InfoQ Presentation: Selecting the Right Methodology and Steering it to Success

It's easy to agree on abstractions like "anything more than 'barely sufficient' in a software process is waste," but it gets more complicated when the rubber hits the road and we need to customize a process for a particular project. In this presentation from the APLN Leadership Summit at Agile2006, Todd Little shared a model developed to help choose the right "flavour" of Agile for different kinds of projects. Their approach included understanding the project's and team's parameters and their relation to the "sweetspot" addressed by Agile processes, and then actively steering as development progresses. Clearly a visual thinker, Little's presentation is full of interesting visuals, including graphs.

Todd Little is an AgileAlliance board member and Senior Development Manager with Landmark Graphics, which creates shrinkwrapped software for oil & gas production companies. Landmark needed to be responsive to a market apt to require abrupt changes in product direction. In applying Agile to their project portfolio, which includes over 60 products and distributed development, Landmark adopted a "cone of uncertainty" approach, a concept used by the weather service when predicting cyclone paths. Accordingly, the energy to be invested in features was based on their "nearness to execution" within the product backlog and portfolio backlog. This resulted in an approach where, according to Little,  "we know generally where we are going - setting out for a general area, and narrowing in."  This approach therefore required ongoing analysis, and to reflect new information as projects unfolded, it was important to continue incrementally adjusting project and portfolio plans.

Little cited observations from Boehm and Turner on "balancing discipline and agility":

• Neither agile nor plan-driven methods provide a silver bullet
• Methods are important, but potential silver bullets are more likely to be found in areas dealing with
• People
• Values
• Communications
• Expectations management
• Agile and plan-driven methods each have home grounds where they clearly dominate
• Future developments will need both agility and discipline
• Some balanced methods are emerging
• It is better to build your method up than to tailor it down

Landmark developed a set of criteria for categorizing projects, including both project and team attributes. With time they have seen their "Houston Matrix" prove useful. The criteria of the Houston Matrix were:

Project Complexity	Uncertainty
Team size	Market Uncertainty
Mission criticality	Technical Uncertainty
Team location	Project Duration
Team capacity	Dependents
Domain knowledge gaps
Dependencies

To make the matrix more usable, they gave easy-to remember names to the quadrants:




Landmark also adopted or developed practices and tools to help management adapt as time passed and parameters changed, including an Aggregate Product Plan, A/B/C (Prioritization) List, Quality Agreements, Continuous Integration, Expert User Involvement and Project Dashboard.

Little also went on to look beyond software agility to other applications for "agile leadership," as envisioned by the APLN.


Watch the entire one-hour presentation, full of images, graphs and real experience with many projects: Context Driven Agile Leadership: Managing Complexity and Uncertainty.

(Images are excerpts from this presentation itself)