A mechanism of estimation I’d not heard of.
The general idea is you make 3 estimates: optimistics, pessimistic & most likely. It assumes that these follow a distribution like this:
Then you do a weighted average. (optmistic + pessimistic + 4*most_likely)/6
So if we think: “Oh, this will probably take a week. Maybe as soon as 3 days. Maybe as long as 2 weeks.” You’d end up with 5.5 days.
Or if your pessimistic version was “maybe 3 weeks”.. it would bump the estimate to 6.3 days.