The need for smaller protocol specifications

[Erik Nordmark]

> We HAVE been setting higher and higher standards for our work. I keep
> expecting the next trivial protocol proposal to have to solve the
> distributed-PKI problem while proving the NP-completeness of general
> relativity and demonstrating a solution to world hunger all in order 0 time.
> 
> Seriously, I think the bars are set at the wrong levels -- the first rung is
> just too high, the second is unachievable, and the third is unimaginable.

Devil is in the details I suspect.
The hard question is what should the criteria be.

While I can see the community agreeing on some general requirements
(like reasonable security; congestion control) it is a lot harder to ssee
where to do incremental work when it comes to things like
 - doesn't handle all the aspects of the layer's it's built upon
 - doesn't fit together with other protocols sitting next to it
 - doesn't provide the assumed set of services to other protocols e.g.
   those layered on top of it

While prptocols that don't have all of the above might be useful
to specify and implement as part of a better understanding of the
problem and solution spaces, it would seem odd to call somethhing
an IETF standard it is doesn't fit into the puzzle of existing
standards.

So perhaps there is an argument that such "components" be experimental
RFCs and when they fit better into the puzzle they can move to the standards
track. Making them fit might involve modifications to the protocol
itself, or the addition of some more components that provide the glue.

  Erik