Upvoted for content. Though it strikes me as slightly redundant (gut feeling), I think that it covers some important points. Organizationally, I would change your thesis so that it better reflected what the rest of the post was going to be about -- the post struck me as rambling a little too much.
Thoughts:
Since I cannot prove my value system on an absolute scale, I sometimes restrict use of the word 'should' to technical applications where there's no ambiguity about the objective. For moral arguments, I like to use 'I think that you would want to' as a replacement for 'You should' -- it seems more intellectually honest, less opaque, and less provocative.
I interpret 'should have known better' as a data-sparse, unnecessarily antagonistic comment -- it doesn't tell the listener much about how they should revise their updating system and risks causing negative emotions toward the speaker. I also think that it causes future action to be more pain motivated because 'avoiding harsh correction' may become one of the listener's motivations for improving.
It should be obvious that people need to learn what they know, but a tremendous amount of effort goes into blaming them for not knowing whatever would have been useful if only they'd known it.
A lot of this appears in the form of "should have known better". Sometimes (especially after a low-probability disaster), it will be people tormenting themselves about ordinary behavior. If only they had left a little earlier or later, they wouldn't have been in the car crash. As far as I can tell, people are apt to expect themselves to be clairvoyant, and they expect other people to be telepathic.
The latter comes into play when they're shocked and infuriated that other people don't have the same views they do, or don't understand what is vividly clear to the telepathy expecter. This plays out both politically and personally.
One thing I've learned from arguing about torture is that moral intuitions don't transfer very well. To some people, it's completely obvious that torture is a bad thing and doesn't work, and to others, it's completely obvious that getting important information for one's own side is an emergency, people can't lie if they're suffering enough, and people on the other side don't deserve careful treatment. (The overlap between what people believe is moral and what they believe is effective is probably worth another essay.) When you're in the grasp of a moral intuition, it can be very hard to believe that people who don't share it aren't trying to hide bad motivations.
On the personal level, it's probably more complex. You do want to be around people who are reasonably clueful about how you want to be treated, but on the other hand, I've seen and done a certain amount of ranting about how people should just know how to behave. How are they to know it? They just should.
I suggest that as a matter of epistemic and emotional hygiene, downplaying "should have known better". As far as I can tell, if taken literally, it invokes an unavailable counterfactual and just leads to fury. I think I have better sense when I don't deploy it, but it's hard work to inhibit the reaction, and it's tempting to think that other people should know better than to invoke "should have known better".
It's possible that "should have known better" is one of those normal people things that geeks need to figure out. I can interpret it as "I am lowering your status because I want you to never make that mistake again". I think it can be effective for specific cases, but at a cost of cranking up anxiety and despair because it isn't actually possible to know what one could be blamed for in the future.
This post is a rough draft. It could probably use more examples, and I'm not sure whether the material has already been covered. I think the idea of knowledge needing a source has been, but the emotional and cultural effects of not having a gut level understanding of the fact haven't been.
A philosophy professor recently told me that one of the few things philosophers agree on is that there can't be a moral obligation to do the impossible-- ought implies can. On the other hand, there hasn't been significant work done on figuring out what actually is possible.
On the epistemic side, I've been distracted by whether there's explanation required for how people recognize sound arguments, or if it's enough to say that they just do it some of the time.