Peter Molyneux has plenty of flaws, and while his overstatements are among them, I just don't think he's a very good game designer.
Let's take the first Fable into consideration. The whole aesthetic, from the literal art style to the whole storybook mock-up approach, was very charming. The game was nothing if not inviting. But the mechanics, while tight, are almost as shallow as mechanics can be. For instance, combat revolved around a very limited amount of functions, which would be fine if those functions had diverse applications -- but they didn't, and winning combat was a case of finding the square hole for the square peg. And I can't see why Molyneux couldn't see this from the design stage itself, not just the execution.
If one intends to make a game to last the ages, isn't the absolute first thing to perfect going to be the mechanics? A work of digital media can be excellent without mechanics or with few mechanics or even with shallow mechanics, depending on the work (such as The Walking Dead). But a truly excellent game, in the most traditional sense of the word, must have mechanics that are centre-stage in contributing towards the user's experience. Fable's mechanics were indifferent towards itself, so much that another set of real-time RPG interaction mechanics could have been put in and nothing would have been lost.
This is what Peter Molyneux doesn't get, but what defines Shigeru Miyamoto as among the absolute finest game designers of our time. Molyneux has his big ideas and seems to consider them to be perfect upon conception, rather than potentially excellent upon development. Miyamoto, on the other hand, has an excellent understanding of how to iterate on broad concepts to create an excellent end user experience. Their objectives are different -- Molyneux seems to want to express himself to his consumers and for them to thank him for it; Miyamoto wants to create the best experience possible for the consumer base, with his own expressive qualities filling in the gaps left behind.