Impact Factor and the power-low tail of citations

I pulled the data of A simple proposal for the publication of journal citation distributions and produced probability density function using geometric bins. The result is:

They have power-law tails, so I belive the suggestion of Time to remodel the journal impact factor is a good quick fix.
If we replace mean with median in the definition of impact factor, I think it has a better interpretation. If you publilsh an article in Nature, there is 50% chance that it get more than 24 citation in two years time.

For more details about math and methods, please see the source file.