I would correlate the measurements of a particular site with how a particular headphone you are familiar with sounds. Then evaluate other cans by looking at the plot differences using the can you are familiar with as a baseline. I would avoid absolute sense comparisons across sites due to differences in measurement methodology.

ON SECOND THOUGHT #1: Rin Choi seems to strongly feel his plots are moar bettar than others in the absolute sense.

ON SECOND THOUGHT # 2: In the absolute sense, Rin Choi's plots don't seem to correlate well with how certain cans I'm familiar with sound to me.

ON SECOND THOUGHT #3: Rin Choi has a tendency to all over other peoples work if it doesn't agree with his methodology (even when AFAIK an absolute measurement target for headphones is still an area of research)

ON SECOND THOUGHT #4: I have lot's of second thoughts.

According to their site, Audeze uses a Neumann KU 100 Dummy head:

http://www.audeze.com/audeze-lcd2-waterfall-plots

They seem to use quite a bit of smoothing and emphasize their low frequency extension and performance. Not sure what compensation method they use.

There are many compensation methods, and I don't think there is an industry standard. There are perhaps a couple of popular compensation methods though depending on measurement approach, which are discussed and described to some extent here:

http://www.innerfidelity.com/content/headphone-measurement-proceedures-frequency-response