In the real world, where many people of varying backgrounds and skill levels are editing the same code, if the OOP method becomes a mess, why wouldn’t the functional approach also become a mess? I think that is more the point OP was making. In a vacuum with a single perfect Adonian programmer, seems like the OOP and functional approaches would becomes the same level of maintainable, because we’re in a vacuum of perfection anyway.
In my experience, the mess created using OOP is harder to untangle than the mess created using a functional approach. With the latter, it's simpler to replace a different function for any part of the logic, and the data is always just data; whereas with OOP the method is usually tied up with shared state and other functions.