Re: [Curdle] Time to Review IANA SSH Registries Policies?

Some general thoughts on this:

(A) When I'm implementing new ideas, it would be nice to have a more
straightforward way of registering and documenting them.

(B) Registration is needed primarily for coordination. Primarily,
incompatible ideas must avoid using the same (numeric or string)
identifiers on the wire. This primary purpose can be met without
documentation. If the identifier space is large compared to demand, there
need be no specific standards that registrations must meet.

(C) However, registration also helps discover ideas developed by others. In
this case, documentation is essential if one wants to understand and use
those ideas.

(D) So quality of documentation is important. Lots of people who try
writing specs, write specs of questionable usability without lots of
guidance, review, iteration, and cajoling from others. The spec writer
experiences this as suffering. However, this suffering ensures a useful
result.

(E) Nevertheless, there is suffering. Therefore, if the process is too
onerous, the spec writer learns to avoid it.

I'm thinking (E) is a problem for the IETF - the registration + spec
polishing burdens are onerous. Two years to publish an RFC is a lot. It
vastly reduces the number of specs people are willing to write.

For most things, I'm not willing to go through this effort. Instead, I
might publish an internet-draft, and then make no effort toward an RFC
because the draft, even though informal, is discoverable, and costs 5% of
the effort of an RFC. There are lots more things for which I'm willing to
write an internet-draft, than things for which I'm willing to write an RFC.

I think the IETF could ease the registration and documentation burdens to
increase the likelihood that:

- Registration happens, for coordination purposes.

- Documentation is published that otherwise would not be, even if it's not
perfect.

I think there are opportunities here that are currently being missed.
Changing some of the SSH registries to expert review could be a step in
this direction, but it depends on what the expert review practically looks
like. If it's going to look like Peter described, where you can't actually
reach the experts, then I'm not sure if it's an improvement. :)

denis

On Thu, Feb 4, 2021 at 5:44 PM Peter Gutmann <pgut001@cs.auckland.ac.nz>
wrote:

> Phillip Hallam-Baker <phill@hallambaker.com> writes:
>
> >Restrictive registration requirements contradict the principle of
> >'permissionless innovation'. If the IETF is the place you go to get
> >permission to innovate, we are doing it wrong.
>
> +1.  To give a real-world experience of "expert review", when I tried to go
> through the process on $unnamed-wg there was an experts group that you
> couldn't contact who had a non-public mailing list where decisions were
> made
> in secret, but that didn't matter since there was no way to contact them.
> Eventually it got sorted out by one or two people involved in the process
> going out of their way to help, but when done as it should have been the
> process was more reminiscent of Kafka than the IETF.
>
> Peter.
>
>
>
>
> _______________________________________________
> Curdle mailing list
> Curdle@ietf.org
> https://www.ietf.org/mailman/listinfo/curdle
>