It's difficult even with that. The mean temperature is calculated via inference (obviously you can't have a thermometer everywhere). And there's occasional adjustments to how the global temp. is inferred. The NOAA recently did one [0].
They have a nice comparison list of the ten warmest years [1].
[0] http://www.ncdc.noaa.gov/sotc/global/2015/5/supplemental/pag...
[1] http://www.ncdc.noaa.gov/sotc/global/2015/5/supplemental/pag...