What I seriously wonder is, just how easy was it to write a virus in the days before protected memory? And by that I mean a virus that could do some pretty horrible things, I'm sure it was possible to destroy a disk or a printer or something by making them do things they should never get up to ...
Was it that before then people simply weren't writing malicious code and nobody even thought about safeguarding against it?
Furthermore, were banks and stuff ever running on OS's without protected memory? Because that ... that would've been awesome in its own little naive way.
Corrupting the OS's memory is rarely a problem. It's corrupting your application's memory that's a major problem, and that's still very easy to do on modern hardware with "modern" development techniques.
People in the finance industry treat C++ like they were abandoned as children in the woods, and C++ brought them to its home, fed them, raised them, and sent them to college. Someone I work with claims that Stroustrup is God.
If you're an exchange or some other data source and you want to compromise a shit-ton of machines, start sending invalid messages to the banks. You will have a botnet in about three nanoseconds.
It was really easy to erase or damage files, completely destroy the data on a disk and, on an Apple II, cause misalignment of the Disk II disk by repeatedly crashing the head against the drive armature. On a PC with an CGA/MDA you could damage the monitor power supply by putting the CRTC in a mode incompatible with the video input. Legend says you could make smoke go out of an MDA monitor.
At that time, it was also much harder for viruses to propagate - they would have to infect files and disks. Also, most IIs were booted from the application floppy and were thus completely stateless: whatever the virus did in the last session would have no effect on the next one.