We've briefly seen the paper, it's good to see time taken to do outsider's analysis.

regarding fingerprint, problem is the data used:

• the doc takes all harvested, but balance/importance of an account could change over time... (i.e. take a look at funny outliers presented in a table above, that have high hBlks/r_imp_H, balance is very low, what does it tell you? the doc even states Balance:  r_hBlks / r_imp_H  has a weak correlation 0.344 should it really be surprising if you take ALL harvested blocks, but take only LATEST importance?)
• fingerprint graph on x has balance, on y has hBlks/r_imp_H,
  • mind that there's correlation between balance and importance...
  • as you can see pattern emerges for low amounts (10k - 100k), so emerging pattern shouldn't really be surprising,
    with low amount you probably generate only few blocks (10-100) so what you have on this image is / importance x balance,
    i.e. compare two accounts with same r_imp_H, where one harvested 9 blocks and other 10 blocks: (9 / 1601000) / (0.00003) = 0.1873 vs (10 / 1601000) / (0.00003) = 0.2082, thats' where the vertical gaps come from,
    horizontal linearity should be obvious as well, bigger balance == bit bigger importance, i.e (10 / 1601000) / (0.000031) = 0.2014 vs (10 / 1601000) / (0.000030) 0.2082