There may be a few minor code changes (less than 1000 lines per data loss) that got lost over the last few decades. And i don't have my earliest pices of garbage code from the 8 bit and DOS eras any more.

But for the last 20+ years i've been very meticulous when it comes to tracking code i care about. There's a great deal of code i deliberatily don't backup (i have a src/temp directory), this is mostly one-off tools, test code, or code i write live to teach an old collegue new tricks. If THAT stuff goes away, it's a bit like someone sneaked into my house and cleaned the junk out of my attic. Yes, there may be a certain sentimental loss, but i knew i didn't need to hold on to that stuff anyway...

All tracked code is also auto-synced to multiple servers in different locations, just to make sure even a house fire doesn't automagically wipe out my rather huge codebase. With all the ~270 repos on my server (and the the correspondig default base databases for the projects), i'm currently at about 30 GB of code backup.

In the last 10 years, i only had one major loss, and that was of an OpenSCAD project that took me a few days to design. I accidentaly deleted the wrong project, so i was "forced" to design something much better in a fraction of the time.

To be clear, most of my project stuff in my personal mercurial SCM doesn't need to survive me. It's all my hobby stuff, as well as some projects of other people i so regularly use in my own stuff that i started to implement my own non-public modifications to it. All the open source stuff that is also used in commercial applications by my employers gets additional, in-company mirrors.