Agile in the Very Large – Sam Guckenheimer

I am sitting in Sam Guckenheimer’s talk at the Agile 2009 conference entitled, “Agile in the Impossibly Large – Experiences in Microsoft Developer Division.” Sam is effectively the product owner for Team System and an avid Agilista. He has been sharing experiences in Agility from DevDiv for years and freely publishes stats on Microsoft’s teams.

Here are my ad-hoc notes from Sam’s talk.

• Organizational structure is a better predictor of defect density than any other measure.

  Here are ad-hoc notes from Sam’s talk.

• Get clean, stay clean

  • Understand your technical debt
  • Put limits on acceptable technical debt
  • Clean your house
  • Keep your house clean by putting limits on defects with a stop-the-line mentality.
• Scenario > Value Proposition > Experience > Feature
• Scenario – Our business would like to
• Value Proposition – How can I accomplish …
• Experience – Here’s how you …
• Feature – As a… I want… so that…
• When prioritizing features to add to products, clients vote VERY differently than MVPs. This was illustrated in an amazing slide showing how customers vs. MVPs voted for new features.
• Red bits – development on product already in the field. Must be stringently controlled and reviewed to ensure backward compatibility
• Green bits – development on features not yet released, so fewer constraints, but always reviewed to guard against rick after the feature turns into red bits.
• Sizing for features had to take quality gates into account
  • Code complete is a dead term. You must now be feature complete.
  • This includes things like threat analysis
• It takes several weeks to run all the automated tests that execute against Visual Studio Team System.
• OGF – Overall Gut Feel – A subjective measure based on expected customer experience.
• BVT – Build Verification Tests. Tests that verify a check in – typically unit tests.
• VS2008 vs. VS2008 SP1 – SP1 saw an improvement in quality of 10X
• While it is important to celebrate success, never declare victory.
• Do not undersell the investment to start a project well, and the cost of being truly done.