This is why the existence of the medieval warm period has been such a battle ground between sceptics and AGW hypothesis proponents. Good evidence from worldwide sources such as the Greenland ice cores are ignored or marginalised in the literature.
Here is a file containing the individual country records from 1900 before they are processed by CRU's HADcruT2 'code', but after homogenisation by GHCN.
As you can see, there are many countries which show little warming over the C20th according to this data. The Russian data in the very bottom graphs only represents 40% of the landmass of Russia, which itself is over 10% of the global landmass. Stations north of 70 degrees which show no warming have been ignored.
To summarize so far. Sceptics contend that:
1) Proxy evidence for previous natural warmings and coolings has been minimised or ignored.
2) Instrumental evidence for warming has been exaggerated by selectivity and poor station keeping. We can expect a paper on this from the Watt's team in the new year.
3) The co2 record is dominated by the Mauna Loa data, which is kept by Charles Keeling's son. Ignored data shows much greater variation in co2 levels over the last 150 years.
4) The natural recovery from the little ice age, 60 year oceanic cycles, and multidecadal variation in cloud cover have not been properly factored in to calculations of climate sensitivity, which has led to an exaggerated figure for the effect of co2.
5) The solar variability has had it's effect on climate variation underestimated due to it's supposed non correlation with temperature since 1980. If this is understood as largely an effect of (2), then the sun comes back into play as a more dominant climate influence. This will further diminish co2's role in (4).
So finally to your question about disproving the A in GW.
6) Sceptics are saying that the foregoing issues mean that whatever role co2 does play in global temperature, it is much smaller than the claims made by proponents of the AGW hypothesis.
7) We are discussing climate analysis, rather than some kind of lab operable climate science, which cannot deliver definite prognoses of real world climate, due to the number of confounding factors in the data, and a low level of scientific understanding of potentially large factors.