That's why this algorithmic approach is actually a kind of interesting idea. Obviously the all-caps technique is too simplistic, but it would be interesting to try and give political opinions a hysterical rating based on the vocabulary used.

Obviously it wouldn't be a foolproof method; it's hard to talk about the consequences of global warming or NSA overreach without scoring pretty high in hystericalness, but it would still be a neat tool.