[PREV - WILD_COG]    [TOP]

PLANNING_FOR_TROUBLE


                                             September 5, 2007

Some tarnished bullets:


   o   Centralized systems can lead to big failures.
       Avoid having critical components.

   o   This also applies to human beings: leaders need to
       delegate, to recruit assistants.

   o   Experiments should be kept small.  If successful,
       work on scaling them up next time.

   o   Pin down what can be pinned down relatively early:
       Draw diagrams (e.g. floorplans).
       Make numeric estimates (e.g. hours of manual labor).

   o   Tasks under adverse conditions can take 2x longer
       of more, and adverse conditions are common.
       Where possible, do pre-assembly, test equipment, etc.


Parts break.  People get sick.  Orders get lost.

   These problems are inevitable:
   the question is how well you
   can deal with them.
                                            But then, things like
                                            this are easy to say,
                                            but not always so easy
                                            to comply with.

                                                 Redundancy has it's costs,
                                                 and a new project may need
          Plans and estimates are nice, but      to do without for a time.
          do you need to revise them under
          every little change in the scheme?
          If so, you can spend your life on
          the revisions.  If not, they become
          useless pretty quickly.


                                             It sometimes seems like
                                             an effective strategy to
                      (In XP doctrine,       delay decision making as
                       they call this        late as possible so you
                       YAGNI: "you ain't     really know which
                       gonna need it",       decisions need to be made.
                       i.e. don't go
                       beyond the               I've had the experience
                       minimum until you        of repeatedly re-working
                       know you have to.)       mechanical drawings with
                                                elaborate solutions to a
                                                problem, where a simpler
                                                (but notably, harder to
                                                draw) solution emerged
                                                later during assembly.
                   Sometimes you need
                   to stop planning and
                   just play with the blocks.


--------
[NEXT - NO_FUTURE]