People constantly get polar amplification wrong. The most obvious mistake is to assume it applies equally at both poles. It doesn't in both models and observations: there is far more warming in the Arctic (at present, and expected in the future. This picture is somewhat complicated if you look back to, say, the last glacial, where climate change was amplified at both poles; but thats because it was for long timescales, hence the stabalising effect of the Southern Ocean doesn't apply (e.g. this RC post)).
But they also get the mechanisms wrong, too.
Hence, as OM points out in Natures blog, the apparently bizarre trumpeting (and not only by the septics) of a new result showing that warming in the mid-trop is contributing to Arctic warming as supposedly proof that its not anthropogenic. As OM puts it there is an idea that the ice-albedo effect and anthopogenic warming are effectively synonymous, and that if warming isn't due to the feedback but to increased atmospheric transfer of heat then it is more "natural" and less "man-made". But in fact increased mid-trop forcing is perfectly consistent with GW (e.g. Alexeev et al) and is arguably a signal of it. Which is one reason we were quite pleased to find enhanced trop warming around Antarctica... it suggested links to the wider world rather than just local effects.
I never really worked out exactly how much Arctic amplification was supposed to come from this or from that, and I'm not sure if anyone else has. But the (sea)ice-albedo feedback (I think the seasonal snow cover on land may be more important) is such a nice easy idea that everyone loves it.
[Update: thanks to H for noticing this over at RC
I read the Nature paper. It is an intersesting paper and I agree with you that if the authors had analyzed the CMIP3 output, it would be much better. I also don't think it is right to say that in the CGCM models the ice (snow) albedo feedback is the main mechanism for polar amplification. We  have recently published a paper and theoretically proved the feedback from heat transport greatly influence the vertical and meridional structure of global warming. It is important to point out the heat transport feedback does not have to warm the atmosphere only, it also can cause a larger surface warming by the concurrent other thermal-dynamic feedbacks.
 Ming Cai & Jianhua Lu (2007),
Dynamical greenhouse-plus feedback and polar warming amplification. Part II: meridional and vertical asymmetries of the global warming
Climate Dynamics Vol 29:375-391
doi 10.1007/s00382-007-0238-9 -W]
I noted the Graversen, et.al, paper was submitted to Nature on March 27, 2007. It would be safe to assume the research and writing required several months or more. It could not have included any of the 2007 melt (post-submittal) and likely not much of the 2006 max melt either. So, an update on the research is in order.
The Arctic sea ice melt records of the past decade confirm the fact that new ice in the western Arctic ocean predominates.
New ice melts first and fast.
No problem with natural causes melting the sea ice.
Dr. Chapman's Cryosphere Today provides evidence that Arctic sea ice melts each year - both from the east and west - along the ice margins. As that melt area increases by millions of square miles, the extent of new ice increases, in direct proportion with the expansion of the melt area. That creates lower albedo over a greater surface and more heat uptake in the ocean surface.
As I review the daily images on Cryosphere Today, there have been events of massive sweeps of ice melt on the Eastern margin stretching almost to the North Pole. I am no scientist but I attribute that to 'natural causes' since there can be no other explanation, given the very short period of time - days -. Regardless, ice melt means new ice next summer, in that eastern region.
New ice melts firstest and fastest.
So, now that we have been offered new evidence that natural causes contribute to Arctic sea ice melt, how about giving some time and attention to studying what impact Arctic sea ice melt means to the world's grain basket in Western North America.
John L. McCormick
i thought a part of polar amplification also resulted from the relatively more important additionnal greenhouse effect at high latitudes, since there's less water vapour there ?
[I know of no evidence for that, though its a popular idea. I did address that on the old-blog post I referred to -W]
This one and its predecessor, perhaps? (I can only see the abstract and this RC post that Gavin thanked the author for making): http://www.realclimate.org/index.php/archives/2008/01/new-rule-for-high…
The direct CO2 forcing at the top of the atmsophere (TOA) is largest in subtropics as shown in IPCC (2001), but if we notice the vertical structure of the forcing, then the CO2 forcing at the land(ocean) surface is larger in high latitudes (about 60N, or 60S), we recently developed a new method [1,2] to decomopose the contributions from vairous processes (dynamic or thermodynamical) to the climate sensitivity by considering the vertical structure of forcing and feedbacks. Also we have recently finished a somehow idealized CGCM simulation and use the method to analysis the ploar amplification even without including the ice-albedo feedback in the model. This, of course, does not mean that we ignore the role of ice-albedo feedback in actual climate change. The manusript will be submitted soon.
 Jianhua Lu, Ming Cai
A new framework for isolating individual feedback processes in coupled general circulation climate models. Part I: formulation
 Ming Cai, Jianhua Lu
A new framework for isolating individual feedback processes in coupled general circulation climate models. Part II: Method demonstrations and comparisons