EDIT
Needless to say, Mark spoke to a huge ballroom, and it was packed to the gills. This really is the Year of the Ice. He opened by showing some quotes from his earlier papers. Seven years ago, big changes in the arctic were clearly observable, but he wouldn't have said with confidence that they were due to anthropogenic global warming (AGW). The problem is that the Arctic is subject to a lot of natural variability, notably the North Atlantic Oscillation (NAO). There was NAO-related warming in the period 1920-1940, so how could we know that the current warming wasn't part of the same thing? What changed Mark's mind was that improvements in climate models made it easier to see the anthropogenic component of climate, as did a few more years of the AGW signal rising above the background variability. This is true not just for global means, but for regional climate. More importantly, there as been a persistent warming and sea ice retreat despite circulation changes (e.g. NAO going neutral) which should be forcing ice the other way. Mark said that the circulations that used to help you don't help you build sea ice anymore, and the ones that hurt are now hurting more. That latter effect is intimately connected with a general thinning of sea ice,which makes it more vulnerable.
Models are getting good, he pointed out, but reality is exceeding expectations. Since 2005 there has been a 25% decline in Arctic sea ice at the time of minimum, equal to the combined area of Texas and California. What's more, the drop in 2007 was way below the already steep trend line for 1978-2002. Factors contributing to the unusual 2007 drop include an unusual pattern of atmospheric circulation, with high pressure over the Central Arctic and low pressure over Siberia. This brings a lot of warm air into the Arctic. How does this fit into the longer term pattern? It comes back to the thinning again: the sea ice was so vulnerable to this situation (which has happened before) because it is so thin. He compared sea ice trends between observations and the IPCC Fourth Assessment Report models. The observed rate of ice loss is much faster than expected. Even before 2007, the downward trend is at least twice the GCM trend, and 2007 is of course way more extreme than even that. The GCM's say that 40% of the sea ice loss is anthropogenically forced between 1953 and 2007, and if you just count from 1979, it's more like 50%. But that's the GCM's. In reality, given the extreme observed loss, the anthropogenic component may be much greater. It's the ice thickness again: GCM's seem to be overestimating ice thickness.
Other big changes are happening. Igor Polyakov's results show pulses of increasingly warm Atlantic water, seen in moorings at Svinoy and Fram Strait. Also, there are feedbacks associates with Pacific summer water: Loss of sea ice means the atmosphere warms more, and that helps shunt more Pacific summer water farther into the Arctic Ocean, melting yet more sea ice, etc.
Are we at a tipping point, Mark wonders? He pointed to an NCAR simulation with the model CCSM3, which shows a gradual sea ice decline until 2020, but then a kick from an atmospheric fluctuation causes a sudden near-total disappearance of the thinned ice, which was primed to pop. The state of ice thickness observed in 2007 is quite similar to the modelled state in 2020. Mark suggests (my paraphrase) that 2007 is the new 2020. Is the 2007 drop the first harbinger of a shattering drop like the NCAR model has in 2020? We're close to the critical ice thickness now, based on IceSat data. Will 2007 be remembered as the tipping point for sea ice? Only time will tell. Based on a detailed comparison of observed patterns of Arctic warming with GCM patterns including anthropogenic forcing, Mark declared confidently that "Arctic Amplification is Here." The simulated patterns are spot-on the observed patterns, both with regard to seasonal cycle (warming is greatest in the cold seasons) and vertical structure (warming is most pronounced near the ground). But — the amplitudes observed today are most like the GCM forecasts for around 2020. Not something to send you off to bed with sweet dreams, exactly.
EDIT
http://www.realclimate.org/index.php/archives/2007/12/live-almost-from-agu%E2%80%93dispatch-2/