I wish people would quit using this as an example of a fallacy. With the gross uncertainties involved in predicting presidential popularity, "current popularity" is probably the best predictor available.

Not true. Popularity isn't an efficient futures market.