Apparently, the graph designer isn't familiar with the y vs. x convention. But that's not really the point (math errors annoy me, while grammatical ones don't for some reason). It seems that a linear plot isn't really appropriate. This graph and the one mk links appear to be biphasic more than anything. That is, there are a lot of countries bunched into two distinct groups. One in which the Gini coefficient is about 35 and GDP is uncorrelated to it; the other in which GDP is ~10k/capita and Gini is uncorrelated to it. It would be interesting to see if there are any qualitative distinctions between these countries. I don't think a linear regression yields any interesting data here, beyond that supply side doesn't work (which is a no brainer for anyone, well, with a brain).