The basic idea seems to be: In some cases it might prevent more harm to punish innocent people i.e. family members of criminals, therefore consequentialists should endorse punishing family members in those cases.
The author seems to think this logic doesn't apply to their own position because they believe that the reason to punish wrongdoers isn't to prevent further harm, but because the punishment of wrongdoers is justice which is good. The opposite side of this would be punishing innocent people, which would be injustice. This contrasts with pure consequentialism which doesn't care about justice vs. injustice, it only cares about the end result.
The key part is from the definition from the Stanford Encyclopedia of Philosophy, which says:
> 3. that it is morally impermissible intentionally to punish the innocent or to inflict disproportionately large punishments on wrongdoers.
But then, the author then goes on to say:
> What if harming Hitler’s baby was the only way to save the world? As a moderate deontologist, I reluctantly endorse this implication. But only in dire hypotheticals with little real-world relevance.
So it seems like the author also endorse harming the family members of wrongdoers, but for them it's ok because it's only in "dire hypotheticals". This feels like a bit of a cop out to me. It seems like you could just as easily call yourself a "moderate consequentialist" and say that punishing people is only good as a deterrent, but also punishing innocent people is wrong.