Cognitive load is what matters

Large software projects built by humans will always be doomed to fail, because humans like to build the new, and nobody likes to maintain the old.

I'm pretty sure this entire thread is filled with "nobody likes to maintain the pile of ifs", since I doubt very seriously it's the age that jams people up, it's finding the correct place to make a surgical change that only produces the net-new behavior without blowing up the world. I guess the rest of that is that often the older a codebase is, the more revenue stream in impacts if something goes wrong

I've very much enjoyed maintaining or optimizing or hardening existing systems--I can just never convince my leadership to let me do that.

My current org has a terrible case of not-invented-here syndrome, and it's so easy to pitch new projects that solve something that there's already an existing tool for, or building a new feature. We would love to spend time just working within our existing systems and fixing crap abstractions we made under the deadline-gun, but we're not "allowed" to.

> [...] humans like to build the new, and nobody likes to maintain the old

I think this is certainly true at organizational scale, but most of the people I've met are change-resistant overall.

Humans are the worst programmers, except for all other programmers.