Re:Back of the Envelope Calculations [pdf]
by erydo on Tuesday April 09, 2013 @03:19PM
(#5520478)
Errors cancel out when they're errors of magnitude, which is what I believe he was referring to, and is often the most important property of Fermi estimates.
If one estimate is off by 2x and another is off by ½x, those cancel out because the exponents are additive.
If one estimate is off by 2x and another is off by ½x, those cancel out because the exponents are additive.
Edit: Parent deleted comment…oh well.