We may have to agree to disagree on this point. From my...

We may have to agree to disagree on this point. From my perspective (as I believe I've demonstrated here) he completely screwed up his interpretation of the temperature graphs he presented.

As to his argument that you can't derive an average temperature of the Earth to that level of accuracy - I haven't touched on that until now, but it's pretty half-baked, too. An analogy that might help demonstrate:

Imagine you've got a large balloon full of a mixture of helium, neon, argon and xenon. Somewhere on the order of 10^23 atoms, whizzing around with speeds that vary over an enormous range:


You want to measure the temperature of the balloon. Surely, with such a huge number of atoms, and with the fastest moving well over 1000 times faster than the slowest, that must be impossible? Surely the average you measure must be meaningless, or only accurate to within tens of degrees? Yet, when you stick a thermocouple in there and wait, you'll find it settles at a temperature of 298.15K (25 degrees Celsius).

It's a basic rule of statistics: the more points you sample and the longer the timeframe you average together, the more precise your estimate of the average becomes.