Let's say I have two friends, call them Bab and Jesiah. One day I contact Bab, and I say, "That guy Jesiah, he's taunting us on Twitter all the time. Here's what I suggest: I have hold of this new virus that can wipe things off the Internet. Remember [this is all just made up for this example, mind you] that terrible public bet you made a few years back? I can wipe out all evidence of that... if you take out Jesiah when he visits you next month."
Now, in the kind of simplistic view found here, "without a doubt this [offer] is making its participants better off!" After all, I wouldn't have made the offer if I didn't prefer to make it. And Bab can turn the offer down, and be in the same situation he was in before I made it, or take the offer... in which case he must think it makes him better off.
But the fact is that my offer is evil, and makes me worse off just by having made it, whether or not I "prefer" to do so. And while Bab can turn the offer down, by my putting temptation before him, he may be worse off merely having heard it: even to begin to seriously consider the offer is itself a moral error, and so he might have been better off had I never tempted him. That is why the Lord's Prayer contains the line "and lead us not into temptation," and not, "only present us with situations that in my model offer Pareto improvements."
And it is confusing a model with reality that is at the root of Hall's error in the post I link to above: there is nothing wrong with creating a model in which some new offer makes the person receiving it "without a doubt" better off. But it is a terrible mistake to equate that model with reality.
In the real world, sudden, unearned windfalls often leave the recipient worse off, and terrible events that a person never would have chosen to undergo may leave them better off. Reality, unlike our models of it, is complicated.