For my Agile and TOC  friends, a neologism: Lesstimating. 

It's tempting, for some, to abandon estimating entirely because the answer is always "wrong".  

It's tempting, for others, to spend ages estimating because, the more you understand things, the more accurate your estimates feel.  

I prefer to lesstimate: spend considerably less time estimating, spend more time ensuring you can cope, elegantly, with the fact that your actuals will not equal your estimates. (If they did then we wouldn't call them estimates, we'd call them facts.)

Estimating suffers from diminishing returns. You quickly come to an accurate range (2-4 weeks, most likely 3) but after that you start adding pointless precision (2-4 weeks, most likely 22.5 days). If you are not meaningfully moving the range, or the midpoint then move on.