Yeah, I know, it sounds funny that it has to be stated so explicitly, but many FReepers don't really get the concept (that concept, and the idea that if you can't take out any one data point and still get the same "trend," then your trend is likely just the result of cherry-picking the data (like how--as is pointed out in the story--this supposed "cooling trend" requires selection of a specific year as the start year, a specific dataset, a specific length of comparison, etc. Any variation of these and the "cooling trend" disappears, which is a good indication that it's not likely a true trend.
But believe it or not, that's over the head of some folks. In fact, if you'd plotted the same numbers as polling data, I guarantee we'd see several FReepers claiming that the one point is an obvious outlier, and that there's a definite rising trend if you took that out or if you look at any other dateas a basepoint.