My answer for chapter 3 exercise 6

[whitestork26]

In the long run I see a fundamental disadvantage for the procedural approach.

Assumed that you have a perfect and bug-free implementation in a procedural programming language. But one day you have to enhance your code to support a new feature (e.g. new vehicle type). In this case you have to touch and change a (up to now) bug-free code. In more complex implementation projects this might be a risky thing as probably enter some bugs with your new code that even influence and invalidate old matured features.

Whereas within the object-oriented approach you add the new feature without the need to change anything in your existing code. Either you make use of behavior already available in the super class. Or, if that does not fit with the special behavior of your new class, you (re)define only the behavior of the new class without changing anything for your existing classes. Means that you do not invalidate any of your existing features.

So in the long run the object-oriented approach eases the enhancement of your stuff and limits the risks doing so.