It's easy to forget that, for the most recent two previous years, 2016 and 2017, the ice on this date was in bad shape for close to the North Pole and on the Russian side.

Thanks for this reminder. 2016 was a total mess near the pole, and in terms of a hypothetic index of latitude-weighted ice area or difficulty-weighted ice area, IMHO was lower than even 2012.

OTOH, this also made it vulnerable to early refreeze.

Tech note: the melting-difficulty weighted ice area index would calculate the average concentration of a given grid cell on September 5th-20th, and calculate weights for each cell based on its resistance to concentration loss. The higher the September concentration, the higher the weight. The index then sums current ice area on a given date, weighted by these weights. So a year that loses ice early in the CAB but delays Hudson and Baffin will show these losses much more clearly with this index.

Around the September minimum, the index will show how far the season was from a BOE, in terms of required additional energy/melting punch.