Elon, you are known for taking over projects within your companies when you feel they are not executing well enough (autopilot, manufacturing process, etc).

There are often several broken projects in a startup; how do you decide when to take over a project and fix it personally?