The idea was better stuff, quicker and cheaper. It turned out - like some of the lessons Boeing learned with the 787 - that agile development may work great at Facebook but it's a train wreck when applied to aerospace, military systems and gigantic procurements. Oops.
One of the basic ideas of agile dev is to get a partially-working system in the field ASAP. Doing so lets you figure out much sooner (10% into a project) that the design requirements are wrong, and that you need to rethink what you're doing. In this case, the loop wasn't closed - there were plenty of early signs that it was going wrong, but the project just kept going forward without reevaluating the basic requirements (VTOL being the most obvious case).
I don't think that means agile dev won't work for aerospace generally. It's more an indication that the organizations involved (governments and military contractors) are too heavy to handle an agile process: imagine trying to go back to congress once a month to get the requirements updated based on dev feedback. Smaller, independent companies like SpaceX can manage a much faster, cheaper cycle on the space side, and I think it's possible for a new military aero supplier to do the same.
There were also plenty of f***ups in assumptions the program made that were only really recognizable in hindsight, like the fact that trying to mesh the Marines' requirement for a V/STOL aircraft with the traditional designs for the Air Force and Navy hobbled the plane's performance for all three constituencies.
That wasn't only seen in hindsight. It's obvious that adding complicated, heavy components to something that's supposed to be fast and reliable is going to create problems. It was more of a "let's see how well we can apply modern materials and design to make this work" kind of thing... Initial tests showed it was possible, but a bit further into the program it was clear that it was still too much of a tradeoff to be worthwhile.