A first step to mitigate some of the risk would be to move the system to a virtualised system. This could be in each location or more centralised which would make the maintenance of the fleet of old computers easier. Floppy can be copied to hard disks and will not have to worry about failures of mechanical parts involved in reading floppy drives. Developing a brand new system would take quit a lot of time. As all sys…
I wonder if anyone makes a virtual floppy drive that replicates the performance characteristics. I.e. to avoid a faster virtual drive uncovering dormant race conditions. Something like a developer assuming "I have enough time to do this processing before the disc makes another rotation" etc.
This is to get rid of the media only. You'll still be using the original compute hardware. But it would be an interesting step.
I feel that most of the desire to upgrade is cultural and not technical. People love to talk about the floppies being used while its just a small part of the equation. Cost and risk of creating a new system with the same reliability expectations is hard when the incumbent has decades of iteration. For systems that do not require more performance or energy efficiency the accounting on upgrading looks very different.