Frankly, the claims made in this article are a bit ridiculous. To anyone with any kind of math or science background, the claim that one can accurately measure the temperature of the earth down to two decimal places is hubris.
But it's even worse when one considers that the distribution of temperature recording stations isn't uniform, that large swaths of the earth have barely any coverage at all, and that the most accurate way of measuring temperature has only been available for the last couple of decades (satellite measurements). And I might add that the article above uses surface temperatures and not satellite measurements.
In many of these temperature stations, temperatures are still recorded by hand, on a piece of paper, by observing a thermometer. One has to deal with different measurements made by different people and equipment as things are replaced or repaired.
Additionally, lots of these temperature stations have had their recorded temperatures adjusted upward. I always wondered why, and I did some investigation and found out that it was because the temperature stations had been moved from an urban environment to a more rural environment. However, the measurements have not been subsequently adjusted down as those new environments urbanized. What we see is a ratchet effect, not in actual temperature but in observed temperature.
What's the upshot? The supposed 0.02 increase in temperature is well within the margin of error for both the surface level temperature measurements and the estimation made in the upward temperature adjustments at many sites.