|
Abstract:
There
are basic ideas about designs that we can acknowledge, teach and
practice. I would be surprised to find serious disagreement about the
principles, but I would be surprised to find serious conscious practice and
teaching of these principles.
The
core concept is that designs have multiple impacts on our requirements and
can only fully understood in those terms. |
|
1.
The
‘Design is Means’
Principle:
Design
is a selected means to satisfy your ends.
This
has a number of implications:
-
if
any requirement changes, this can potentially invalidate all designs
chosen up to that point
-
the
justification of a selected design is the degree to which it help you meet
your requirements
-
if a
design has unknown or uncertain attributes in any of our requirements
dimensions then that design choice may be invalid or not as good as other
alternatives;
-
if a
design is required, then it can have arbitrarily bad impacts on any
performance and cost requirements
|
|
2.
The ‘Valid Design’
Principle:
A
valid design must contribute to performance goals within all constraints.
This
has implications:
-
you
must be able to prove that a design does not violate any defined
constraints
-
you
must be able to prove that a design contributes to at least one
unfulfilled performance requirement.
-
the
design cannot be justified if its only contribution has been made by
another accepted design already
-
a
change in the level required for a performance requirement can invalidate
a design or make a discarded design valid
|
|
3.
The ‘Good Design’
Principle:
A good
design contributes more value than it costs. It is a profitable design.
Implications:
-
it
is not sufficient for a design to contribute to the performance
requirements; it must not cost more than the value of the level of
performance it contributes
-
a
designer must be able to estimate the cost impacts of a given design on
all critical resource dimensions, in order to justify it
-
failure to correctly estimate the design cost impacts can cause budget
overrun or schedule overrun
-
if
there is clarity about design cost, but no numeric clarity about
performance impacts, then there is no clear argument for the design
|
|
4.
The ‘Best Design’
Principle:
The
best design of a set of alternative designs is the one that contributes most
value in relation to cost. It is the most ‘efficient’ design.
Implications:
-
selecting alternative designs is a matter of comparing both performance
impacts and cost impacts – of getting best benefit for cost
-
the
choice of one design over another depends on multiple performance and cost
attributes, and cannot be evaluated in a single dimension
-
any
design chosen on a single dimension is a ‘risk for failure’ in other
dimensions of performance and cost
-
if
the uncertainty – worst case – impact is not evaluated, then the selected
design risks may be worse than you expected, and the design may be a
totally wrong choice
more...
© Tom@Gilb.com
2005 |
