Some good points/ideas from this MS talk: - cone of uncertainty - estimates early on suck, but slowly converge on accurate - ample use of probabilities and co

Some good points/ideas from this MS talk: - cone of uncertainty - estimates early on suck, but slowly converge on accurate - ample use of probabilities and confidence intervals in estimates is a must - stand up for your estimates