Comparison
AnchoringvsEstimate
Anchoring
someone said two weeks before anyone had estimated, and every number afterwards was near two weeks.
The first number mentioned pulls all subsequent judgements towards it, whether or not it was informed. It is why an executive's guess out loud damages an estimate before it is made, and why estimates are better gathered independently and then compared. The simple mechanical fix is to have people write numbers down before anyone speaks.
Full entry →Estimate
you said about three weeks, and heard it repeated in a meeting as the date it will be done.
A probabilistic statement about how long something is likely to take, given what is known now. It is not a promise, and the entire dysfunction of software planning comes from the two being conflated somewhere between the engineer and the roadmap. Stating the confidence and the assumptions along with the number is what keeps it an estimate on its way up the organisation.
Full entry →