I've Been Thinking

Cargo Cult Architecture
  • Home
  • Architecture
  • If you don't know about Cargo Cults, then see below for some references:

    There is a commonly held view that best practice for developing an enterprise or other architecture should be:
    • Document the as-is situation
    • Define the to-be environment
    • Perform a Gap Analysis
    • Initiate a project to close the gap.
    There is a naive belief that following this process will deliver some sort of benefit to an enterprise. I assert that this is simplistic and analogous to a Cargo Cult.

    The problems with the "best practice" approach are:
    • Architectures can be defined by identifying "requirements". See my comments on requirements;
    • There is no clear connection between "to-be" environment and any sort of business problem, and hence has some sort of business value;
    • A "to-be" environment can be defined, relatively easily. In practice it is likely that the "to-be" is a Wicked Problem.
    There are probably many reason for Cargo Cult architectures, here are but a few:

    The enterprise (and too often the architect) does not understand the optimal role for an architect

    An architect works in the space between the business, who owns the problem, and the solution providers, who can deliver an appropriate solution. The role of the architect should be to facilitate the business in describing their problems and the solution providers in matching their solution to the problem. Applying standard, simplistic processes to complex and wicked problems just won't work.

    Cargo Cult architects are very familiar with technology solutions and tend to think in terms of solutions. They are adept at producing pictures of current or “as-is” environments and at developing future or “to-be” architectures based upon patterns or products with which they are familiar . These documents are meaningless without the context of the business problems that need solving. Architecture should be part of a problem solving process. These architecture deliverables are all well and good but they are not solutions to problems.

    Background and References to Cargo Cults


    Cargo Cult Management
    Mike Speiser Jun. 21, 2009

    Cargo Cult Science
    Richard P. Feynman
    Engineering and Science

    Feynman Cargo Cult Software Engineering
    Steve McConnell
    Aug 15, 2003

    Bernard Robertson-Dunn 20011
    Home        Architecture