On 1/23/13 1:27 AM, Hannes Tschofenig wrote:
Hi Brian, Hi Joel,
the point of my mail was not to start a discussion about the examples I provided but to note that the suggested "let's reduce complexity by reducing options" is not as easy as it sounds in practice.
The prototypical human enterprise has need to innovate which means
finding a way to bring products and services which people want buy to
market. This process is both one of discovery and failure (market
signals). Part of the process which respond(s) in an inverse fashion to
market signals is effort required for standardization (e.g. increased
demand and therefore resources makes it harder not easier). It is is
relativity easy at the margins to wing documents for things which nobody
really cares about through the IETF (one can contrast this with other
standardization processes if you like), the result being that you have a
record of a diversity of lightly used but possibly important to someone
or potentially no-one solutions. When you have a great number of groups
involved in the development of what is currently a hot area of work, the
amount of effort associated with getting those drafts through a process
is dramatically larger, they are subject to evolutionary pressure, and
they have to satisfy the needs of a great number of constituents, that
probably if not always produces a better outcome as far as utility, and
interoperability as concerned but the efforts at the margins don't stop
because of it.
In the context of the document Stephen wrote and the proposal that was made on the list I was wondering whether a bit more analysis about what the problems we are trying to solve would be helpful. As some folks had raised implementing specifications is not a sufficient condition for interoperability, or successful deployment. I personally believe that IETF working groups are a fine job in writing code along with their specification work.
Ciao
Hannes
On Jan 23, 2013, at 9:58 AM, Brian E Carpenter wrote:
On 23/01/2013 04:14, joel jaeggli wrote:
On 1/22/13 12:34 AM, Hannes Tschofenig wrote:
Another example from a different area: Why do we need so many
transition technologies for the migration from IPv4 to IPv6? Wouldn't
it be less complex to just have one transition mechanism?
You mean no transition mechanisms...
That was, in fact, the plan (dual stack during coexistence).
It went wrong.
Brian