This is quite hard to grapple with fresh out of school, where you worked mainly on pristine implementations of eternal algorithms, proven to be optimal in multiple ways, elegantly crafting each loop and feeling accomplished by expressing the algo with even higher clarity, reading and idealizing Knuth etc.

Then you see the real world and you think it must be because people are stupid, the bosses are pointy-haired, they don't understand, they don't value elegance, they are greedy, they etc. etc. But once you spend enough time on your own projects, and they evolve and change over the years, they turn more and more into a mess, you rewrite, you prioritize, you abandon, you revive, and you notice that it goes much deeper than simply laziness. Real solutions depend on context, one caching algo is good when one medium is x times faster than another, but not when it's only y times faster. It makes sense to show a progress bar when downloading a file if the usual internet speed is X but not when it's Y. Over years and decades the context can shift, and even those things can change that were only silent assumptions of yours when you made the "perfect" program as a young'un, looking down on all the existing "messy" solutions that do a bunch of unnecessary cruft. It's an endless cycle...