June 11, 2004

Abstract Models and Concrete Tests

At XP2004, Someone (Goldfish attack!) mentioned that only 5% of people can think in terms of abstract models. This is probably why we get such poor feedback from the business when we present them with models.

I explained our approach to the requirements on my current project. The business users have specified a list of representative trades. The business analysts are preparing the trades for sessions where the business users are specifying how they want the system to react for each trade. In effect, they are specifying their requirements in the form of user acceptance tests.

Henry C.T. Andrew suggested that the business coach should present the users with the abstract model and then the concrete results. e.g. 2 + 2 = 4. This way the users would not need to think in terms of abstract models. They could express their requirements in concrete terms that they are more comfortable with.

I said that it would be better to present the abstract model and then 2 + 2 = ?
and et the business user to specify "?". There would be useful learning from the exercise of discussing the answer. The discussion might be "Its four", "No, its four point zero zero to two decimal places", "No, it does not really matter, four is close to five so lets round it up to ten."

Posted by chrismatts at June 11, 2004 4:55 PM