"But half of the end-of-summer summer sea ice found in the Arctic in 1950 is already gone." In terms of the albedo feedback the September sea-ice extent is really not very consequential, as it is then around the time of the Autumn Equinox, and the Arctic isn't receiving that much solar radiation compared to the peak in June at the time of the Summer solstice. That's why this melting season is so very interesting/scary. It has set large records for May, and could do the same for June - peak months for solar insolation at the Arctic. Thus far the rate of decline for these months has been much slower than for later in the melt season. If that is now changing with this year then we will really see the albedo feedback take effect. If I had time, what I would want to do would be to calculate the theoretical maximum albedo feedback each melting season by integrating the sea-ice concentration observations with solar insolation (under a clear-sky assumption). I suspect that this would show that, for all of the dramatic decreases in September sea-ice extent, we ain't seen nothing yet. When we do that will really heat the Arctic up - with obvious implications for the Greenland Ice Sheet.