Skip to main content

Minutes IETF118: gendispatch: Thu 12:00
minutes-118-gendispatch-202311091200-02

Meeting Minutes General Area Dispatch (gendispatch) WG
Date and time 2023-11-09 12:00
Title Minutes IETF118: gendispatch: Thu 12:00
State Active
Other versions markdown
Last updated 2023-11-16

minutes-118-gendispatch-202311091200-02

GENDISPATCH Hybrid Meeting @IETF-118

Thursday 9 November 2023

Room: Congress Hall 2

13:00 - 14:30 Local

Log into the IETF datatracker to access:

GENDISPATCH Meeting

13:00 Status and Agenda Bash - Chairs and ADs (10 min)

The Agenda is updated by moving the third one as the second.

13:10 Removing Expiration Notices from Internet-Drafts (20 min)

Presenter: Martin Thomson, Paul Hoffman (On site)

draft-thomson-gendispatch-no-expiry

Messages

Proposal: Do away with the notion of expired drafts entirely

Replacement Mechanisms: reminders sent to update docs

Robert Sparks: Amendable to notion of removing expiry, dont see value in
putting in effort in doing so without putting in extra effort. "Expiry"
is givng newcomers and participants a way to avoid looking at time
wasting docs. Yes - and. We need the and. We need community behavior
changes. Needs to be incentive to give people a reason to do this. Need
clean up efforts.

John Klensin: Sympathetic to the problems here, agree need further
discussion, conversation foused entirely on WGs which is not the whole
community - just a subset of docs.

Paul Hoffman: We know that there are many chairs that make drafts, not
just IETF. Please send text.

John: will send text.

Ted Hardie: Like basic idea. Need mini WG to work out details. Figure
out if this will be ongoing work. Personally almost never needed expired
doc as chair. Docs are no longer deleted. Mini WG is suggestion.

Martin Thompson: we do have the ability to mark docs as dead.

Mark Nottingham: Agree with mini WG. Status of drafts/docs in IETF not
always clear. This is a bigger problem that needs to be discussed. Docs
are spread across multiple depositories in other groups, the way IETF
does it is more effective. Vote Yes.

Lars Eggert: As Individual. Open to exploring ways that are more useful
than current, lots of things to consider. Mini WG is a good idea.

David Schinazi: Vote Yes. Speaking as both draft author and Chair.
Creates extra busy work. If author has to stop work for personal reasons
draft can expire and create issues.

Rich Salz: Expire system is no longer useful. WG or AD sponsored design
team? Support doing something in this area.

Jon Peterson: View this as anachronistic. Focus for Mini WG? What amount
of metadata needs to be embedded vs live in the track? These docs get
disseminated. Valuable discussion. Need WG.

Andrew Campling: Would it be simpler to extend by two years instead of
just six months? If nothing has happened in two years then it probably
should be marked expired.

Cullen Jennings: Mini WG vs AD sponsor. Not AD sponsor - would be very
difficult to deal with all the comments. No AD would want to deal with
this. Mini WG would work better because of consensus process. Cisco had
a draft for 10 years that was just refreshed. Many times have been in WG
where expired draft is being discussed and causes confusion. Vote Yes.

Eric Rescorla: This is from bygone age. Things no longer need to expire.
Originaly was going to say AD sponsored - Cullen convinced otherwise.
Dont drag this out.

Anthony Somerset: Concur with working group. Echo Cullen's thoughts.
"Expire" may not be the right term. "Stale"?

Colin Perkins: IETF makes decisions for IETF. We cannot make decisions
for other streams. Need to consult with other streams.

Lars Eggert: While drafts are individual, they also belong to the IETF.
They are then adopted to other streams, and follow their rules. Want to
follow the rules of the Chairs, not have to make the decisions.

Dispatch outcome: Support in favor of a small working group.

13:30 Request to the Trustees of the IETF Trust to Permit the Creation of Derivative Works (20 min)

Presenter: Martin Thomson (On site)

draft-thomson-gendispatch-rfc-derivatives

Messages

Why re-license RFCs? It is the right thing to do. Other SDOs have this
option as a default.

Open Standards: like open source, except our licenses are more
restrictive.

Change Control: hard to give up control with a guarantee that you will
not get it back. Open source projects manage this risk - give anyone the
right to fork.

Lars Eggert: For open standards, if it forks the marker deployment will
figure out a winner.

Interrupt - will cover this in slides.

Martin: Allowing derivative works gives us control. copyright means that
people cant copy RFCs, but they cant always write specifications.

Risks:

  1. A fork is successful. Great! Happens rarely. Reason - neglect. Not
    always interoperabbility risk.
  2. Fork introduces market confusion. Seen attempts with MPLS and TLS.
    Tend not to be successful. Should Put safeguards in place anyway.
    IETF and technical restraints.
  3. The trust can just license - maybe if you know the right people.
    Does not address open participation concerns.
  4. We need people to stop being wrong in the internet. Good luck.

John Levine: Considerable sympathy for desire to move drafts , but as it
is wrtten this would be a disaster. This would be an open spec for bad
actors. Yes they can already, but not using our documents. Yes to draft
saying that IESG can give org authority to make derivatives.
This doesn't happen all that often.

Rich Salz: When this first came up I was opposed, now believe this is a
good thing. Protecting brand, intellectual property and open forks. Lots
of discussion over naming because of this. In favor of loosening grip.
Needs WG. Not able to get control back is an issue.

Ted Hardie: Sent mail which is in archives. Should not be dispatched.
The way this is written now, none of the pre existing RFCs would
inherit. Reread section 5. Would be difficult to retain other rights
associated with docs if we made retroactive changes. Licenses for IPR do
not travel with forks. Need to be able to say that x stream can take
control but not y. WG is about the right size to solve this. Happy to
contribute to WG.

Martin: Not my intent. Left out anything prior to RFC5378.

Glenn Deen: Chair of IETF trust: trustees follow guidance of IETF. As
written today the doc is structured so could not be turned into license
agreement. If we were to go forward it would need radical changes.
Personal opinion: conflating code limitations and the protocal specs and
protocal within. Open source software has a copyright. You can license
this code. Within RFC we hold copyright. 1. copyright of doc itself. 2.
copyright of code snippets inside doc. The copyright does not prohibit
anyone from implementing the protocol within the document. The controls
you are trying to do to enable things do not quite fit. If the intent is
for other groups to take our work and adopt it, that is a pathway here.
Recommend not doing a blanket thing. The name: The IETF does not
trademark protocol names. There is not protection mechanism for multiple
people to make "quic". Happy to help give guidance.

Martin: Consulted with lawyer, distinction between protocol and doc is
clear, if not that is mistake on my part, we need to fix it.

Cullen Jennings: This is long and hard to think about how to achieve
this. Go back to other SDOs. We should learn from this. If there is an
issue in a WG, ultimately people who decide is steering group. Right to
enable open source action. Not fully true when trustees say they only
listen to advice received from IETF. Needs more discussion.

Martin: RFC5378 has clause that says may consider more open licensing.

Cullen Jennings: Highly legally challengable. Granted ability to make
broad decisions. Main point is, can you say more about an use case where
this would be a good thing for the internet?

Martin: Every few years we publish NFS - do we want to keep doing that?
Is this the right place to do that?

Harald Alvestrand: Sympathetic to presentation. Wrong: this has not been
extensively discussed. This has definitely been extensively discussed.
Put forward in 2006. Consensus then was wrong. Need to consider
carefully what we do here. Cannot be AD sponsored thing that consumes
fewer cycles than IPR of 2006 did.

Richard Barnes: Current policy is absured. I am in violation of current
policy. It is over zealous and ineffective. Draft is in the right
direction. Value of IETF is not in documents. Focus of this draft should
not be on docs. Value is in consensus.

Jon Peterson: There are a lot of uncopyrightable ideas, point of trust
is to own a set of docs. Non WG forming BOF.

Glen Deen: You can site freely without restriction. You cannot take
snippets under the current license terms and stick them into other
bits.So in some cases, there have been special licenses done upon
request. But in general, the rule is you can take an RFC, reproduce it
in whole, but if you want to take little bits, that one we don't give
general licenses to.

Jon: Other than in fair use

Glenn: Yes

David Schinazi: Hypothetical: want to build the next version of quic,
dont want to do work in IETF. Current rules say I would spend more time
and it would not be as good. If we do this proposal, I would start from
quic RFC and deploy to internet, doc would be better. Current doc does
not prevent, just prevents from doing a good job.

Eliot Lear: Doomsday scenario is when work is taken and modified by an
organization that regulators pay attention to, and then the regulators
say, Do this version instead. Worse is when some regulators say do this
version. Another standards body picks it up and says, do this version.
And now you have three versions. You think that might not happen. Only
we see standards organizations and regulators pressing in that direction
right now in various different dimensions. This is a problem that needs
to be dealt with judiciously. Lets do something in an incremental
fashion. Where trust and IESG is more involved. Middle ground. Has
limitations. ISE has liberal policy, not implemented.

Dispatch outcome: More work is needed.

13:50 Reflecting on experiences in NOMCOM 2022 (20 min)

Presenter: Rich Salz, Geoff Huston (On site)

draft-rsalz-gih-nomcom-2years

Messages

Nomcom has two tasks: How to conduct business and perform selection
process.
Hard to set up mechanics. Language barriers, locations, etc.

Multiyear Nomcom membership terms

3 year chair role - see one, do one, teach one.

Strengths: greater level of continuity year to year allows for faster
startup. Allows members to gather experience and understanding. Allows
members to undertake role with greater confidence over time.

Weaknesses: Greater time commitment - may reduce volunteer pool. Greater
potential influence. May pose risk to confidentiality.

Questions: Should members be unable to volunteer after expiration of
term? Should casual apts be drawn from original volunteer pool?

Mallory Knodel: This is intention with value of randomness. This noncom
has no women, so now we have to wait another year. Randomness is there
because we need fresh and new. Having someone in the role for two rounds
reduces impartiality. Not necessarily bad idea, but introducing new
thoughts. Wants to continue diversity conversation.

Rich: Agreed.

Harald Alvestrand: When design protocols, always wise to consider what
happens if they are attacked. Nomcom has resistance to attack as it is
now. Randomness guarantees that you cannot pack the board. Even if you
succeed for one year, random nomcom next year means that you will not
succeed next year. Making two year positions weakens those protections.
A single bad selection can now replace entire IETF leadership. Perhaps
unrealistic, but it can be a worry.

Rich: not planning on changing the limitations. Example - no more than
two people from Akamai per term.

Leslie Daigle: Commitment of nomcom members is intense - takes over and
takes people out of work flow for a couple of years. Thank you for
presenting as problem with pros and cons with strong proposal. Worth
having a BOF to figure out what changes can be made.

Ted Hardie: This draft should be dispatched with no action. Not ready.
Not convinced that it is convincing for the right thing, agree with
Harald.

Jon Peterson: +1 to Ted.

David Schinazi: Discussed complication of nomcom process. Will add more
complication?

Tara Tarakiyee: Agree that it is already complicated. Crux of the
problem is that nomcom is actually two processes. Simplify the system by
breaking out the mode of operation to make a more permanent, contiguous
working group or similar Secretariat-type of position that would be
non-voting and have the voting members rotate.

Peter Koch: There is prior art. Dont understand problem, sees two
proposed solutions. Staggered model of chair is attractive. Risk: if you
have continuity on voting nomcom members and bootstrapping process then
you might want to see if continuty is a good idea, or you want fresh
people. Past Chair would actually review their own assignments based on
proposed model.

Robert Sparks: One of the features the current system has is a complete
shock replacement. You are identifying a problem having to bootsrap from
zero. Feature rather than bug. If you dont do this you have a new system
that develops a culture that self propogates. We might not want this out
of nomcom. Want radically different thought from last nomcom? If we go
forward, caution to think about periodicity of replacements.

Rich: Good points, thank you.

Suresh Krishnan: Dont like the 2-year thing because it kind of narrow
down the volunteer pool more. Less diverse is my biggest concern. I like
the incoming Chair shadow idea.

Samuel Weller: Had nomcoms with various degress of experience. Give them
the chance to appoint advisors. They can do this if they want guidance.
Don't think we need to move forward with this.

Geoff Huston: Part of nomcom last year - lots of time wasted at the
beginning learning what to do, appointments are rushed at the end. How
to provide framework for continuity. How to get both chair and members
to lend experience going forward so not everything is a complete
surprise. Still random selection of 10 people, just happens over two
years instead of one. Last year was not an anomoly, this happened years
ago when was a chair.

Rich: Agree. Foreshadow of likely outcome, please get in touch in
Brisbane.

Dispatch outcome: More work is needed for more clarity.

Summary of the Dispatch outcomes:

  1. Support in favor of a small working group.
  2. More work is needed.
  3. More work is needed for more clarity.