This strikes me as an excellent idea — there are a lot of things in climate that will remain really hard to figure out, but a scientifically and statistically sound approach to creating a surface temperature record should not be among them. It is great to see folks moving beyond pointing out the oft-repeated flaws in current surface records (e.g. from NOAA, GISS, and the Hadley Center) and deciding to apply our knowledge of those flaws to creating a better record. Bravo.
Warming in the historic record is not going away. It may be different by a few tenths, but I am not sure its going to change arguments one way or another. Even the (what skeptics consider) exaggerated current global temperature metrics fall far short of the historic warming that would be consistent with current catastrophic high-CO2-sensitivity models. So a few tenths higher or lower will not change this – heroic assumptions of tipping points and cooling aerosols will still be needed either way to reconcile aggressive warming forecasts with history.
What can be changed, however, is the stupid amount of time we spend arguing about a topic that should be fixable. It is great to see a group trying to honestly create such a fix so we can move on to more compelling topics. Some of the problems, though, are hard to fix — for example, there simply has been a huge decrease in the last 20 years of stations without urban biases, and it will be interesting to see how the team works around this.