Lately I have been thinking about the subtleties/psychology that lives between-the-lines in answers and comments.
There are algorithms to try and warn posters of subjective questions, as well as a few token mechanical filters in comments for "problem cases"...such as preventing people from linking to "let me Google that for you" (it rarely represents a sincere attempt to be helpful, but prohibiting me from linking it to illustrate a point in this question would be foolish...yes?)
My hypothesis is that there should be no mechanical censorship, only warnings you may disregard (at risk of moderation). Further...I believe there would be broad impacts on the success of StackExchange if there was a warning system--similar to what runs against question titles--which checked your comments and answers prior to posting. If potential problems were detected you'd get an "Are you Sure?" with links to explanations of the issues.
The warning filter might start off simple enough. Basic example: it could look for words like "stupid". But the idea would be to imagine the warning step as ideally passing through someone who knew the person you were responding to really well, and knew their triggers. (Again: This shouldn't stop you from posting what you want to post, just to raise awareness...the question title warning works in the same way.)
I'm only suggesting the introduction of this step, and the importance of being able to ignore its feedback (even in the lmgtfy scenario). I bet some simple heuristics could remove roadblocks to communication.