It is, but you should keep in mind there isn't just one "the data" and in this case, what data source you pick affects what answer you get. The people you're arguing with might be looking at something other than GISS. For instance, here's the WoodForTrees index:
http://www.woodfortrees.org/plot/wti
the WoodForTrees Index is an average of four popular temperature series. This index was created specifically in order to reduce the need for arguing about which series to use. I believe the current formula is:
WTI = mean(GISTEMP-0.35, HADCRUT3VGL-0.26, RSS-0.10, UAH)
Playing around with the data sources, it seems like 1998 was an all-time high within the period of the satellite record - not surpassed by "a few years even warmer in the 21st century" - according to three out of four of those. Specifically, UAH, HADCRU3, and RSS. It was surpassed according to GISS or HADCRU4.
So not only is it not inaccurate to say "Since 1998, the warmest year of the twentieth century", it also would not be inaccurate to say "Since 1998, the warmest year for which we have satellite records."
UAH and RSS measure lower troposphere, while GISS and HADCRUT measure surface temperatures. I'm usually more interested in the temperatures in the first 10 feet of atmosphere rather than the first 10 miles of atmosphere (of course, I don't make climate models). HADCRUT4 is a different model of the same source data that HADCRUT3 uses and in theory should be more accurate.
In any case, you've got the hang of it now. By selecting a convenient data source, one can make many creative and not inaccurate statements.
http://www.woodfortrees.org/plot/wti/last:120/plot/wti/last:...
That one works even if you use GISS - the trend for the last ten years is cooling. (The trend over several other recent periods is "warming, but not very much".)
Note: the way I plotted "the last ten years" was to plot the last 120 monthly samples. The WRONG way to do that is to specify a "From" year and plot to the current day, because the resulting chart wouldn't start and end on the same month.
http://www.wmich.edu/corekids/images/icecoredatahistorical02...
However the 21st century has been an outlier in that temperatures have suddenly (for some reason :) failed to correlate with the amount of CO2 in the atmosphere.
Clever. "the warmest year of the twentieth century" is completely true. But it seems a little misleading, since there's been a few years even warmer in the 21st century.
Whenever I see the year 1998 mentioned in a discussion about climate, I am immediately suspicious. The temperature recorded in 1998 was a serious outlier.
Let's pretend the datapoint from 1998 did not exist. Then we could accurately say, "The last decade has been hotter than any previous year on record!". But with 1998, you could accurately say, "Most of the last decade has been cooler than it was in 1998!"
It's always nice to look directly at the data. http://data.giss.nasa.gov/gistemp/graphs_v3/Fig.A.gif
It's true, if you take the past 10 years only, the trend is pretty flat. But if you look at a 10 year running average we're in the middle of an upslope. We can even make it look like we're in a cooling period if we pick 1998 as the first year! Fun with data.