That's pretty clear to me; they are irreversible on times scales of interest to our survival. (realclimate.org)
Once you fix the «long» to be a particular time scale of interest (say 10 years or 100) you'll see non-periodic oscillations on greater time scales. (realclimate.org)
Has anyone ever demonstrated convincingly that global temperature is bound by the central limit theorem on time scales of interest? (judithcurry.com)