> People who are comfortable with fast evolution, this ability to be
> comfortable with ambiguity while doing the work anyway, like the good
> software product people I've known, seem to be able to get a picture in
> their minds, and put new pieces in as the data comes in.
When doing simple programming work, there seems to often be a window
of obviousness. (I need to add one to the counter each time through
this loop.) Then there is a greater context where the particulars are
not kept in mind. (Data is received over the modem. Well, what order
is it in? I don't know; we'll figure it out when we work on the the
communications module.) The process involves tests of the data stream
and then use of the test results in designing the program. As soon as
the test is begun, the lack of knowledge begins to go away. The
question of the best design of the overall system may not be solved
as easily as writing a particular function. Methods - such as use of
program models and design and development plans - may be necessary.
Maybe more related to the quote above, it's possible to work on the
parts of projects that are defined and wait to proceed with the rest
of the project. The consequences of the many efforts or loops within
loops may be impressive, yet it may be very much like putting a
mechanical object together.
> How do we make this seen as a useful skill? Is there a way to teach
It seems to apply to any task that benefits from iterative
development and those might be tasks that are not easy to deal with
all at once. Continuous improvement is an example not in the computer
programming field. Research and writing may also be examples.
Your emphasis was on the ambiguous nature of the work, though my
response has tried to treat the question as having to do with
limited, yet working, understanding. Is my response relevant to the
Have a nice day
John Paul Fullerton
"John Paul Fullerton" <firstname.lastname@example.org>
Learning-org -- An Internet Dialog on Learning Organizations For info: <email@example.com> -or- <http://world.std.com/~lo/>