Requirements now evolve so fast that a "functional useful piece of software" without a maintainer will become useless in about 5 years. Imagine trying to compile a project you haven't touched in 5 years. I know for a fact that even Go, which offers arguably the strongest backward compatibility guarantees out there, wouldn't compile my employer's codebase from 5 years ago.
I mean it's like a lot of jobs: a taxi drivers cant really pinpoint his accomplishments, a nurse will see all person they ever cure die one day, things like that.
Contrary to what someone said above, software is not a machine, it's the state reconfiguration of a machine, one of its many inputs. As a programmer all you did was align the knobs to make some inputs produce some outputs, but upstream and downstream changes will change input shape and output requirement, or the machine will change its constraints or the programmers will have figured no way to state change it.
Your contribution was important, but thanl god we dont use the state configuration you wrote 20 years ago to solve today's problems.
Sorry for a tangent: this is what I really love about some languages like Common Lisp where very old software (that may still be very useful!) still builds and runs.
You build something fast because marketing or sales wanted it, not because it was a good idea or anybody found it useful. Most people probably hated your feature.
But, OFC, giving it away for free is nearly impossible to actually make money off of.
Like the taxi driver, probably nobody will remember you unless you said something particularly impactful to someone. Someone working as a pro-bono ambulance driver is more likely to receive accolades, but that's still difficult.
This makes me want to invest time into hobbies where I create things that endure. Of course, I don't have time for that because I'm too busy building temporary solutions to what feels more and more like unimportant problems.