Problems that remain persistently insoluble should always be suspected as questions asked in the wrong way, like the problem of cause and effect. Make a spurious division of one process into two, forget that you have done it, and then puzzle for centuries as to how the two get together.