On Mon, Mar 21, 2016 at 10:17 AM, Jari Arkko
<jari(_dot_)arkko(_at_)piuha(_dot_)net> wrote:
Yes, but I would approach this from a slightly different end-point. It is
always the case that convincing others of something new is a lot of work. In
my mind, standards or open source work both succeed best when performed by
those who actually are deploying and in charge of building mainstream systems
(commercial or open source) for the topic. So it is not so much about IETF
folk reaching the adopters, but having the adopters drive the IETF work…
One of the problems I see is that 'consensus' is wielded as an input
to the process rather than being the output.
People are already telling me that there is 'no consensus' for my
proposal for the LURK BOF. How on earth is there supposed to be
consensus if we haven't even met yet?
The mode of failure I keep seeing in IETF is the following:
1) A very narrow scope is decided 'to focus'
2) Poorly thought out aspects of the proposal are defended because the
problems they cause are 'out of scope'
3) The resulting RFC describes a protocol that is worse than useless.
4) The proposal fails in the market.
5) The experience is used as 'proof' that the problem is insoluble. (optional)
What should be obvious, I hope, is that the narrower your scope, the
narrower the appeal of your solution.
The risk in considering a wide scope is that the solution becomes
overly complex. But it can also make the solution simpler by forcing
the designers to consider what the fundamental issues are and deal
with them in a modular, extensible fashion.
I have a proposal:
* Use cases are not subject to consensus calls.
* Requirements derived from the use cases are.
What I want to use a specification for is really none of anyone else's
business. When a person proposes a use case, what they are essentially
saying is 'this is what the specification has to do if you want my
support'. So why should that be subject to a consensus requirement?
During the SAML design, there was a working group on use cases that
spent a lot of time voting on use cases. Some of the work was useful
in that it did help clarify what the use case was. But the time they
spent arguing over whether given use cases were in or out was a total
waste of time because I was writing the architecture specification and
I made sure that I could meet every use case anyone ever proposed.
In design, I actually find the more absurd use cases turn out to be
the most useful because they can lead to a different way of thinking
about the problem.