User blog comment:Chronolegends/Chronolegend's Shrinking Hierarchy/@comment-1605058-20170211175133/@comment-24920136-20170212163415

You are right, i've also encountered some other minor inconsistencies, such as f_a(b) > f_c(d) for some cases where a>c and b>d. Though it terminates correctly for all valid inputs, some tuning is in order.