My counter part at work told me he was among the first PLC-5 users to use MSG instructions I believe the story goes, perhaps a certain but documented usage that he claimed altered ladder logic without faulting the PLC. This was immediately fixed as a bug, not something many customers ever saw.
I have never seen a Allen-Bradley PLC in run mode have it's code altered (aside from the multi-step RSLogix5/500/5k Online editor method) or fail in any way other than the sole red LED and complete fault mode all all outputs. And, frequently it will clear it's own memory.
My former employer used hundreds (perhaps 500+) of the PLC-5.
I saw one failure among the 200 or so of them in three departments.
It failed due to a flood that literally drowned 7 machine panels with a waterfall type flood, right through the big square holes were cut out to clear the wiring to the servo drive brake resistors on top with expanded metal covers.
The other six machines using with that setup: A 5/40L +16 slots of AC I/O...came out fine after being dried out and powered up again. I had to reload two of them, and one of them I had to go in serially to revive the DH+ ports which showed to be disabled. (Yes, all of them, but hey, they didn't drown, just turned off).
The only time I ever saw a dry one with a known fresh battery dump it's program was due to a failing power supply that eventually died six months later when the machine dumped its program and went to FAULT status the 2nd time.
The PLC-5 is the Muhammad Ali of PLCs in my book. If that bit changed, it happened for a perfectly explainable reason.
Prior to the PLC-5s, 19 of my machines had the ISSC IPC (310 I think) controller. It could, and did, literally go apeshit mad and show ladder logic with open branches on the CRT:
I am at desk, reading email. Operator enters the lab and says "Hey, Paul, this machine over here is going nuts?"
I am thinking yeah, whatever, and when I ask him what it is doing as I follow him to the door to he floor, he says "going nuts" which kinda pisted me off.
So I go over there, the operator resets the e-stop, and every motor, solenoid valve, indicator light, you name it started randomly going off and on.
This was a big complicated tire assembly machine with lots of screw drives and appliers, conveyors, web handlers, around 500 I/O probably 60 motors.
All of them going wild. I opened the panel door, and the outputs in all three racks LEDs (plc driven) were all flickering like Christmas lights. I Fonzied it once on the big aluminum housing, and it quit, went right back to normal. I found the watchdog module* working and jumpered, so I yanked the jumper, they built a tire. It was fine. I tried to make it fcuk up again for 30 minutes tugging twisting and looking for loose plates...I couldn't make it screw up. I shrugged and said run it. They nicknamed me The Exorcist for the day, and we left it at that.
*The watchdog timer was a capacitor timer powered by the backplane the was reset between scans. This timer took up one slot and was to be installed in the last slot in the parallel I/O bus. If the logic stalled, the timer would open a contact you wired into mid 70's era e-stop strings.
The stuff was old and often flaky and suspect and left jumpered by the floor guys.