The data doesn't appear to be new, just selected and processed differently. The common thread on comments on Curry's blog, the Register Article (earlier post) and
here at DotEarth seems to point out the low frequency of historic data which could mean shorter events and changes (under 500 years) are not shown. The questioning is around the validity therefore of effectively gaffer taping under 250 years of high frequency data to the end which show changes which may have also happened earlier but aren't being shown.
Wonder how many times those graphs will appear in the next IPCC report ?