Sussman goes over some interesting ideas on the provenance of calculations and asserts that "exact" computation is possibly not worth the cost.