The job of a poll isn't to predict the future, unless they're specifically asking about a future event, they're about measuring exactly what public opinion is at a specific point in time. That's why you see a lot of polling questions with stuff like "if the election were held tomorrow..." and such. You then use a series of polls to predict the future. Good polling is like digging core samples when you're looking for coal. You go along an area digging holes and use the data you get to predict trends.You seriously don't understand why it would be more impressive to predict an even a year ahead of time vs a couple of days?
As such there isn't much value in polls that try to predict events too far out. Whether those polls end up being accurate or not is irrelevant and to hold up the results as proof of their merit is just a selection fallacy, since you're only using single polls at a time.
If we are going to debate the quality of their polling though this is a pretty big deal: http://fivethirtyeight.blogs.nytimes...rmed-strongly/
Rasmussen's polling was the farthest off the across the board in the last major election. I mean they had 13 polls off by more than 10%, and one off by 40%
On top of that their methodology leaves a lot to be desired.