UML really died because modern software development is less like designing an aircraft than it is a research program to design a new kind of aircraft. To the pr

UML really died because modern software development is less like designing an aircraft than it is a research program to design a new kind of aircraft. To the programmer, the plafroms they are going to build on are so vast and complicated, that their behaviour is not fully known to the programmer. So to design a specific application, the programmer has to performa series of experiments on the platform to determine the manner in which it will have to be used to support the objectives of the application. When those experiments are complete, the programmer can then spend some time in design, but for the most part programming-in-the small is akin to conducting research and then doing a little desing. For large systems, someone has to do design, but the best way of doing such design is through whitepapers, and large scale experiments and possibly simulations and analysis. UML isn't particularly useful, beyond establishing a common vocabulary for diagrams -- a task that can be done more simply and informally.