hand played the opcode, the right hand, the address. After each strike, hit the motorized switch register reset and get ready to strike the next codes.
I wrote a program that simplified this process by keeping the store address in an index register. Strike an instruction, toggle a sense switch.
I remember noticing that each instruction was stored three or four times. Could this be switch bounce on the sense switches? I asked John. He admitted that in all his years with the 7094 he had never seen anyone read the switches so rapidly.
I modified my program to delay while the switches bounced. It worked fine. I wrote a few more programs that did nothing interesting.
I remember John coming by again and explaining the two data channels that occupied two boxes behind my seat. They had buttons and lights too. John punched them. Data transfers happened.