The "late surprise" problem

[Jari Arkko]

Henrik Levkowetz wrote:

> Hmm. In software development, it has been recognized for 8 years or so
> that the waterfall model gives far to long development cycles --
> iterative development lets you work on specification, design,
> implementation and testing in parallel, giving faster results, better
> throughput, better understanding and better solutions. So why this
> rather heavy reluctance within the IETF to look at possible solutions
> until the problem statement is polished and gilded?

Good point, Henrik. I agree.

In fact, I suspect it applies even more generally than just for
the problem definition WG itself. We try to do the perfect
protocols -- start with requirements, then do a full-featured
protocol, and beat the specification until it is perfect... and
in many cases leave the RFC at PS and not revisit it. Classic
warefall model.

We do have some bit of "running code" in the side of the
RFC development work. So hopefully some of the experience
filters to the final RFC as well. But the RFC development
itself is strictly waterfall. No smaller pieces can become
ready before the full-blown thing is done.

Jari