This recent paper contains my best current description of what a methodology is, and forms the basis for my tutorial on Designing A Light Methodology
. -- AlistairCockburn
The paper addresses the more serious matter of creating a fair dialog between people of opposing views, and the interesting but less critical matter of creating a taxonomy for methodologies. What is a "methodology?" Is there one or must there be many? If many, how would one know which one or elements of one to adopt? Is one "better" at some thing or in some way than another? Must they be big, and if so, how should one best deal with their size? What are the key questions to ask on the matter?
The main result is that there are necessarily multiple methodologies. Methodologies sort along two essential dimensions: staff size (the number of people on the project) and system criticality, the potential damage of undiscovered system flaws. For any point in the size/criticality space, the designers of the methodology (a) select a scope of concerns to address: which project roles, activities, deliverables, and standards to cover and which to omit, and (b) work from their beliefs, previous experiences, fears, wishes, and philosophical principles, attempting to optimize some quality of the project.
Methodologies therefore differ by the size, criticality, scope, optimized quality, and the grounding beliefs of the authors. Comparison of methodologies should focus on these different dimensions, and their relationship to the needs of the project or organization.
The ten essential topics contained in a methodology. A set of questions and curves that can be used to discuss the optimizations of the methodology. The third is what to do with the bulkiness of a methodology. I suggest that most of the bulk should go into training courses, and the rest should fit easily onto a few sheets of paper per person. The sheets name the reviews the person must attend, and the standards they must adhere to. Very few people should have to sit down and read the entire methodology.
I couldn't agree more about "One Size Does Not Fit All". I'm not sure
if the space
is continuous and parameterized. I trust MichaelJackson
's weaker hypothis that certain classes of problems can be solved by well defined methods.
My interpretation of the secondary result is that basically each member
of a project should be given:
- tools to perform their tasks, and
- the measures by which their performance will be judged
Outside of that the group member is free to perform their tasks.
This could be used as a further criteria by which a methodology can
be measured, i.e. how well it defines these attributes for each role