The graph has a link to show where it got the data from;
http://earthquake.usgs.gov/regional/world/historical.php If you follow this link, you'll see at the top:
"Selected earthquakes of
general historic interest"
If you stay on the same site (earthquake.usgs.gov) and look at the "boring" earthquakes as well, you get this graph;
earthquake.usgs.gov/earthquakes/eqarchives/year/graphs.phpScroll down and you'll that the average is pretty constant. Compare the two graphs, for example 1990, and your graph has 2 quakes happening. The site where the data comes from has 127.
Obviously 125 of these weren't of "General Historic Interest".
Check out the FAQ I originally posted, it sort of explains why more earthquakes are of interest than others, especially recently.
I think someone went to a lot of trouble to generate a graph that used the wrong data I'm afraid.