This question is prompted by a practice lab I was just doing. In it, I had a backbone router that was generating a load of /24 routes that were to be propagated throughout the BGP.
Part of the exercise was to filter certain specific routes from being propagated in certain parts of the network. But one of the side-effects of the filter in the answer book was that it would block anything that wasn't /24. So in a sense, the filter blocked more than was required, although in this case it did not make any difference because all the routes were /24. But in a certain sense, the solution was not "future proof". If the backbone decided to inject a /16 sometime in the future, it would not be propagated.
Question: when approaching such an exercise in the CCIE, should we assume that the routes we are getting from the backbone are the only ones we will ever receive, or should our solutions allow for other routes that might be injected in the future?