[harveywi]

> proven, reality-based statistics

Ex falso quod libet. The statistical mechanics can be rock solid, but if the data is garbage or the model is garbage then the prediction will likely be garbage. Hiding behind mechanical minutiae is no excuse for the overtly bold and wrong predictions.

[pfortuny]

So, why did they not call the data garbage instead of giving us a number?

As with credit agencies... "We do not give advice on investments", yeah but everybody invests based on your ratings.

[woodchuck64]

> if the data is garbage or the model is garbage

There's no better reality-based approach than aggregating polls across the political spectrum. But who would have thought that all polls would be wrong in the same direction?