A friend and I were talking about a DOS POS system we wrote decades ago. It was crude, handwritten using Turbo Pascal DBF files and somehow still running in a few places since it got pirated.
We no longer have the source. No install disks. No backups. The software survived longer than our memory of it.
It made me realize how often "temporary" work outlives its creators, while the source vanishes.
What’s the most important thing you lost because you assumed you’d back it up later? Code, data, research, art, configs, anything.
Did that loss permanently change how you handle backups?
I carried the drives around in hopes of recovering the codebase for a virtualized+distributed SSI OS (Kerrighed-module-based), which had been in the works for about a year at that point.
Due to changes between 2.4/2.6 and 3.x kernel; the rise of user-level distributed computing in C/RIU, kubernetes+docker, I never really recovered the work: there were partial backups of some of the features scattered across three contributors' systems, but no coherent backup of the unifying components; and well "life" with one of the key contributors becoming 'Justice Impacted' stalled any real progress.
I use this as a personal motivation for RAID!=Backup.
I worked on it all day, then tried running it - whereupon my choice of default output filename (main.cpp) caused it to overwrite its own source code. There were no backup copies.
I told no-one, stayed late to re-write it, and demo-ed it the next day. I like to think I learned some kind of lesson, but I'm not entirely sure what.
Result? I learned how to use make. (Borland had a nice one). I also started using my first VCS tool, sourcesafe (pre-Microsoft).