That's the GISS 5-year average. It may look somewhat stuck, although it's important to remember that the now-measly-looking 2002 5-year average was at the time, the highest in the instrumental record. The 1998 5-year average was 0.44 C above the 1951-1980 mean.
The 1979-1986 period was similar to the 2003-2011 time.
The .06 C drop from 1982 to 1984 is interesting - it's the biggest in the instrumental record since the warming restarted in the 1970s. It's a safe bet (as in, I'm willing to bet) that we will never ever get back together with a 5-year average lower than 2002's.
And speaking of betting, while that upstart antipodean Brian Schmidt hogs all the attention, my own 2007 bets plod along to their 10-year through 20-year finish lines. My bet with David Evans wasn't as generous to my side, comparing five year averages to 0.1/decade and 0.15/decade increases (details here from my 2007 post). Last year, fortunately, I was getting creamed with slightly declining average, -0.06/decade. Unfortunately that's now changed to straight zero, which still means I'm getting creamed (temps have to warm substantially or I pay out). If you compare even noisier data of 2012 to 2011, that's a warming rate of 0.2/decade, and I win.
I think the real lesson is not to worry about year-on-year changes. I said in 2007 that my worst-case personal outcome was to lose one bet, void some others, and win most. If the 2017 average is 0.68 or less, then I lose two bets. An increase of 0.09 in six years or less has happened plenty of times in the last 40 years. The warming we've been sticking in the ocean is going to come out at some point, and I think China and India will start to wrestle with their particle emissions soon.
I no longer think I'm safe from losing more than one bet, but unfortunately I feel pretty good about the ultimate outcome.