Sometimes you have to NIH to understand why things the way they are. Granted, it's hard to justify on the company dime. From the perspective that 99% of software is awful and writing more is making things worse, I'm sympathetic to your position.
On the other hand, in my little corner of the world I've seen plenty of damage done by an "invent-nothing" mentality and an approach to software architecture that permits only plugging together popular open source components with fancy websites and corporate backing. For lack of a better term I call it "noun-driven development". You needed a disk reachable through an ssh server and an Apache directory listing, you ended up with Artifactory. You needed a library to replicate data with a Merkle tree, you ended up with someone's Kafka + etcd + memcached + redis + glusterfs Frankenstein monster, etc.
