I think a simple answer can be found in the complexity of the average MR New story.
For example, imagine a story which announces "Back to School Promotion Begins Today- Mac sales expected to double" or something to that effect.
Many would vote positive because of the prospect of being able to get a free iPod, or because they are AAPL shareholders and like the idea of doubling Mac sales.
Conversely, some might vote negative because they expected the sale to begin earlier, or if they expected the sale to feature newly updated macbooks/pros. Or if they wanted the iPhone to be included in the promotion, etc.
Most stories by their nature offer nuance and conflicting sets of interests, and for some that can elicit a negative response while for others it can generate a positive one.
So in short, the ratings are nothing more than an expression of the vast number of possible reactions any one of us could have to any number of things we hear on a given day.
