There's obviously a balance here which involves understanding how words make people feel and also understanding that people don't have bad intentions. You don't have to go all 1984 on it.
The point is that intent has been rendered irrelevant; all that matters is the impact of the words on the audience, and therefore the audience can unilaterally determine the impact (and thus the moral valuation) of those words. Good intentions are not sufficient to exonerate you from mob invective. Social incentives further encourage reading maximum offence into language, as being offended offers one moral leverage over others in public discourse.
