An interesting article on InfoWorld called "How to salvage a (nearly) hopeless software project " discuss how to rescue a software project that is not delivering the expected. When is it better to create a complete new software than fixing the old one? How shall old, no longer supported, hardware or software dependencies be handled? The article gives a few pointers, but much of the essence seems to be deal with it and do the best of the situation...
A blog about software development and architecture. The main focus is .NET and Open Source.