But isn't that what the Hubbert peak has always been? In fact, the Hubbert model is called elsewhere in the literature the "Logistic Model," satisfying an equation equivalent (after changes of scale) to the ordinary differential equation dq/dt =q((1-q). It initially grows exponentially (oil fields don't actually behave this way, so Hubbert taught us to discard the "early-early" production), but it "saturates" and it gets harder and harder to produce as the total supply is exhausted (far past peak).

The peak in lower US without Alaska production about 35 years ago arose from an inability to get new wells to replace fully the depleting old wells. The biggest, easiest fields are pumped out first; as they deplete, the industry drills more, less productive wells,...

Alaska overwhelmed that depletion when it came on, but now it's depleteing rapidly, too. BTW, it seems to me that the early (pre-peak) tail of the Alaska production shifted the 1971 peak in total US production a little later than Hubbert's original prediction, as well as introducing a smaller, secondary peak on the down-slope. I think Deffeyes' more recent books show a slightly different version than Hubberts original paper, but I don't have the numbers to check this hypothesis.

When near the all-time peak in production (let's use the word "peak" ONLY for the maximum of some identified mathematical function unless we're talking about mountain ranges), the capacity to bring new production on line is limited, so spot shortages and price volatility reign. And we won't be sure this was the all-time maximum production until years later, when we're well below it and can no longer hope to return to it.