I have four comments on Seilevel's presentation I attended last night:
First, Jerry Aubin effectively argued that requirements management is vital to the success of a software project. Clearly, one of the main problems with product development is a failure to understand what to build.
Second, Jerry implied that good requirements management requires a great deal of formality. He seemed to contrast formality with agile methods. I don't entirely agree that agile methods are informal. While it's true that proponents of agile methods point out disadvantages of formality, they also tout a structured, test-driven, disciplined, and iterative approach to product development. The bottom line is that agile methods can be every bit as formal as other methods, but just in a different way.
Third, Joe Shideler insightfully pointed out that using "models" to elicit and document requirements helps to ensure completeness and accuracy. When you impose structure on specifications, you expose gaps that you can subsequently fill.
Fourth, the presentation highlighted the differences between what Cauvin, Inc. does and what Seilevel does. Seilevel doesn't do requirements! They are extremely knowledgeable and talented about writing software specifications, but these specifications are largely business process and user interface design specifications, not requirements specifications.
Joe Shideler's presentation made this last point abundantly clear. Almost all of the examples and models he presented contained detailed user interface assumptions. His click-action-response table example went so far as to specify all of the behavior of a particular button in a UI.
When a member of the audience raised the question of the line between requirements and design, Joe defined "requirement" as "something that the business user cares about". Does the business user really care about buttons in a UI? I would argue that the business user doesn't care about the UI at all except insofar as it helps her accomplish her goals within certain constraints.
I am grateful to the Seilevel folks for raising the level of discussion about requirements and specifications. And I'm glad that we have such talented people here in Austin. But I respectfully suggest that Seilevel changes its tag line from "requirements defined" to "specifications defined".
Random and Scott Sehlhorst also attended the presentation. See Random's review here and Scott's review here.
First, Jerry Aubin effectively argued that requirements management is vital to the success of a software project. Clearly, one of the main problems with product development is a failure to understand what to build.
Second, Jerry implied that good requirements management requires a great deal of formality. He seemed to contrast formality with agile methods. I don't entirely agree that agile methods are informal. While it's true that proponents of agile methods point out disadvantages of formality, they also tout a structured, test-driven, disciplined, and iterative approach to product development. The bottom line is that agile methods can be every bit as formal as other methods, but just in a different way.
Third, Joe Shideler insightfully pointed out that using "models" to elicit and document requirements helps to ensure completeness and accuracy. When you impose structure on specifications, you expose gaps that you can subsequently fill.
Fourth, the presentation highlighted the differences between what Cauvin, Inc. does and what Seilevel does. Seilevel doesn't do requirements! They are extremely knowledgeable and talented about writing software specifications, but these specifications are largely business process and user interface design specifications, not requirements specifications.
Joe Shideler's presentation made this last point abundantly clear. Almost all of the examples and models he presented contained detailed user interface assumptions. His click-action-response table example went so far as to specify all of the behavior of a particular button in a UI.
When a member of the audience raised the question of the line between requirements and design, Joe defined "requirement" as "something that the business user cares about". Does the business user really care about buttons in a UI? I would argue that the business user doesn't care about the UI at all except insofar as it helps her accomplish her goals within certain constraints.
I am grateful to the Seilevel folks for raising the level of discussion about requirements and specifications. And I'm glad that we have such talented people here in Austin. But I respectfully suggest that Seilevel changes its tag line from "requirements defined" to "specifications defined".
Random and Scott Sehlhorst also attended the presentation. See Random's review here and Scott's review here.
Comments