When discussing an issue such as climate change, I am often amazed at how people can believe statements such as "the Earth will warm up by between 2 and 6 degrees in 100 years". The reasons for my amazement first, then an example to elaborate on point (3):

1) The warming is very precise. We believe computer models that say "X ppm CO2 rise = 0.XoC warming" without questioning what the numbers themselves represent.
2) The numbers involved are vast. Thermodynamically, the difference between 2 and 6 is enormous. When considered in terms of the size of the Earth, 1oC of temperature change represents an astronomical amount of energy.
3) Anyone with a grasp of such numbers, and the Earth's systems (I read somewhere that there are around 2 million "feedback" effects in the climate system), should be aware that it is just not possible to predict what will happen in weeks, let alone years.
4) Making policy decisions as a result of this "science" while completely ignoring the economic and political implications is frankly mindblowing.

Point 3 in particular requires some elaboration. I often hear that while we can't predict the weather next week, there is a "general trend" that can be followed over time. But this totally ignores the history of forecasting, and ideas such as forecast degradation and the law of iterated expectations.

Nassim Nicholas Taleb's "The Black Swan" elaborates on this much more. There is an example he uses regarding billiard balls on a table that explains what I am getting at very well, fortunately someone else has already blogged it here:

http://www.anecdote.com.au/archives/2007/10/the_billiard_ba.html

Most people are aware of the rice-on-a-chessboard example, but the billiard ball example provides us with this statement:

"to compute the fifty-sixth impact, every single elementary particle in the universe needs to be present in your assumptions!"