Gregory's Server
@gooba42 @Alice The error margin is supposed to statistically even out over lots of tickets. Which would work if the estimation errors were randomly distributed, but they never are, because software engineers are incurable optimists. So your best bet is to take the median and double it. That's usually a solid estimate of the actual costs.
@kolya @gooba42 @Alice
I practice the Montgomery method. Under-promise and over-deliver. My estimates are conservative, and most of the time I get it done, and then some. If there's an unforseen issue, I still get it done, I just don't have enough time to go above and beyond.