There’s no formula for making great games. Especially if it’s a brand new IP or a first in the series, developers are essentially just experimenting with any old idea that seems vaguely good to see what works, all the while trying to put out fires that keep popping up because the game engine keeps exploding every five minutes.
It's a mess, and often even the people in charge don't have any idea how a project is shaping up until its final few months.
In the same way that movies are described as “miracles” because of just how difficult they are to make, getting a game right is equally as hard, and even the very best ones have certain stretches, levels or sections that come close to tanking the whole thing.
They're the kinds of missions that stop you from wanting to go back and revisit them, because if they were bad at the time, they'll be even worse now.
They don't completely overshadow or really take too much away from the grand heights those games are still held up to, of course, but they're still met with a groan every time they're even mentioned.