Re: agile/xp question (formal analysis)

From: AndyW (foo__at_bar_no_email.com)
Date: 11/26/04 

Date: Fri, 26 Nov 2004 11:48:19 +1200

On Thu, 25 Nov 2004 09:05:04 -0600, Robert C. Martin
<unclebob@objectmentor.com> wrote:

>On Thu, 25 Nov 2004 13:12:40 +1200, AndyW <foo_@bar_no_email.com>
>wrote:
>
>>I do requirements analysis and I dont work on small projects where we
>>would be able to deliver anything to any customer that early. Maybe
>>after a year, but not during the startup phase.
>
>In a study of 23,000 projects it became clear that project duration
>was a significant predictor of failure. None of the projects longer
>than 36 months succeeded. Of projects whose first delivery was within
>6 months about half succeeded. And at the one year mark only 20%
>succeeded.
>
>(See Craig Larman's "Agile and Iterative Development: A Manager's
>Guide" for more information on this study.)
>
>>And I suspect thats the difference, large project experience vs small.
>>At the end of the day XP/Agile use all the techniques out of
>>waterfall, just re-arranged to make it seem faster, but as soon as you
>>scale you end up with waterfall again (well actually mini-waterfall).
>>
>>They just wont admit it.
>
>Because it's not true. In his 1987 report to the DOD on Military
>Software Fred Brooks said: "...the document-driven, specify-then-build
>approach [] lies at the heart of so many DoD software problems..."
>
>The underlying principle of waterfall is *phases*. The transformation
>of requirements to design to implementation through a series of phases
>that are independently scheduled and measured. The "allure" of
>waterfall is the mistaken belief that one can measure the progress of
>the project by measuring the progress of the phases. Waterfall uses
>the completion of the analysis phase, and the design phase, as an
>indication of true project progress. XP does not do this.
>
>XP's *only* measure of progress is: passing acceptance tests. You
>cannot make progress in XP unless you produce real code that passes
>real tests that define real features. No other measure of progress is
>acceptable.
>

You know as well as I do that XP/Agile = Mini-Waterfall, so comparing
it against waterfall isnt really the right thing to do.

Its like comparing a ferrari with an model T.

It was well recognised during the early 80s that the lifecycle of
waterfall needed to be iterated in a cycle - this is where the spiral
methods came about and where the origin of the Agile lifecycle was
formed.

The document driven methodology has nothing to do with waterfall as
such, its just one of the techniques used at the time to implement it.
One could have been using Yourden struction design, or Ward Mellor -
Or in 1987 we were working in using the basis of formal OO methods.

I think fundamentally when you trot this study out you are confusing
the process with the techniques used to implement the process. The
latter changes quite a lot from company to company and in the 80s it
was well known the DoD were having problems with their implementation
- but it was due to management and organisation structure and politics
more to the actual techniques being used.

Relevant Pages

  • Re: agile/xp question (formal analysis)
    ... >scale you end up with waterfall again. ... the project by measuring the progress of the phases. ... XP's *only* measure of progress is: passing acceptance tests. ... "The aim of science is not to open the door to infinite wisdom, ...
    (comp.object)
  • Re: agile/xp question (formal analysis)
    ... > the completion of the analysis phase, and the design phase, as an ... > indication of true project progress. ... mini waterfall environments - no need to mention XP here, ... Regards, ...
    (comp.object)