> The idea that multiplying a series of uncertainties together does something other than blow your error range out is such a weird idea Might seem weird but i

> The idea that multiplying a series of uncertainties together does something other than blow your error range out is such a weird idea Might seem weird but it is massively useful. Having an educated estimate that is _only_ out by an order of magnitude is a massive improvement over wild-ass guesses that are 4-5 orders of magnitude off -- especially in systems design where over- or under-provisioning leads to millions wasted in effort.