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

96509Re: I'm just _so_ Bad. Estimating.

Expand Messages
  • jrb32002
    Sep 15, 2004
    • 0 Attachment
      For what it's worth ....

      --- In extremeprogramming@yahoogroups.com, John Carter
      <john.carter@t...> wrote:
      > In an industry full of really lousy estimators, I'm a really lousy
      > estimator even by the industry's terrifically low standards.

      Don't tell my boss this, but I'm not as good at it as I'd like
      either. ;-)

      > a) I don't know my interrupt bandwidth. I'm a helpful guy, people
      ask me
      > stuff. When I say, "It will take me a day to do that" that's a
      joke. I
      > never have had an uninterrupted day in my life. I need to get...
      > i. A good estimate on how much time I spent in "interrupt mode".
      > ii. I need a good estimate of the variability.

      Time logs help me for this. I have one right next to me as I type
      here. For instance, so far today I've spent 15 minutes on
      administrative things, 70 minutes total on my main reason for being
      here, and 15 minutes on general training (which includes posting to
      this list, go figure :-).

      > b) I give "gut feel" estimates. Bad, bad, bad. The best phrase I
      > _ever_ invented to do with estimation is this...
      > "I don't know how long it will take me. But I can tell you how
      long it
      > will take me to give you a good estimate." I must use that
      _every_ time.

      I give gut estimates too, but only after having taken some time
      toying with it.

      > c) I _always_ have a "todo" list 10 or more deep with differing
      > frames and priorities. It's too complex for me to merge that list
      into my
      > estimation.

      I don't even try any more, citing Tom deMarco's "Peopleware"
      observation of the study results that the most productive developers
      were working on self-selected projects where the manager essentially
      said "wake me when you're done".

      Been thinking of working out a techno-solution. I expect to
      prototype it on index cards, and suspect it won't shift far off of

      > d) The goal posts are shifting, priorities shift, I just plain get
      > wrong. Therefore estimates must shift, but they don't.

      I shift my estimates all the time. Not sure yet whether the general
      culture here around me at my new job will take that in stride. After
      much inspection, it's a heterogeneous mixture. Much of my trouble is
      in making sure that larger estimate changes get communicated to the
      correct level(s) of awareness.

      > e) If it were simple, it would have been done already. Therefore
      > programming tasks are inherently more complex than what we did

      Much of what I'm trying is simpler once I get someone else to explain
      certain pieces to me. A "divide and conquer", which sometimes
      includes handing off pieces to someone else entirely.

      Joseph Beckenbach
    • Show all 16 messages in this topic