The PROCSY command interpreter was not designed to have loops so little regard was given to the need to break out of a loop that wouldn't terminate.
As I would write longer and more intricate commands I found that all the other Model 33 teletypes in the computing center's terminal room would fall silent for brief periods.
I wrote commands that would intentionally take a long time to see under what circumstances I could starve the other terminals of the computer's attention. I found that a short burst of computing might interrupt the flow of characters to one or two terminals, but, five seconds or so would silence the room.
I wrote a variation of this command that would silence the room until a given time expressed as seconds after the minute. I tried 10. The room silenced briefly and came back to life right on time. When all the teletypes resumed together it was more obvious that someone was controlling thing.
I tried 20. Same thing. I tried 40. A longer pause. I tried 60. The teletype chatter never came back.
I packed up my books trying to look more frustrated than guilty. I walked by the operations desk and asked what was up with PROCSY. Crashed again, they said. We'll reboot it.
If they had wanted they could have looked in memory to see what terminal had control when PROCSY froze and then come next door to see who was or had been seated at that terminal. The didn't. Or if they did, they didn't confront me.
Later when I worked for the computing center I got a walk through the source code for PROCSY. The various buffers were just the right size to explain the time constants I had observed.
I told another office mate about this foolish mistake. She was amused and remembered the incident many years later when she had assembled a handful of now wealthy alumni in her role as a fundraising volunteer. This was such small potatoes as modern hacks go I think they wondered why she had taken their time for the memory.