Hi Phlinn
HADCRUT3 has even less coverage than HADCRUT4. Why not use the latest and greatest? Regarding mathematical artifacts, replication over many different reconstructions using different methods and different data gives us confidence in the results. Regarding malfeasance, I'm not sure that the resignation of a journal's editor when it becomes clear that the journal is pushing an agenda at the expense of the truth is malfeasance. I'm not inclined to discuss conspiracy theories although I know these narratives are popular. Suffice it to say that I disagree.
Regarding plotting from 2002, yes the trend line is negative for some data sets. It is more negative if you plot from 2010. What does that tell us? Note that the data is consistently above the trend until about 2007. Note that the data cycles above and below the trend as PDO and ENSO wax and wane. What we are seeing is a steady upward trend with natural variability superimposed on top. We're below the trend line now and the indicators show that we should be. That means we will go back above when the indicators flip back to the positive part of their cycle.
When you look at the data, do you have any expectation that the next El Nino will not be the new hottest year on record? That's even with the PDO strongly negative. You can subtract ENSO and PDO from the trend with this tool and you end up with something closer to the real trend: http://scratch.mit.edu/project...
P.S. if you have kids then you should introduce them to Scratch. I've been showing it to my kids and I've become addicted :) Please pardon the Scratch evangelism :)