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

RE: Metrics: d(LOC)/d(stories) (second try)

Expand Messages
  • acockburn@aol.com
    (Trying again, this time typing before sending) ... The third peak is more interesting. I took almost a full week just ... Question: it is not clear that
    Message 1 of 1 , Dec 10, 2000
      (Trying again, this time typing before sending)

      "Barb and Alan Lund" <ablund@...>

      I took the liberty of substituting story points (SPs) for stories in the
      metric..
      I took the net number of lines added over the previous three weeks and
      divided by the number of story points completed during that time

      The third peak is more interesting.  
       I took almost a full week just
      refactoring before really starting on the story.  

      Question: it is not clear that "refactoring" would cause a rise in 1,000
      lines of code.  You did, however, write that you counted "net" gain in LOC.  
      So the newly refactored version really was that much larger than the previous?

      First, having periodic peaks and valleys is partially due to doing some new
      kind of thing for the first time, followed by doing several similar things
      immediately after.  


      Right.  Completely obvious "after the fact" (but not before) :-)

      For purposes of this discussion, it's enough to say that estimates in ideal
      engineering days that take into account the current state of the system are
      going to muddy the waters as far as the d(LOC)/d(SP) metric are concerned,
      since (in some ways) SP becomes a function of expected LOC, which out to at
      least correlate with actual LOC.  


      Very good point.  This never occurred to me... it may be connected with your
      using story points instead of stories, since your story points are based on
      work-hours to produce.  It reminds me of two good sw companies that are
      having trouble with their rates, because they charge by the hour.  As their
      programmers get better, their net profit goes down!!  They are at least
      discussing using Function Points Delivered to get out of this quandary.  
      Stories are perhaps more like the Function Points in that regard.  
      Question: does your graph look about the same if you use Stories instead of
      Story Points?

      It still seems to me that the shape of the graph (in general, not mine
      specifically) ought to trend downward over time if one is obeying
      OnceAndOnlyOnce,


      I'm not sure any more, given that periodically, new stories will show up that
      force a round of "new" writing, causing a spike like you observed.  I would
      like to hope that the Stories negate the Story Points effect you described.

      Overall your graph reminded me of my business bank account balance - I graph
      it each year, but never learn anything, because the jumps up and down are so
      severe that no monthly trend is usable.

      Thanks a huge amount for the data and discussion.  Perhaps someone else will
      comment on theirs, and we can see what other effects show up.
      cheers

      Alistair

    Your message has been successfully submitted and would be delivered to recipients shortly.