I concur with your conclusion that there is little practical motivation for a change to the status quo, since observational datasets have at most 3 truly equivalent dimensions (and more often just 2, our ability to measure distance being what it is).