Analysis (Learning) should be barely sufficient. What does this mean?
As described earlier, an analyst uses a model to learn. They create a model and then test it. At any point, there will be a number of:
1. Known problems (issues) with the model that have not been solved. If an issue needs to be resolved before successful completion, leaving the issue unresolved is a risk.
2. A number of unknowns (risks) associated with the model.
Barely sufficient means that the risks associated with the project are understood and deemed acceptable by the risk owners of the projects. XP practitioners are risk takers and as a result are prepared to move ahead with almost no analysis (learning) about the business domain in the belief that they will be able to iterate quickly to the final solution. Their problem is convincing the real owners of the risk (the business) to allow them to proceed. Early business adopters of XP have been prepared to take on the additional business risk for two reasons:
1. They are natural risk takers.
2. The project has already failed and the downside of another attempt is not considered significant. i.e. Failure has been established as the standard against which they will be measured.
The decision to progress with known risks on a project is the responsibility of the ultimate holder of the risk. If the risk is a development risk, it is the responsibility of the development team. If the risk is a business risk (including requirements change), it is the business owner of the project that should take the risk.
If the development has been linked to the strategy of the organisation, changes to requirements should only occur due to a change in market conditions (external events), or a change in strategy. Linking an individual requirement (e.g. an XP Story or Use Case) is explained a future paper by the Agile Business Coach called "The road to Agile - Working Title".
Posted by chrismatts at June 22, 2003 8:10 AM