Loading ...
Sorry, an error occurred while loading the content.

6840Re: [agile-usability] Real data

Expand Messages
  • Michael James
    Feb 1, 2010
      On Feb 1, 2010, at 8:34 AM, Dave Rooney wrote:

      United Flight 232 DC-10

      This is a great case study of the kind of collaboration we want to create on agile teams.  For those who haven't read it, the crew made the best of a disastrous situation by learning how to fly what was left of the airplane by varying the throttle settings on the two remaining engines.  Once they had some mastery over that, they attempted a landing which was not 100% successful but saved a lot of lives.  A similar situation occurred to a DHL cargo plane over Baghdad after it was crippled by an enemy missile.  The crew collaborated to learn how to fly the plane only with differential thrust, and then landed with no loss of life.

      The learning process requires lots of low-stakes failures.  Looking at the Apollo program, we see that encouraging lots of low-stakes failures reduced the risk of high-stakes failures.  If someone asked me to build avionics software today, I'd start with the tightest possible feedback loops, such as test-driven development, and pair programming.  I'd require much more frequent inspect and adapt cycles with the customer, so developers wouldn't suddenly discover they're 9 months behind schedule and work 70 hour weeks to catch up (which leads to sloppy thinking no process can correct).  I'd probably retain Independent Verification and Validation *in addition to* cross functional teams.

      --mj (who is not Michael Feathers, but liked his book)

    • Show all 218 messages in this topic