When melt ponds are not a problem area still has problems with cloud cover, and with not distinguishing between tightly packed floes with significant but small amount of water in between and continuous ice sheet. While I would agree that area tends to be a better reflection of how much ice there is under such conditions, I do think that extent is more accurately measured. The instrument errors for extent are lower, even though area is measuring something more useful, and closer to what we actually want to know.
I don't want to go on about this much more, but unless I've misunderstood, extent and area - if from the dataset - are exact twins except for the 15% threshold. Same instrument, same algorithms to interpret brightness temperature data, same land mask, etc. Extent may be a more accurate
representation at certain times of the year, but that doesn't mean it's a more accurate
measurement. Both area and extent derive from the same measurement, but are different interpretations.
Even Gompertz was showing nearly ice free in 2017.
I take it you mean this graph:
As you can see, volume goes below 2 thousand km3 around 2017. I don't know if there's a similar definition of ice-free for volume as there is for extent/area (1 million km2) that is generally accepted. I haven't ever heard it being 2 thousand km3, simply because if this volume would be spread out as 1 metre thick ice, extent would 2 million km2. That's not ice-free.
Most discussions involving ice-free from a volume perspective is simply 0 km3, and here are the suggested dates for that using different extrapolations:
If we assume that 1 thousand km3 constitutes ice-free, Gompertz suggests 2023.
The original Wadhams projection of 2016+/-3 was made based on a linear extrapolation, but restricted to recent years.
I'm assuming you mean Maslowski, and no, that wasn't based on linear extrapolation, but on model results.
Let's continue focussing on the 2016 melting season.