Definitions

In order to communicate more effectively, it will help to build a common understanding of some terms. I see many arguments that seem to revolve around what is, and what is not, an estimate. In the noun forms, a dictionary gives us the following definitions:

Estimate

an approximate judgment or calculation, as of the value, amount, time, size, or weight of something.

Forecast

a prediction, especially as to the weather; a conjecture as to something in the future.

Prediction

something declared or told in advance.

Conjecture

an opinion or theory without sufficient evidence for proof.

Projection

calculation of some future thing, usually based on past results.

Guess

an opinion that one reaches or to which one commits oneself on the basis of probability alone or in the absence of any evidence whatever.

These words are near synonyms, though some people assign specific differences to them in their own use. Estimate, conjecture, and guess are all very general, applying to the past, present, or future. Forecast, prediction, and projection specifically apply to the future. I might estimate how many jellybeans are in a jar, but I might forecast how many will be left tomorrow after a children’s party. Some domains prefer a particular word; both weather and company profits are generally foretold as forecasts. Future state is often termed a projection when generated mathematically, and a prediction when it is not.

I’m fond of saying that the abbreviation of “estimate” is “guess.” That often gets a chuckle, because the general feeling is that people are pretty bad at estimating and particularly bad at estimating software projects. It seems that grabbing a number out of thin air has as much chance of being helpful as anything.

Beyond that joke, estimates are distinct from guesses. At worst they are “educated guesses,” based on slim knowledge or experience thought to be relevant. Sometimes the application of past experience is very “seat-of-the-pants,’’ with insufficient consideration of the details. With a little more attention to what’s similar and what’s different between now and past experience, estimates can become truly useful.

In conversation, it appears that many people have a more restricted concept of an estimate. They often associate it with whatever type of estimate has been most prevalent in their experience. For software developers, the most prevalent form is usually “when will you be done?” Given that this “when question’’ is often asked before knowing what will be done, it’s understandable that the word “estimate” has a bad connotation for many software developers.

Nothing about estimate or forecast states how either is formed, but many people closely associate estimates and forecasts with the technique they most frequently use, or see being used. A weather forecast may be based on the amount of pain felt in the knee, knowledge of last week’s weather, or on a complex model of atmospheric humidity and pressure. A software delivery estimate may be based on knowledge of a similar project, on a detailed decomposition into tasks that are independently estimated, or on a model of project attributes.