To be compared to 2008/508 that the authors are not fair and respectful enough to take into consideration.
So, instead of understanding, developping and applying Shannon's high-order attacks, the authors still wonder how to "combine signals" (instead of combining probabilities), still rely on 100% ad hoc attacks such as MIA, etc.
One problem is that, by the Theorem of Total Probability/Plausibility, the marginal posterior probability mass function for the subkeys writes as a sum over all possible values of all masks of all encryptions.
Therefore the authors can probably feel or guess that the gap between their attacks and Shannon's increases with the cardinal of the mask space, the order of the attack and the number of side-channel signals. And the purpose of masking is precisely to make this number as high as possible!
So, Would the authors simply stop violating Probability Theory for a while and compare Shannon's attacks to this sad, undecent folklore, they would certainly make sure very soon that their current analysis is absolutely irrelevant.
How long should we wait? That is the question.
Errare humanum est, perseverare diabolicum.
F. Pautot