Should it be done by the ISP? By the IaaS? By the software provider? By developers?
Ideally, moderation would occur at just one level. That way, there are less avenues for abuse to occur. Leaving moderation rules vague may lead to some services that could be beneficial (whistleblowers, etc) to be easily taken down.
Spam that could overload a system would need to handled higher in the tech stack. But some content moderation could at the level of the individual user, who could filter what he wants to see based on his own preferences.
The fact that KF is so beyond the pale that it can't appear to find any stable business partners is more a condemnation of KF than the system.
Oh, no doubt that a massive online campaign surfaced the question of whether KF needed to be evaluated.
The fact that they've all come to the same conclusion is because of KF's content, not because individual service providers are not independent. The common thread is the site under evaluation, and the site is that bad.
But from all that's been written on the topic, there's been very few concrete examples put forward of dangerous content on KF. It seems to me more likely that businesses have simply seen that the twitter campaign reached a critical mass and made the simple decision to cut them off to reduce reputation risk without doing much of an investigation themselves.