Anomalies Relative to the Earliest Possible Baseline

A recent Facebook post from Stefan Rahmstorf showed temperature anomalies relative to the earliest period in a temperature record -- I think it was GISS, which starts in 1880. I can't find his link, but found it interesting and remembered it.

Here's how NOAA surface data looks in that scheme:

This makes it clear that we're at or above 1 deg C of global surface warming.

And 1880-1909 isn't really "pre-industrial," which is probably (?) before 1850. Maybe 1750. I think BEST goes back that far. I should probably look at that. Someday.

So why don't the various data groups use this for a baseline? I'm not sure. Perhaps because the data from back then have some relatively large uncertainties -- see HadCRUT 4.6, column 2 is the monthly anomaly, and columns 3 and 4 the upper and lower bounds (sigma or two sigma, I don't remember). These bounds get smaller as time trends to the present. But not exceedingly fast. I need to learn more here -- or maybe someone in the comments can correct me.