What's more, the FAQ selects quotes from the article; the FAQ offers no additional information. In particular, from the article and FAQ:
>Without filling data gaps, our Standard 5×5 reconstruction (Figure 1A) exhibits 0.6°C greater warming over the past ~60 yr B.P. (1890 to 1950 CE) than our equivalent infilled 5° × 5° area-weighted mean stack (Figure 1, C and D). However, considering the temporal resolution of our data set and the small number of records that cover this >interval (Figure 1G), this difference is probably not robust.
It's disingenuous to imply that this clarification was tacked on after the fact, as the blogger suggests.
Finally, there's this from the blog:
> If your methods can’t resolve data points in a given period of time, then DON’T REPRESENT DATA POINTS IN THAT GIVEN PERIOD OF TIME.
Marcott, et al clearly state (in the paper) that there is zero preservation of variability in 300 year resolutions, 50% at 1000 year resolutions, and nearly full preservation in 2000 year resolutions. This blogger takes this to mean that the last 2000 years should be arbitrarily trimmed from the graph. Following that logic, we might as well continue trimming along the x-axis until nothing is left.
In fact the authors have used the instrumental record of the last 130 years to justify the incorporation of the statistically uncertain data from the last two centuries. From this we can understand their scientific model. They know their conclusions first, and then prove their conclusions. Good science is developing hypotheses and testing them.
Yes, you're right, but now we're talking about science media.
The data is still useful in determining the statistical confidence of the set, but not reliable for representation as its representation completely changes the understanding conveyed by the data set. So the data stays in, and the X-axis doesn't get trimmed. Only the data point, as represented in the last, non-robust interval, gets trimmed.
This correlation of data acquisition appears to have been done until the conversion of recording stations to MMTS stations. The correlation stop, at that point, and we are now entirely dependent on thermometer data. Coincidentally this is is also the 'hockeystick blade' time frame.
