“You shouldn’t find yourself distinguishing the [timelessly] winning choice [(as calculated from expected utility over infinite attempts)] from the reasonable choice.”
In your example, your friend picked the choice that won once. It was luck, and he’s happy, and all is well for him. However, the expected value of box B was $100, which does not win over $1000. Arguably, the gambling in itself may have nonzero utility value, and the certainty of obtaining $1000 may also have nonzero utility value, but that seems irrelevant in your example from the way it was formulated.
TL;DR: It seems like you’re disagreeing more on the formulation or wording than the actual principle.
I believe it should be read as:
In your example, your friend picked the choice that won once. It was luck, and he’s happy, and all is well for him. However, the expected value of box B was $100, which does not win over $1000. Arguably, the gambling in itself may have nonzero utility value, and the certainty of obtaining $1000 may also have nonzero utility value, but that seems irrelevant in your example from the way it was formulated.
TL;DR: It seems like you’re disagreeing more on the formulation or wording than the actual principle.