It only works when you need to calculate some round numbers or to quickly see if the calculations are logical. When it comes to a real life Rule of 72 is too approximate.
The number of external factors involved in predicting CO ski lift tickets a decade into the future so immensely overwhelm the model implied by the Rule of 72 as to beggar imagination.
So, yeah. I'd be surprised.
Does some cool work with it. I've seen many examples from his lectures where it worked quite well.
Edit: wrong sign, doh