I don't see how this helps to understand why the NSIDC promotes extent more than it deserves. Let's collect points why extent might be superior or not...
1) Extent is what matters to ships. They need to know which areas are ice-free, which areas are impassable, and which areas are loose pack "proceed with caution"
2) Extent is what matters to the public, visually. It's a simple "you can see it / you can't see it" metric.
3) There is good historical data for extent which (with care) can be integrated across multiple sources from passive microwave satellite data to radar data to ships' logs. There is much less data for area, and it's hopelessly confounded by melt ponding. There is next to no historical data for volume, comparatively.
4) Extent is what he was asked to comment on. He was addressing the question of when the Arctic will melt out. That is intrinsically a question of extent, i.e. whether the Arctic ocean is covered with ice or not! It seems particularly perverse therefore to slate him for saying (paraphrased) "According to our best guess,
this much of the Arctic will still be ice covered up until 2030".
All the above are fairly simple, obvious answers which make sense to me. However, above and beyond that, you're missing a key point about that article. The claims Walt makes for the date of Arctic melt-out are not based on extent: the claims are based on the GCMs. The predicted extents are the
output, not the input. The GCMs model as much as they can, in as much detail as they can, and that includes volume - in fact, the better ones model the water, ice and atmosphere together. If you look at the papers, you'll see that the GCMs also model the ice as becoming thinner, i.e. the volume drop precedes the extent drop - but that wasn't the question he was being asked to comment on.
Finally, his article (if you bother to read it in detail) makes the important point that while the
model ensemble predicts ice hanging on for a couple of decades yet,
individual model runs frequently show rapid ice loss events over the course of a few years. Any individual model run might well have a period of 5-6 years which appears at least as dramatic as 2006-2012, but then slow back down again. Since the real world is the equivalent of a single model run, scientifically we simply cannot say whether the trend is wrong, or whether we've just had an unusually sharp "downward wiggle" in a trend that will ultimately zero out in the 2030s. The best guess based on the GCMs is the latter.
It would be very interesting to look at some of the data for individual model runs, rather than the complete ensemble, and have Tamino, Wipneus and all the rest run Gompertz fits (and linear, and exponential...) for individual runs. Possibly the most important question would be the long-term autocorrelation of the runs: whether individual runs that show a sharp early dip are also the runs that zero out soonest, or whether they return reasonably rapidly to the average trend. It might be (for example) that the world was indeed on a trend set to zero out in 2030, but that the recent spell of several bad melt years in a row has tipped us into a more rapid regime of melt-out. Tietsche et al looked at this after imposing artificial perturbations (removing all sea ice) and found that it took about 3 years to "decay" back to the long-term trend: it would be interesting to know if the same applied to internal fluctuations within the model itself.
...and bring back the necessary objectiveness the W.M. article lacks.
This is what I mean by a lack of respect.
I opened this thread to find out what might qualify extent to tell the future and declare all other parameters excrescent.
And you would have a point if Dr Meier was doing a naive extrapolation of the extent figures with no underlying model for how climate works. He's not doing that - however that
is what a lot of people here are doing with the volume figures.