CS190 Essay  --- Ning Ge (nige)

In chapter four of The Mythical Man-month, Frederick P. Brooks, Jr. talked about conceptual
integrity in software development.  As he points out, conceptual integrity can be achieved
by separating the archirectural effort from implementation.  He said, "Conceptual integrity
in turn dictates that the design must proceed from one mind, or from a very small number
of agreeing resonant minds."  I agree with him completely on this matter.

I believe architecture--the detailed specification of the system--should not be completed
by a large group.  Having more people involved in this process increases neither efficiency,
nor quality of the specification.  And specification of a system is such an important part
in a software develpment process that without a clear specification, the implementors will
 not know what exactly they should do.  I did an internship last summer working on a driver
program for a prototype machine.  It was a really small system with only two programmers and
two enginners working on it.  I was not familiar with the field of engineering and machinery,
I wanted to first find out about what exactly this program does.  But there was no specification.
I had to start from scratch to work out the requirements with the two engineers, who kept
changing what they wanted in the program.  The process was unefficient and time consuming.
I do think that 'conceptual integrity' may not be such a problem if the target program is
relatively small.  Brooks said in chapter four that 'Architecture must be carefully
distinguished from implementation'.  It is true for large-scale systems, but not necessary
 for small projects, in which case implementors could also be involved in building
architecture.  During my CS32 project, architecture and implementation were not exactly
separated and it worked out fine in a group of four.

In the case of large-scale system developments, I think it's not neccesary, maybe even harmful
 to have a large group of people involved in design, because when too many ideas are proposed
 during an discussion, people natually try come up with more creative ideas.  During this 
process, they sometimes forget about the most important base specification.  It's easier to come
up with new features, but not so easy to make a solid, expandable base.