XpMeetsBusinessAnalysts

There was a great deal of valuable and lively discussion from Chris' presentation. Thanks ChrisMatts for being brave enough to present to XtC. Please place comments or opinions below.

Is the BusinessAnalyst? a coach for the customer, much as we have a coach for the team?

The title of ChrisMatts paper might be better as "The Customer Coach" than "Extreme Analysis" to avoid confusion with Scott Amblers work on UML, etc..

Would like to see more thought as to the BA's role in short iterations, where each iteration requires some business knowledge transfer.

"Ultimately we would like lots of BA's to be thinking in this manner and tuning the way they work as part of an XP team."

It seemed like a lot of the work ChrisMatts described was to establish an XpMetaphor? for the project. -- RachelD

On our XP projects, we used the BA as a "Customer Proxy" because in an organisation the size of ours it is impossible to have a single (or even small number) of "Customers". There are many stakeholders from a variety of areas of the business and they all have a (very often different) view on the detail underneath each User Story. They are usually geographically distant across many sites. On the other hand, the developers need to be allowed to get on with implementing the stories. They don't have time to mediate in the "lively debate" that happens between stakeholders. So, we use Business Analysts to mediate between stakeholders and then act as the onsite customer to inform the development effort. We did have to take into account the downside of this approach which was a loss of the speed of response that is associated with a real onsite customer. We did this at the story estimation phase, and on the whole it worked well for us. - MattStephenson

Many thanks to ChrisMatts for being willing to talk to what he feared was a hostile crowd. Not only was it an interesting talk that kicked off interesting discussions, but it turns out we agree on many fundamental issues.

One thing that came up in the presentation and discussion was the role of documentation. Both ChrisMatts and the XPers agreed in that they did not like generating reams of documentation that nobody ever read.

ChrisMatts presented his idea of documentation as teaching aids that could be discarded once the lessons they taught had been learnt by the team.

XPers argued that XP does not want no documentation. Actually XP generates lots of documentation, it's just that the documentation is in the form of readable, executable code that either specifies what the system or classes should do (tests), or how it does it (functional code).

XPers wanted the BA to provide documentation that can be manipulated by automatic acceptance test suites, or even executed as acceptance tests.

Ward Cunningham has recently been writing about his FIT framework (http://fit.c2.com) which seems to embody this very idea. FIT stands for Framework for Integrated Test, and is a framework for extracting acceptance tests from documentation.

--NatPryce