I believe if the community does not have confidence that the protocol will
actually work on the Internet, then we are experimenting. I think this
definition would cover a number of protocols we would now consider for
Proposed Standard (rather than Informational), and pushes us back towards
A protocol that has no implementations and is significantly new and
different that we cannot inspect the document to have confidence it will
work as intended would be Experimental. This is both a complexity and an
exposition test as well as a measure of how similar a protocol is to other
protocols we know work as intended.
I think that when an author, or an entire work group, dreams up a
significant new protocol and brings forth a complex document with no code,
then we have a research and development effort underway, which should be
allowed to complete; that is, test it. We have a product (a real standard),
when we show that our research worked. As with most research and
experimentation, it is not necessary to have market acceptance to have a
successful research project. It is nice, and a good indicator that the
[mailto:ietf-bounces(_at_)ietf(_dot_)org] On Behalf Of
Brian E Carpenter
Sent: Wednesday, February 15, 2006 10:07 AM
To: IETF discussion list
Subject: What's an experiment?
When considering some recent appeals, the IESG discovered that
we have very little guidance about the meaning of "experiments"
in relation to Experimental RFCs. RFC 2026 refers to work which
is "part of some research or development effort" and the IESG
has adopted some guidelines to discriminate between Experimental
and Informational documents (see
But beyond that, we do not know what constitutes an acceptable
experiment on the Internet.
The IESG notes that the community could establish a variety of
guidelines describing what is and is not acceptable in experiments.
Historically, the IESG has made decisions based on its perception
that there is a strong desire in the community to publish technology
that is being deployed experimentally. We encourage community discussion
and development of more specific guidelines on operational conflicts
caused by experiments and how this should affect what we choose to
publish. (However we recommend that such discussion
focus on the general issue rather than the specifics of any case.)
for the IESG
Ietf mailing list
Ietf mailing list