jmh comments on Problems with predictive history classes