For me, a lot of the fun evaporated when I could find others had written a dozen other programs doing exactly what I was thinking of, often better than I could have.
The rest of the fun evaporated due to (this was the 1990s) compiler churn; every year there would be new compiler versions, some of which required nontrivial changes to existing code, and the compilers often had such poor documentation that their own examples wouldn't compile. "Solving problems" turned into "trying to use broken tools", of which Microsoft Macro Assembler 4.0 was surely the worst offender.
I did my time as a sarariman writing software, but it was just a job by then.