My first job was doing mostly Lego-pieces greenfield development. I learned a lot, but also didn’t learn as much as in my second job where I started having to w

My first job was doing mostly Lego-pieces greenfield development. I learned a lot, but also didn’t learn as much as in my second job where I started having to work with a much more complicated, slightly “legacy” system with a big codebase. At small code sizes, OO does not really have any apparent advantages to doing everything imperatively or in a hacky way. In fact I’d argue it tends to overly complicate and obfuscate things. But there is a point at which you really start wanting OO instead because you can no longer reason about the binary as a whole, and need to start thinking in terms of interfaces with separation of concerns. Even if you did understand the whole state, there are too many people working on it to keep up with changes to the whole state in a way that lets you reason about the thing as a whole. At my first job (and in college) code never reached such levels of complexity so OO always seemed like some dumb fad to make things more complicated than necessary. I still that is often the case, but now I absolutely see the benefits when they present themselves.