Re: prerequisite for change (was Re: draft-housley-two-maturity-levels)

[John C Klensin <[e-mail]>]

--On Sunday, January 30, 2011 1:01 PM +1300 Brian E Carpenter
<[e-mail]> wrote:

> Hi Scott and John,
> 
> I don't see this as inconsistent with the current 2-stage
> proposal, if the latter's omission of a requirement for
> independent interoperable implementations for stage 2 is
> corrected.

I won't try to speak for Scott, but I agree.

> I don't, however, believe that the problems are separable.
> The bar for PS has crept up, IMHO, precisely because the bar
> for DS/STD has appeared too high to be readily attainable.

I think one can read the symptoms, and cause-effect
relationships, in different ways.  However, I think it is safe
to say that --

	* as the bar for PS creeps up, there is less energy,
	inclination, and motivation to move toward DS/STD
	
	* as the percentage of documents --out of those that
	represent specifications of protocols anyone cares about
	-- that are actually advanced to DS/STD goes down, the
	incentives to raise the bar for PS increase.

In other words, regardless of what one believes about cause and
effect, there is a positive feedback loop operating here.

Collapsing STD onto DS is unlikely to affect that loop.  There
has been no evidence offered that such a collapse will increase
the number of documents that advance to that level.   Indeed, by
_adding_ requirements to a combined DS/STD level, it is at least
as likely to raise the perceived threshold for getting to DS/STD
without significantly increasing the motivation to push
documents into that level.

On the other hand, reverting the definition of PS both makes
that level faster and increases the motivation to move a
document to DS/STD because that second level become the first
one at which a reader can really expect a complete and
comprehensive spec.

That is why we used used the term "prerequisite".

> So I see two ways forward that hang together:
> 
> 1. draft-bradner-restore-proposed +
> (draft-housley-two-maturity-levels + independent interoperable
> implementations)
> 
> 2. draft-loughney-newtrk-one-size-fits-all-01 (i.e. simply
> abolish the second and third stages, and make interoperability
> reports optional)

> I prefer #1.

So do I.  But we agree that, absent the commitment to
interoperability testing as a critical part of the standards
process -- the one thing we claimed for years was what made IETF
work almost unique and almost uniquely successful among SDO--
there is little value in having a multiple-step process.

regards,
      john