Why can't URLs start from the top domains and dig deeper into other subdomains and then finally resources? Today we start from lower subdomains and make our way up to top level domains, all to then go further back down into directories and items. Could something be made to experiment this without refactoring infrastructure?
Because Tim Bernese-Lee did it this way and stated he wished he had done it the other way. It was somewhat arbitrary then, and makes less sense now than before.