Read the whole paper. Conformal coating does not always contain whiskers. The thicker it is the less likely penetration is (and the more maintenance/repair trouble it causes). It does reduce the likelihood of shorts because it does almost always prevent a single whisker from penetrating from the outside, and it helps keep detached whiskers from shorting adjacent conductors.
Shorts remain possible from two causes in conformal coated boards. Two whiskers can contact each other after penetrating the coatings from within. A detached whisker can contact two penetrated whiskers.
In the absence of a meaningful theory, or a repeatable series of of test data reflecting actual durations and environments no one has any idea how likely any of this is. Is conformal coating better than nothing. Absolutely. How much better?

Is a conformal coated lead free board as reliable as a non-coated PbSn board?