I've pondered this question as well. What I've concluded is that, instead of assigning "blame", "fault", or "responsibility", it's better to simply take a more consequentialist view, and ask: what are the likely outcomes of this person's actions? Should I convince them to do otherwise? Would it lead to an overall better outcome if something were done to stop them from doing it (again)? What should that something be?

By focusing on these questions, we can sidestep the question of who to hold accountable and instead look at what would be the best thing to do overall.

However, what's interesting is that answering that first question, "what are the outcomes", can be very complicated given the chaotic nature of human behaviour ("chaotic" here means "deterministic, but unpredictable in practice"). We have to use rule-of-thumb approximations for this sort of thing, instead of precise calculations. And it turns out that concepts like "accountability", "blame", "fault", and "personal responsibility" are very useful rules of thumb; in effect, when you blame someone for something, you are asserting that their behaviour requires some internal changes, or they'll just do it again. Even if the underlying causes are far outside that person's control, the logic works out the same.

To put it in maybe a more whimsical/poetic way: if we are but the fingers of the hands of Fate, then we cannot be judged for our sins, for they belong to Fate just as we do. But, since Fate doesn't have a mailing address, we'll have to settle for cutting off its fingers as necessary.