I doubt if more than 25% of my work-hours as a developer have gone toward any kind of program or product that went on to provide enough benefit to have been worth the effort and expense. It feels like being one of the monkeys in some kind of million-monkeys Shakespeare project. Like the system just tries random shit and sometimes something works but mostly it just wastes everyone's time (lives).

I can tell you that working with legacy turd (that will probably run for another decade or more) isn't necessarily much better.

Again the deadlines and lack of planning ahead bites. Look we just need this little feature now, it can't take long? And again, and again. You never get enough time to take a step back and re-evaluate design decisions (or, God forbid, plan ahead and do it right in the first place) and fix what wasn't done right. No, it's always this little feature that needs to be quickly hacked in, another branch, another special case, how it interacts with other past or future special cases is anyone's guess, and the code base that was already full of hastily hacked together barely-functional crap grows more and more poorly thought out crap. Tests? What tests? No tests, so if you dare go back and try quickly fix some poor design.. well, someone is eventually hopefully going to find out what you broke and give you a completely uninformative bug report. Updates? Well there's no ticket for that. And besides, something might break, because there are no tests. There's no ticket for tests.

Sometimes you might even wish that the whole thing would be discarded and done from scratch..

The code I wrote in my first job out of college is still running in thousands of locations 25 years later. But then, it was Windows software for a particular industry, not a website. Most stuff I've written since then has been replaced or will be soon.

> code I wrote professionally hasn't been superseded

Well, the state of New Jersey is looking for programmers to update COBOL programs that were written 50 years ago, so you might be underestimating the sheer power of inertia in software development...