One of the failure modes I see often in today's discourse is that using a group term as a subject makes the associated verb profoundly semantically ambiguous. I

One of the failure modes I see often in today's discourse is that using a group term as a subject makes the associated verb profoundly semantically ambiguous. If I say: "Brunettes like smooth jazz." It can mean any of: * There is at least one brunette who likes smooth jazz. * Some brunettes like smooth jazz. * Most brunettes like smooth jazz. * All brunettes like smooth jazz. * Brunettes are more likely to like smooth jazz than people with other hair colors. * Brunettes are more likely to like smooth jazz than people as a whole. * Liking smooth jazz is a defining characteristic of brunettes. * Liking smooth jazz is a defining characteristic of people who identify themselves as "brunettes". * Disliking smooth jazz is a defining characteristic of non-brunettes. * Liking smooth jazz causes (some|most|all) people to dye their hair brown. * Having brown hair causes (some|most|all) people to like smooth jazz. When the group term has a long history of power imbalance (unlike brunettes for the most part) and when the verb has deep moral implications (like "murder"), then obviously these different interpretations connote wildly different things. When you take that ambiguity and place it in the context of the Internet where context is stripped and nothing is known about the audience who will be interpreting it, you are setting yourself up for misinterpretation. When you do that in a political environment where people are seeking power and stand to benefit from willful misinterpretation, you get, well, much of what US online culture looks like today. There are obviously many deep systemic problems, but one technique to try to improve the quality of discourse is to simply avoid using groups as subjects in sentences. It almost never conveys anything that can't be better expressed in some other form.