I thought the idea was good, the execution was poor.
As somebody who used to live off the nightly builds, I can say that the problem wasn't that the execution was poor - it was that the ideas changed too much, too frequently. Have a vision, document it, and then build it. The programmers appeared to be tasked with executing visions that changed dramatically on a monthly basis.
I would test whole featuresets only to find them disappear completely out of the next build. My (least) favorite was the RSS integration. It's integrated with the bookmarks, no wait it's a separate thing, no wait it's gone altogether. Huh? Development time was wasted, testing time was wasted, and users got flaked out.