If the model of dice are perfectly fair and unbreakable is correct, then the Swedish king is justified in assigning very low probability to losing after rolling two sixes; but this model turns out to be incorrect in this case, and his confidence in winning should have been lower.

Of course it would be silly to apply this reasoning to dice in real life, but there are cases (like those discussed in the linked article) where the lesson applies.