Data clustering and the glassy structures of randomness