On a general evolution criterion in macroscopic physics