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

Re: [XP] Interesting story breakdown experience

Expand Messages
  • Ron Jeffries
    ... Yes, interesting example. When you stated the problem but hadn t given your solution yet, I was thinking, Hmm, I wonder if we could replace the X-Ray
    Message 1 of 4 , Jan 14, 2005
    • 0 Attachment
      On Friday, January 14, 2005, at 9:08:12 AM, William Pietri wrote:

      > Suppose you're building a digital medical X-ray machine. For the first
      > release, there's one feature: you put the person in, press one button,
      > and you see one perfect picture on the screen. Simple, right?
      > Unfortunately, people who have built a prototype to figure out whether
      > the project is even doable tell us that under the hood there's a lot of
      > work, 30-70 points of work.
      > Since there's only one feature, we can't split the cards that way.
      > Normally I might break cards down by speed (image in 1 hour; imagine in
      > 1 minute; image in 1 second), by scale (1 square inch of image, 10
      > square inches, 100 square inches, full size), or by domain fragment
      > (view an arm, view a liver, view a skull, view a whole body).
      > But the researchers tell us that the algorithms they've come up with
      > aren't amenable to splitting that way; the first card on any one of
      > those is still 90% of the work, after which the rest of the cards are
      > easy.

      > So the insight my colleague, Brian Slesinsky, had was to break things
      > down by quality. We start by taking the raw sensor data and putting them
      > directly on the screen. That would be pretty much useless for the end
      > purpose, but it would be a step forward from a blank screen. Then all of
      > the cards are about adding specific data processing techniques that
      > incrementally improve image quality until we reach something of
      > shippable quality.

      > This opened my eyes to something I hadn't really considered before: you
      > can still build a high-quality piece of software that initially produces
      > low-quality results.

      > Does that help clarify things?

      Yes, interesting example. When you stated the problem but hadn't
      given your solution yet, I was thinking, "Hmm, I wonder if we could
      replace the X-Ray camera with a web-cam, to get a picture of
      something on the screen." That would have let me go end to end, but
      would still have left the next step a big one.

      Your example also reminds me of the first paycheck on C3, done in
      the first week of the project. We paid one guy, something like $5
      and hour for 40 hours, and deducted 25% income tax, and printed a

      After that it was all just refinement ... what if he doesn't make 5
      an hour, what if he works different hours, what if there was another
      employee, ... and so on.

      Nice example, thanks!

      Ron Jeffries
      Any errors you find in this are the work of Secret Villains,
      whose mad schemes will soon be revealed. -- Wil McCarthy
    Your message has been successfully submitted and would be delivered to recipients shortly.