A couple of posts ago there was a discussion of IBM green cards. My memory is now directed to the mid-1960's and one of the more entertaining pastimes of those we referred to as "system programmers." They were the real bit pickers, the ones who got right down inside the bowels of the system code, including the code that managed the hardware.
In those days disk drives were great big clunky old things. Look at the picture of the 2311 here (the fourth picture on the right).
The console was perhaps waist high. If you needed a different disk you opened the hinged top of the console, lifted the disk out, stored it wherever it belonged, and replaced it with the drive that contained the data you needed. Then you had a little over 7MB of different data.
The drives' cables ran under the raised flooring, and were of course longer than necessary for the consoles' current positions in case physical reallocation of space for the various devices became desirable.
The bit pickers, those who wrote their own channel control words and channel control commands, realized right away the potential for fun.
Those disks were heavy and slow, with mechanically directed physical read/write heads. It was possible to read *backwards*. You could, with the right channel control, cause the drive to stop spinning and start spinning the other way.
Now normally there would be a built in (the software) pause to allow the drive to wind down and come to a stop before rotating in the opposite direction, but some fun loving programmers wrote their own software to manage the drives, and by dint of causing a sooner-than-intended reversal of spin they could make the entire console jerk, physically moving it along the floor.
When the cats were away the mice would play, and console races were born. Two programmers would each take a drive, write their own input/output software, and race their drives across the floor. Disks would stop and reverse direction much faster than they were intended to, jerking the consoles a little, perhaps an inch or less, causing the consoles to make forward progress.
Naturally, IBM experienced a much higher than anticipated failure rate with drives. I don't know when they figured it out - or found out - but it drove them crazy for a while.