Saturday, February 6, 2010

Frederick P. Brooks, Jr. on estimates


Today's quote: Here's a quotation that is a favorite of mine drawn from Fred Brooks, Jr.'s "The Mythical Man-month"

"It is very difficult to make a vigorous, plausible, and job-risking defense of an estimate that is derived by no quantitative method, supported by little data, and certified chiefly by the hunches of the managers"

Bookmark this on Delicious


  1. This is a great quote, the discussion of which is always important for project management. This discussion can help us to move forward in the understanding of what today is called quantitative project management. In this regard, the main question is as follows. Is it possible to reduce the quantitative project management to the bottom-up statistical processing of project data, or it is something more that represents itself data invariant mathematical theories of projects.
    A detailed discussion of this issue can be found here:

    Pavel Barseghyan

  2. Pavel: insofar as you on a path to invent a better parametric model, as described in your paper at, more power to you!

    Of course, many have gone before. The secret to the successful models seems to be to have a lot of tuning knobs, dozens in some cases.

    Re modeling the human dimension: in addition to linear and non-linear, and deterministic and stochastic, take into account that human behavior is not stationary in the statistical sense [morning people don't function in late night teams], and people are also prone to irrational responses--different from non-linear that is functionally related to stimulus--and chaotic epiphanies.

  3. John: I would like to recommend you to look at my other nine papers on the same topic in PM World Today starting from March 2009.
    Relative to the tuning knobs: I fully agree with you, because in the past twelve years I was doing just that. But in the use of tuning knobs, there are two fundamentally different cases.

    1. There are a lot of tuning knobs at the basis of which there is no a unified model
    2. There are a lot of tuning knobs at the basis of which there is a unified model

    In the first case, the tuning knobs are helping people to draw (just draw and not to build or synthesize) that they would like to see as a desired result.
    In the second case, many tuning knobs and the unified model allow to synthesize the solution as opposed to the arbitrary drawing.
    I favor the latter approach.


Please add your value with a comment