Universal indie pattern. Two questions to ask before starting: (1) Is the new project actually different, or am I escaping a hard moment on the current one? (2) Will I be embarrassed in 6 months if neither is shipped? If both answers are uncomfortable, force-finish the current one, even badly. Shipping a small version of project N beats half-building N+1.
I want to start a new project before finishing [current project]. Interrogate me: is the new one actually different, or am I escaping a hard moment? Will I be embarrassed in 6 months if neither ships? Then help me force-finish a small version of the current project before I'm allowed to start the next.