Australian Scrum Community

Managing Software Development

Posted by Andrew Hallam to 12 July, 09:44 PM

Reg Braithwaite has an excellent post contrasting deterministic and probabilistic approaches to managing software development.

Theory P adherents believe that there are lies, damned lies, and software development estimates.

Reg gives an opinion on which approach wins based on the evidence and which is winning in the marketplace. I won’t spoil it for you, but will say that it is well worth the read.

Comment

    1. Richard Banks

      13 July 2007, 17:36

      Nice one Andrew – that’s a really good essay and says a lot about both camps without descending into emotional argument.


    1. Rowan Bunning

      16 July 2007, 16:22

      You can find a good summary of the two contrasting belief systems at the top of this interview.

      I think that the reason why software development is not a defined, deterministic process is clear when you consider a very clear example of a deterministic process e.g. manufacturing 1,000 more widgets in a factory. In that scenario you have a clear set of steps at a highly granular level and you can be very confident that executing those steps in the same way that you executed them to make the first 1,000 widgets will successfully result in another 1,000 widgets like the first lot minus any variability (e.g. machine malfunction – which is what you are trying to reduce in this environment). There are lots of operational processes that are like this.

      A typical software development project exhibits completely different characteristics from the sorts of completely definable, deterministic processes like manufacturing widgets or processing loan applications.

      Does anyone disagree or have another good source on this?


Commenting is closed for this article.

  :