Nice that you brought this up. I largely agree with the direction of the post (we need to learn how to use and deal with emotional appeals), but I'd make some stronger claims.

the strength of an emotional appeal to believe X and do Y doesn't correlate with the truth of X or the consequences of Y. In fact, we are surrounded by emotional appeals to believe nonsense and do useless things.

Agreed

So, in our environment, emotional appeal is a strong indicator against rational argument.

In the memetic environment one may explain away the other (in Bayes nets once you take two independent things and have them both cause a third thing, when you condition on the third thing, the two become correlated). This means that the presence of one changes your estimate of the probability of the other (in this case, emotion makes you think that it's less likely to be rational), but it does not mean that making something emotional makes it less rational. The causality does not work that way.

If we condition on being in LW, I'd actually say that emotional appeal would positively correlate with rationality. Lukeprog is a better writer and better rhetoritician than most LWers, because he's also more rational in terms of actually going out and doing things to make him better at communicating his ideas. Like rhetoric.

Avoiding emotional appeals entirely helps you avoid irrationality by helping you avoid the problem of separating emotional appeal from veracity, but I think that in the long run learning the mental process of separating the emotional and factual aspects from each other makes you a stronger rationalist. Running away from the problem makes you weaker and less effective.

Can you somehow otherwise filter for emotional appeals that are highly likely to have positive effects?

I think that positive emotional appeals are generally better than negative or reactionary appeals.