A cautionary tale from the punch-card era of mainframe computing, as told in The Register's reader-contributed "Who, Me?" column. A former Sperry Univac UK employee, called Mark, recalled a weekly inventory job that took five or six days to run, including printing. Three senior managers had flown in from the US to observe the important job.
The system had two printers, with one dedicated to the inventory run. When a smaller job went wrong and began generating far more pages than expected, Mark checked with the programmer and got approval to cancel its print stage. He did so — but the printer that stopped was the one handling the inventory job, not the runaway workload. The entire output was lost, including the summary pages.
The consequences were immediate: the visiting managers could do nothing until the job was rerun, which meant waiting until the following week. Mark said everyone was sympathetic and no one blamed him, but he left an ominous "but…" hanging. The story is a reminder that in shared computing environments, a simple action can have outsized consequences when the wrong target is selected.