That's a winner in smart-assery:

That's exactly the point he was making: give more credit to sources which have been shown righter in their previous predictions. But i guess agreeing wouldn't have given you the same ego boost.

A guess? Just because he did not give a time range? Well, he could easily have added "within 10 years". Many people did (I've been taught exactly that in 2002 during a market finance class).
But even without it, who would you rather trust now that it did happen? Someone who said it was going to, someone who said it wasn't, or someone who had no idea whether it would or not?
What you're doing is dismissing a statement claiming it does not contain enough information, which leaves you with either one that contains even less or one that contains none at all... Which contradicts your very point of wise trust allocation based on past record.

"as long as the cost of believing a superstition is less than the cost of missing a real association, superstitious beliefs will be favored."

Rationally speaking, if there's a significant positive expected return, how can that be superstition? It's just rational risk management.
In other words, some form of pragmatic knowledge.

It's pretty easy to prove: drop a _non-superstitious_ mathematician in the rustling grass and see how long he survives...
Hopefully for us, prehistoric men weren't so "clever" and we stood a chance of existing at all.