“We had two options: a conservative strategy, which would immediately please all users, leaving the code basically unchanged, and our more aggressive feature development and code renovation path, which has created some stability problems in the short term but is rapidly leading to a completely new and substantially improved free office suite: LibreOffice 3.5, the best free office suite ever."
Paraphrase: we had two options. Either do it the proper way, through unit testing and refactoring, which takes time but is how software engineering works. Or make some big hacks and release it without proper testing so we could claim progress.
Sounds like every other corporate IT project to me.