I'd like to think you're wrong about the "human nature" part. I know people who seem to put the best construction on almost everything, and I attempt to do so most of the time myself; in any case, I don't automatically assume the worst in what people say. On the other hand, I know that there are a lot of people out there who do assume the worst--although that is in part a function of whether they already like you or not. There are plenty of examples in which people think that someone likes them or really hopes that that person likes them, and so they read everything as more positive than it really is. I generally don't assume that someone is doing something intentionally malicious, and in experience that has been the case even when they are.
The second part, though, is frequently noted regarding online communications. In direct person-to-person communication the majority of what we communicate is non-verbal--facial expression, posture, tone quality, inflection. Some of that is lost even when we are talking over voice-only communications, telephones and radio. Even more is lost when we communicate entirely by text. Because we lose that, we attempt to derive it from the text itself, and we often misread.
That happens in life, as well. I've often noticed that my wife's sarcasm is so natural that sometimes even I don't know whether she is serious or sarcastic. She can say something like, "Well, of course, you would have to do it that way," and she means, "How can you be so stupid to think that would work" while the person to whom she is speaking and those in the vicinity would hear "I understand completely, and approve your decision." The sarcasm is not in her inflection, which is usually very deadpan at that point; but it enables her verbally to confirm her opinion without creating a fight--and, as I say, even I sometimes do not know whether she means what she says or the exact opposite of what she says.
With text, that's true of everyone, and everything everyone says. The inflection is gone; the body language is gone; the facial expression is gone. All we have is the words themselves and our knowledge of the person writing them. The less we have of the latter, the more we read into the former. Add to that the possibility that someone is disagreeing with us, or even merely that we think they might be, and it goes from disagreement to disagreeable, from statements of facts to character assassination, in short steps.
It's a standard rule in all civilized forums that everyone ought to put the best construction on what anyone else says. It is also a standard code in most friendships, that we assume that "he didn't mean what it sounded like", because we know him better than that, and think better of him.
Now, you can certainly use this for such scenarios as communication between a player character and an AI. Skynet might be user-friendly, but that doesn't mean its communications are going to be phrased in a manner which will communicate its intent. Robots, too, are frequently unemotive in their speech and manner--there was a Doctor Who episode in which this was a significant factor, that the robots were built with roughly human forms and countenances, but since they had no expression of emotion humans were unnerved by them, sometimes even frightened by them. So sure, there are plenty of ways this knowledge can be applied to game worlds. Simply creating a people whose emotional expressions are entirely different (and thus seemingly absent) from those of the player character can make communication frustratingly fun.
--M. J. Young
