Complex Problems (Was: Re: Discipline of Internet Protocol Engineering)

[Keith Moore]

> Breaking down a complex problem into smaller pieces that are
> individually useful does two things that are quite good:

divide-and-conquer is all well and good, provided the pieces don't 
conflict when we try to use them together.  it's not divide-and-conquer
I object to, it's our too-frequent failure to provide an architecture
that informs the design of each of the pieces in a way that allows
them to work together.

also, there are too many cases where one group tries to take something
that is already in use by others and change it in a way that favors
the group making the change while harming its functionality for
others.  NAT in IPv4, linklocal addressing in IPv4, and content
negotiation over SMTP, are all good examples.

Keith