A large part of the work of scientists is determining how accurately we know things, in this case how accurately we know the global mean temperature. That is also where Harris got the estimate of the confidence interval ("uncertainty") from. Unfortunately, the press is less interested in this.
The mentioned uncertainty in the global mean temperature is exactly because we do not have measurements everywhere and have to estimate the mean based on a sample. This is normal statistics.
All groups computing the global mean temperature (NOAA, GISS, Berkeley Earth, Japan and the UK) have difference methods to estimate the global mean temperature and find basically the same answer for the long term trend. This sampling uncertainty is important for a monthly or yearly average, but it is not important for the long-term trend, which most people are interested in.