Linux

This is incorrect. It's true that most (in fact, I would say almost all) forks go nowhere but that doesn't mean forking isn't incredibly valuable. Even the example you cite, "original project is dead" isn't just incidentally useful, it's critical to open source. Other examples include:

• project's core team is part of a for profit org that is moving the project in a bad, profit motivated direction:
• project's leader suddenly and dramatically loses respect (maybe he killed his wife or something);
• project's leader dies without leaving a digital will regarding who controls the core repo;
• project continues to direct effort into features while falling to address major security concerns;
• project is healthy and useful in every way but there is an important use case not being addressed, and the fork would address it.

Even if 99% of forks fail, that's irrelevant because 99% of original projects fail in the same ways. Forks are critical to open source.