The Hockey Schtick points to a study on Pennsylvania temperatures that illustrates a point I have been making for a while:
A new SPPI paper examines the raw and adjusted historical temperature records for Pennsylvania and finds the mean temperature trend from 1895 to 2009 to be minus .08°C/century, but after unexplained adjustments the official trend becomes positive .7°C/century. The difference between the raw and adjusted data exceeds the .6°C/century in global warming claimed for the 20th century.
I think people are too quick to jump onto the conspiracy bandwagon and paint these adjustments as scientists forcing the outcome they want. In fact, as I have written before, some of these adjustments (such as adjustments for changes in time of observation) are essential. Some, such as how the urbanization adjustments are done (or not done) are deeply flawed. But the essential point is that the signal to noise ratio here is really really low. The signal we are trying to measure (0.6C or so of warming) is smaller than the noise, even ignoring measurement and other errors.