While I do agree with you that his theories break down after the first book, Facebook was delivering healthy profits pretty soon after IPO. They could have made profit far earlier, but their investors chose to roll that profit back into the business and make a big bet on mobile. That bet paid off and hugely increased the available equity value and allowed them to go on an acquisition spree and diversify into other growth categories.
Facebook is also a unique business in this context because it didn't really disrupt an established market. They were just first to reach scale in an emerging product category with strong network effects. But there was no product that consumers were over / under served by. You could say advertising, but their real innovation was on the consumer side: you don't get to show lots of ads until you have lots of traffic first, and Facebook's traffic and content acquisition strategy were novel at the time.
But I also don't think that Clayton was really thinking about startups back then. He was more thinking about businesses that solves actual problems, not Yo apps :)
Perhaps what is really the problem is our definition of a business/startup is lacking some extra definition hmmm...
I believe this is exactly where the author an many people get Ubers success wrong.
Uber big disruption was in challenging legislation not in providing a better service.
Uber started by allowing limo drivers to make extra money. They then moved on making it possible almost anyone to start as taxidrivers and thus flodded the market with available rides. This wouldn't have been possible if they had used the normal channels.
They also benefitted from the popularity of smartphones and GPS removing the need for a taxi central for people to call in an order rides.
The problem with existing taxis wasn't in the quality of the service but rather with accessibility. This is what they disrupted.
Where some of Claytons theories break down is in his second book. Innovators solution. "Be patient for growth not for profit" which isn't necessarily wrong but Facebook proves that wrong.
I once asked him about this over twitter he said great question but never answered it :)