[atomicnumber3]

I tested this out too. I noticed a few things where GPs explanation is probably abbreviated:

I noticed that finger-to-nose, I was able to do the first time, every time. 100% accuracy, and with no "do a little circle to figure out the last 0.5in". As opposed to touch-fingers-behind-head, I could "feel" the uncertainty and successfully touching fingers the first time, while not impossible, was probably 10-20% success rate. I was able to improve it by doing a little (radius of perhaps 5mm?) circle with both fingers once I knew I was about in the right place, since I was usually within about a fingers-breadth. But the whole process was just a lot less bulletproof than finger-to-nose. Which I think was GPs point.

[johnchristopher]

N=2, nose is a hit 100% of the time with pinpoint accuracy but fingers behind the head only hit if I repeat the movement. Going much slower increase accuracy but it's never as good as hitting the nose anyway. At normal or increased speed I usually miss by a finger width.