I mean there are different kinds of bloats. Some is justifiable, some is not and some is just a symptom of other problems (the last 2 are not mutually exclusive), like mismanagement, incompetence (from management, developers, team leads, etc). This is somewhat similar to cholesterol, there are different types of cholesterol, some might be really bad, some might be harmless, etc.

Bloat (you mean here code duplication?) can be both cause or a symptom of some maintainability problem. It's like a vicious cycle. A spaghetti code mess (not the same thing as bloat) will be prone to future bloat because developers don't know what they are doing. I mean in the bad sense. You can still be not familiar with the entire system but if the code is well organized, is reusable, modular, testable, you can still work relatively comfortably with such code and have little worries of introducing horrible regressions (in a case of a spaghetti code). You can also do refactors much easier. Meanwhile, a badly managed spaghetti code is much less testable, reusable, when developers work with such code, they often don't want to reuse an existing code, because the existing code is already fragile and not reusable. For each feature they prefer to create or duplicate a new function.

This is a vicious cycle, the code is starting to rot, becoming more and more unmaintainable, duplicated, fragile, and, very likely, inefficient. This is what I meant.