It has a net negative effect because people then go around saying (this post will be deleted after policy implementation), "Oh, look, LW is encouraging people to commit suicide and donate the money to them." That is what actually happens. It is the only real significant consequence.

Now it's true that, in general, any particular post may have only a small effect in this direction, because, for example, idiots repeatedly make up crap about how SIAI's ideas should encourage violence against AI researchers, even though none of us have ever raised it even as a hypothetical, and so themselves become the ones who conceptually promote violence. But it would be nice to have a nice clear policy in place we can point to and say, "An issue like this would not be discussable on LW because we think that talking about violence against individuals can conceptually promote such violence, even in the form of hypotheticals, and that any such individuals would justly have a right to complain. We of course assume that you will continue to discuss violence against AI researchers on your own blog, since you care more about making us look bad and posturing your concern, than about the fact that you, yourself, are the one has actually invented, introduced, talked about, and given publicity to, the idea of violence against AI researchers. But everyone else should be advised that any such 'hypothetical' would have been deleted from LW in accordance with our anti-discussing-hypothetical-violence-against-identifiable-actual-people policy."