[codec] Minutes from IETF 75 BoF

Peter Saint-Andre <stpeter@stpeter.im> Fri, 04 September 2009 14:27 UTC

Return-Path: <stpeter@stpeter.im>
X-Original-To: codec@core3.amsl.com
Delivered-To: codec@core3.amsl.com
Received: from localhost (localhost [127.0.0.1]) by core3.amsl.com (Postfix) with ESMTP id 691B03A67F4 for <codec@core3.amsl.com>; Fri, 4 Sep 2009 07:27:33 -0700 (PDT)
X-Virus-Scanned: amavisd-new at amsl.com
X-Spam-Flag: NO
X-Spam-Score: -2.117
X-Spam-Level:
X-Spam-Status: No, score=-2.117 tagged_above=-999 required=5 tests=[AWL=-0.432, BAYES_00=-2.599, J_CHICKENPOX_33=0.6, SARE_MILLIONSOF=0.315]
Received: from mail.ietf.org ([64.170.98.32]) by localhost (core3.amsl.com [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id pkK-BYaT39-B for <codec@core3.amsl.com>; Fri, 4 Sep 2009 07:27:30 -0700 (PDT)
Received: from stpeter.im (stpeter.im [207.210.219.233]) by core3.amsl.com (Postfix) with ESMTP id 528363A679F for <codec@ietf.org>; Fri, 4 Sep 2009 07:27:28 -0700 (PDT)
Received: from leavealone.cisco.com (72-163-0-129.cisco.com [72.163.0.129]) (Authenticated sender: stpeter) by stpeter.im (Postfix) with ESMTPSA id 750374007B for <codec@ietf.org>; Fri, 4 Sep 2009 08:18:43 -0600 (MDT)
Message-ID: <4AA121C2.8020508@stpeter.im>
Date: Fri, 04 Sep 2009 08:18:42 -0600
From: Peter Saint-Andre <stpeter@stpeter.im>
User-Agent: Thunderbird 2.0.0.23 (Macintosh/20090812)
MIME-Version: 1.0
To: "codec@ietf.org" <codec@ietf.org>
X-Enigmail-Version: 0.96.0
OpenPGP: url=http://www.saint-andre.com/me/stpeter.asc
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
Subject: [codec] Minutes from IETF 75 BoF
X-BeenThere: codec@ietf.org
X-Mailman-Version: 2.1.9
Precedence: list
List-Id: "Should the IETF standardize wideband Internet codec\(s\)? " <codec.ietf.org>
List-Unsubscribe: <https://www.ietf.org/mailman/listinfo/codec>, <mailto:codec-request@ietf.org?subject=unsubscribe>
List-Archive: <http://www.ietf.org/mail-archive/web/codec>
List-Post: <mailto:codec@ietf.org>
List-Help: <mailto:codec-request@ietf.org?subject=help>
List-Subscribe: <https://www.ietf.org/mailman/listinfo/codec>, <mailto:codec-request@ietf.org?subject=subscribe>
X-List-Received-Date: Fri, 04 Sep 2009 14:27:33 -0000

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

========================================================================

Minutes for Internet Wideband Audio Codec BOF, IETF 75
Kongresshall C, City Conference Centre, Stockholm, Sweden
Thursday, July 30, 2009, 13:00-15:00 local time, 11:00-13:00 UTC

CHAIRS: Jason Fischl, Jean-Marc Valin

NOTE TAKERS: Olle E. Johansson and Bruce Lowekamp
JABBER SCRIBES: Alan Hawrylyshen and Peter Saint-Andre

MINUTES EDITOR: Peter Saint-Andre

AGENDA:

<http://www.ietf.org/proceedings/75/agenda/codec.txt>

1. Administrivia (5 minutes)
2. Introduction and Scoping of the BoF (10 minutes)
3. Goals of the Working Group (15 minutes)
4. Engineering Work (25 minutes)
5. Codec Expertise in the IETF (15 minutes)
6. Gauging Likelihood of Success (15 minutes)
7. Call for Consensus on Whether to Form a WG (15 minutes)
8. Charter Discussion (15 minutes)
9. Conclusions and Next Steps (5 minutes)

AUDIO: <http://www.ietf.org/audio/ietf75/ietf75-thur-kongressc-pm.mp3>

JABBER LOG: <http://jabber.ietf.org/logs/codec/2009-07-30.txt>

EDITOR'S NOTE: This was a very fast-pased BoF with a great many
speakers at the mic and much back-and-forth conversation. Although
the note-takers have attempted to capture all of the discussion, we
cannot guarantee that the minutes are truly complete. Please refer
to the audio stream and the Jabber logs for details. Approximate
minute markings have been provided for reference, with time in UTC
to synchronize with the Jabber logs.

========================================================================

1. ADMINISTRIVIA (11:06)

1a. Note well slide presented.

1b. Remote logistics, Jabber room, meeting materials, audio, etc.

1c. Volunteers noted for note takers and Jabber scribes.

1d. Agenda bashing.

Chairs note one modification: order of items 7 and 8 switched.

Roni Even @ mic: The agenda topic "Engineering Work" does not explain
what exactly will be discussed. Codecs or requirements?

Jason Fischl (Chair): The topic will be the *kind* of work that we
would be doing. We will look at some attributes that we would use as a
way of evaluating codecs, then show how the 5 codecs that have been
submitted would address those attributes. We will have short
presentations (5 minutes) about individual codecs that have been
submitted to indicate the kinds of work that might be in scope, but
these codecs are merely representative.

========================================================================

2. INTRODUCTION AND SCOPING OF THE BOF (11:11)

2a. Cullen Jennings (AD) presenting

* SLIDE 1: Things for the IESG/IAB to Learn
- - Is there a well-defined goal?
- - Do standardized solutions exist?
- - Is it technically feasible to do this work in a reasonable time?
- - Are there people that can do this work in the IETF?
- - If the work is completed, would it be used?

* SLIDE 2: Patent Related IPR at the IETF
- - If formed, working group would function within IETF IPR policy.
- - Not looking to change IETF IPR policy.

* SLIDE 3: BCP 79
- - [quotation from BCP 79]

* SLIDE 4: Goal is Royalty Free
- - from BCP 79: "In general, IETF working groups prefer technologies
with no known IPR claims..."

* SLIDE 5: No Mandate of Royalty Free
- - from BCP 79: "... working groups have the discretion to adopt
technology ... with no licensing commitment, if they feel that
this technology is superior..."

Xavier Marjou @ mic: Question about overlap of standardization bodies,
referring to RFC 2418 (BCP 25).

Cullen Jennings @ mic: Yes, it overlaps and we have a procedure for
handling this with other liaison SDOs. Generally the IAB focuses on
this. We need to make sure that our processes are consistent with that
BCP, and we're doing that. We have already received two liason requests
from two relevant organizations. New work is discussed on a mailing
list read by other SDOs and issues are addressed between SDOs and IAB
before deciding to form a working group.

Patrik Fältström @ mic (as liaison to the ITU-T): I am the liason to
the ITU-T and I am working together with the ITU-T on this.

========================================================================

3. GOALS OF THE WORKING GROUP (11:18)

3a. Chairs on the goals

Jean-Marc Valin presenting...

* SLIDE 1: Today's Codec Landscape

Jean-Marc (presenting): A representative list of some relevant codecs
received from the ITU-T liaison.

Henry Sinnreich @ mic: Why is the SILK codec not in the list?

Jason Fischl (Chair): The list is not comprehensive.

[unidentified] @ mic: There might be others that are royalty-free.

Jean-Marc Valin presenting...

* SLIDE 2: Applications

Ted Hardie @ mic: How fixed is this list? Is this application list
comprehensive or might there be others? AppArea might have a different
set of requirements.

Slava Borilin @ mic: Just an initial list to given an idea of the scope.
Defining a list of target applications would be part of the working
group's task, if it is approved.

Roni Even @ mic: The charter says only doing "interactive" application,
but not "streaming" applications (which might be similar).

Ted Hardie @ mic: Is this list constraining or not?

Jason Fischl (Chair): This list is intended to be a non-constraining
list of motivating examples.

Roni Even @ mic: I understood from the charter that there are some
limitations.

Jason Fischl (Chair): There is a slide later that defines what is out of
scope.

Joe Hildebrand @ mic: There are a number of AppArea people here as well
and folks should take requirements from applications into account.

Jason Fischl presenting...

* SLIDE 3: Goals
- - Widely and easily distributable and accessible standard codecs
- - Small number of codecs (1 or 2) that cover the operating space

Jason Fischl (presenting): Not proposing to start a "codec factory".

Stefan Bruhn @ mic: What is so unique with these goals for the IETF?
There are other SDOs who do work in this area. Why are we reformulating
these goals here in the IETF?

Harald Alvestrand @ mic: Is the operating space meant to be interactive
voice as opposed to stored voice?

Jean-Marc Valin (Chair): The scope is meant to be interactive audio,
not just voice.

Monty Montgomery @ mic: Within the IETF we would be solving problems
that IETFers have. Also gives us an opportunity for cross area review.
Feedback from application developers, transport protocol designers, etc.

Stefan Bruhn @ mic: I don't understand the motivation. We have the MPEG
(system agnostic), ITU-T (fixed line), 3GPP (wireless transports), so
we have the whole range, I don't see why we need to address this in the
IETF. What is so specific here to the Internet that it can't be
addressed in MPEG, ITU-T, or 3GPP?

Peter Saint-Andre @ mic: The IETF has an open participation model in
which the entire Internet community can participate, many other SDOs
and industry consortia do not have that model.

Monty Montgomery @ mic: The development model in the IETF culturally
matches the way we work in the developer community. BCP 79 avoids an
accidental, systemic bias against producing royalty-free codecs. Here
we have a fighting chance of doing something truly open.

Joe Hildebrand @ mic: It's a valid point that applications considered
at the IETF have requirements that applications considered by other
SDOs don't have.

Colin Perkins @ mic: I was chair of AVT WG. You need to pay attention to
the way the Internet works. I don't know if that is an argument about
whether to form a codec working group, but it is an argument that those
who design codecs need to interact with those who understand Internet
protocols. One could potentially make the argument that those who work
on codecs should interact heavily with the IETF. Maybe that doesn't
happen in other SDOs.

Christian Hoehne @ mic: Need to consider other requirements than have
traditionally been considered by other SDOs.

Colin Perkins @ mic: I'm not saying that other SDOs have not been
involved in that kind of interaction with the IETF, but that it doesn't
always happen.

Roni Even @ mic: IETF should produce requirements for submission to
other SDOs, which have the expertise to do the codec work. As to open
participation, anyone is welcome at ITU-T rapporteur meetings who is
approved by rapporteur (but they cannot vote).

Xavier Marjou @ mic: I confirm what Roni just said. I also add that
there is a liaison relationship between ITU-T and IETF, and the ITU-T
is open to taking on new work.

Anisse Taleb @ mic: The ITU-T or even MPEG are not less open than the
Internet community. Individuals are invited to submit requirements, and
MPEG calls for proposals are open to companies that are not members of
MPEG. I agree as well with Stefan Bruhn -- I don't see what is different
here in terms of requirements.

Joe Hildebrand @ mic: Definition of open used at IETF does not overlap
with needing to be invited to participate.

Roni Even @ mic: You need to pay to attend an IETF meeting.

Joe Hildebrand @ mic: These meetings are not canonical, it is the open
mailing lists that are canonical.

Roni Even @ mic: It is the same in the ITU-T.

Patrik Fältström @ mic (as ITU-T liaison): I'm a bit concerned about the
language being used at the mic. We have agreement between ITU-T and IETF
on how to communicate and how to cooperate regarding the possibility of
doing overlapping work. We are following that process in the IETF and
the ITU-T is following its process. It is not part of the process to
have the kind of discussion we are having in the room, so please stop
and let us move on to more productive topics.

Jean-Marc Valin presenting...

* SLIDE 4: Technical considerations
- - suitability for most fixed-point digital signal processors
- - medium- to high-quality speech at moderate bit-rates
- - high-quality music encoding at higher bit-rates
- - sampling rate profiles from narrow-band to full audio bandwidth
- - real-time adaptive bit-rate
- - source-controlled variable bit-rate (VBR)
- - low algorithmic delay

Christian Hoehne @ mic: What's the background behind the choice of
the chosen technical considerations? This looks like a description
of CELT and SILK.

Jean-Marc Valin (presenting): These considerations were originally
posted in the first proposed charter and adapted based on feedback.
They are by no means final and are open to modification.

Roni Even @ mic: This is a typical list for codec requirements, nothing
specific here to the Internet.

Jason Fischl (Chair): These are technical considerations, there are
other Internet-related considerations.

Ted Hardie @ mic: Is the goal to "pick a winner" (just one or two). Is
there a desire to avoid transcoding? We might want that for technical
reasons, such as avoiding delay or preventing fragility. But let's be
careful that the way to avoid transcoding is not to pick a winner
because then we have a collusion issue. Must be very open about what our
needs are to avoid the appearance of collusion. If this is not the
issue, they do we really need to avoid the codec factory approach?

Colin Perkins @ mic: I agree with Roni Even -- there is nothing here
that is particular for the Internet. Surely there must be technical
factors from the Internet transport that are relevant here.

Francois Audet @ mic: I have a similar point. How does this relate to
SIP or SDP? Issues like end-to-end negotiation of SDP parameters are
important. In-band or out-of-band? Not good to take the most general
approach. What are the IETF specifics? Identify what things IETF is
best at and include those in the scope.

Monty Montgomery @ mic: Transcoding points are excellent. It would make
the system better to avoid transcoding. Second point, those technical
considerations are items that are interesting to those of us who want to
do the work, and this work is not being done elsewhere.

Christian Hoehne @ mic: We haven't discussed transcoding because of the
end-to-end principle for IP packets. If you have e2e instead of some
other network principle (e.g., circuit-switched), then there's no need
for transcoding.

Mikael Abrahamsson @ mic: I'm a network engineer. I spend a lot of time
debugging VoIP traffic. Jitter, latency, packet reordering, packet
drops are Internet issues that are not commonly handled by codecs and
that cause codecs to lack robustness. I'm baffled that this is missing
from an IETF codec initiative.

Jean-Marc (presenting): This is not a complete list.

Julien Meuric @ mic: There are existing implementations that meet all
these requirements. These considerations are not IETF-specific. Why not
use or extend existing standards-based solutions?

Jason Fischl (Chair): Could you be more specific?

Julien Meuric @ mic: Transcoding not mentioned. There are standardized
tools that make this possible. What's the new issue that needs to be
solved?

Slava Borilin @ mic: Existing codecs don't meet these requirements,
either not easily distributable or not high quality. I agree that the
Internet-specific requirements need to be included. Those of us who
worked on the slides probably assumed that the Internet-specific issues
were so obvious that unfortunately we omitted from the slides. New work
will potentially be an extension of something existing.

Anisse Taleb @ mic: These are kind of default considerations for codec
work anywhere. What particular features would an Internet codec have?
Be more specific about what requirements existing codecs do not meet.
Need evidence. Robustness to frame losses is an important technical
consideration, for example.

Jean-Marc Valin (presenting): The requirement for robustness to frame
losses is in the charter but we cut it off the list to fit on the
slides.

Anisse Taleb @ mic: Best way to avoid transcoding is to stop
standardizing codecs.

Keith Drage @ mic: You don't have a slide about non-goals.

Jason Fischl (Chair): We do, it's next.

Slava Borilin @ mic: Two requirements, easily distributable and good
performance. Those are not met now.

Jason Fischl (Chair): Let's move on to the next couple of slides, then
have more comments.

Jean-Marc Valin presenting...

* SLIDE 5: What is out of scope?
- - Very low bitrate
- - Non-interactive music (e.g., MP3, AAC, Vorbis)
- - Unlimited number of codecs

Jason Fischl presenting...

* SLIDE 6: Why do this here?
- - IETF is where you find the people who understand the impact of the
  IP stack on codec design
- - Cross-area review
- - Whole Internet Community can participate
- - IETF will have change control and the resulting codecs will
  be different and better than the initial candidates

Stefan Bruhn @ mic: Are you saying that in other SDOs there are no
people who understanding the IP stack?

Jason Fischl (presenting): Not saying anything about other SDOs, the
only statement is that we have such people at the IETF.

Stefan Bruhn @ mic: I don't see what is so unique in the IETF. We have
all these things (cross-area review, whole Internet community can
participate) in other SDOs.

Keith Drage @ mic: I want to make sure there is a non-goal that we're
not trying to define the default codec for the Internet, and that the
resulting codec is not going to be mandated.

Roni Even @ mic: I don't see codec expertise on this slide. Also, IETF
doesn't mandate codec, this is done by other forums that create
profiles. Free codecs are available. Why do you need a standard codec?

Alan Duric @ mic: As an operator, I've been struggling for 6 years to
deploy VoIP and none of the vendors want to deploy Speex because it's
not a standard. If we don't have this work done as a standard, operators
will not be able to convince vendors to deploy it. G.722 is falling
apart on IP.

Xavier Marjou @ mic: We have already deployed a wideband audio codec
over IP. I think IP considerations are important, but other areas come
into account in codec design.

Alan Duric @ mic: Give me the evidence regarding G.722. Please post to
mailing list what these codecs are that meet these requirements.

Adam Roach @ mic (quoting from XMPP BoF minutes held at IETF 54):
"We should be chartering work based on the number of folks who want to
do things, not the number opposed." I think we've heard from a number of
people here who want to do the work and have competence in this area.

[unidentified] @ mic: [Comment about G.719?]

Patrik Fältström @ mic (as liaison to the ITU-T): Regarding the comment
that we should start work if there are enough people willing to do the
work: there are two different discussions here. One is the IETF/IESG
decision about whether to create a working group. There is a separate
discussion between the IETF and the ITU-T according to RFC 3356, the
agreed-upon procedure between IETF and ITU-T to determine if there is
overlapping work. That discussion is happening and will continue after
this BoF.

Ingemar Johansson @ mic: There is no problem with interaction between
SDOs. Let's define payload formats here and leave codec work to the
ITU-T.

Xavier Marjou @ mic: Why are people in the IETF saying that standards
coming from other SDOs a joke? Does putting "IP stack" in the slide make
the IETF qualified?

Monty Montgomery @ mic: Regarding why to do this within the IETF: there
are people who want to do this, are doing this, and who want to do it
through the IETF. That's not true elsewhere.

Christan Hoene @ mic: Things have to work together. This is a reason to
do the work here. Are there enough Internet, codec, and VoIP experts
sitting together in the same room? That is the question, and we need to
decide whether that can be done in the IETF, between the IETF and ITU-T,
or in some other way. I think it can be done in the IETF, but no matter
what we need to start this work.

Slava Borilin @ mic: Regarding why IETF and why standardized codec:
Because IETF management of change control is important for vendors.
Once change control moves from corporation / vendor to the IETF it
becomes safer for others to implement and deploy. The IETF has a group
of people that are willing to contribute, skilled and able to
contribute. We have expertise here. The codec has to be developed with
all aspects of the transport and IP stack considered for high quality on
the Internet. Existing codecs do not address that factor, and the few
codecs that are close to meeting those goals can't be freely distributed.
We have goals of freely distributable and high quality in the real
Internet. The IETF is the best organization that fits meeting both of
those goals.

Jason Fischl @ mic (as individual): How to we get wideband codecs
broadly adopted? G.722 is brought out as an example and is starting to
become actively deployed. The trigger is expiration of the patents. We
haven't seen widely deployed codecs before. Widely distributed and
freely available codecs lead to broad deployment. The IETF can try to
achieve that, although there is no guarantee.

Wilhelm Wimmtreuter @ mic: We have people willing and able to contribute.
It matters who can fund it and support it. Why not let them do it?

Stefan Bruhn @ mic: I doubt that you have the expertise in this group.
Not traditionally at the IETF. Why are you not proposing this to the
ITU-T, even royalty-free, and face competition?

Slava Borilin @ mic: Because most of our stakeholders are here in the
IETF.

Jean-Marc Valin presenting some slides about codecs that have been
submitted as Internet-Drafts so far...

* SLIDE 7: CELT Codec Characteristics

* SLIDE 8: SILK Codec Characteristics

* SLIDE 9: IPMR Codec Characteristics

* SLIDE 10: BV32 Codec Characteristics

* SLIDE 11: A2DP SBC+PLC

Dave Oran @ mic: Loss robustness is not a boolean. And it is not a
scalar. Loss is a curve, not a scalar value.

Colin Perkins @ mic: How is it relevant to the Internet that we have a
list of codecs? How do these codecs solve Internet-specific problems?

Roni Even @ mic: What is the point of a list of codecs?

Jason Fischl (Chair): Trying to point out that there were a number of
submissions of real codecs.

Roni Even @ mic: That's interesting, but there are no requirements so
everyone is qualified.

Peter Saint-Andre @ mic (as individual, not Jabber scribe): People say
there is no competence in the IETF. The point of the list of codecs is
saying that we do have competence among the people who have developed
these codecs and contributed them to the Internet Standards Process.

Xavier Marjou @ mic: What is really new in terms of new requirements?

Jason Fischl (Chair): These have been submitted in the open and intended
to be open source.

Xavier Marjou @ mic: For a new codec there are likely patents coming.

Colin Perkins @ mic: I question the claim that they are royalty free.

Jason Fischl (Chair): Claim is not that they are royalty-free, but that
they are being provided without royalties on any of the patents. [?]

Slava Borilin @ mic: People are contributing codecs that are believed to
the best of their knowledge to be royalty free, and these people are
contributing / have expertise to contribute royalty free codecs that are
superior to existing alternatives. Based on what we know, we stand a
good chance of doing better than what is out there, and keeping it
freely available. We believe that this can be achieved such that quality
is as good or better than existing codecs, and to do so in a way that is
freely and easily distributable.

Alan Duric @ mic: As to competence, there are people in this room who
have developed codecs that run into the hundreds of millions of minutes
each day or each week. These codecs are superior to the existing codecs
out there and none of those are suitable for Internet or wireless LAN
if there is any kind of packet loss.

Peter Saint-Andre @ mic (as individual, not Jabber scribe): Could we
please try to get through the agenda?

========================================================================

4. ENGINEERING WORK (12:22)

Jason Fischl (Chair): Quick presentations on SILK and CELP...

4a. Koen Vos presenting about the SILK codec (12:22)

(draft-vos-silk-00)

4a. Monty Montgomery presenting about the CELT codec (12:24)

(draft-valin-celt-codec-00)

"If CELT comes out of a proposed working group intact, I will be sorely
disappointed."

Emphasizes the point that all of these codec developers want to merge
their work to produce a best-of-breed codec.

========================================================================

5. CODEC EXPERTISE IN THE IETF (12:27)

Jason Fischl (Chair): These brief presentations were trying to establish
that we do have codec expertise among IETF participants, but there are
also other kinds of contributions -- requirements, testing, etc.

Joel Halpern @ mic: I'm not going to question whether there are enough
people here to do the codec work, because that's clearly true. The
question is whether the *IETF* has the expertise. It's only sensible
for the IETF to do it if it works with the rest of the process. Can the
reviewers provide appropriate technical insights? Does the IESG have
expertise to provide oversight? If we don't have expertise in those
places, we can't do this as an organization.

Henny Sinnreich @ mic: I'm willing to contribute with usage scenarios
for Rich Internet Applications and the resulting requirements.

Slava Borilin @ mic: We definitely have enough people to work on the
codecs themselves, but do have enough qualified customers? I think so.

Colin Perkins @ mic: I do not believe the wider IETF can meaningfully
review the work (IESG and other areas).

Bernard Aboba: We have chartered IETF working groups in this kind of
situation before. For example, TRILL. We only looked at whether we had
people who could do the work. What we put into the charter is review by
appropriate outside experts and liaisons (in that case IEEE) to get
comments. We have been successful in that approach. The only thing
that was relevant is whether people are here with expertise and willing
to do work.

Roni Even @ mic: It's more complicated. You need to provide information
correctly to liaisons and outside experts. For example, need to work on
characterization.

Christan Hoene @ mic: There need to be experts on judging the quality
of the codecs. For example, there are standards for conducting listening
tests.

Jason Fischl (Chair): I want to make sure we are very focused on the
question of whether we have the expertise to do this work in the IETF.

Jean-Francois Mule @ mic: I do believe there is expertise. Anyone who
has done codec development and testing knows how to define test vectors
so that implementers can validate implementations. In this room we have
4-5 companies with codec test bench labs that could validate that the
codec has been implemented correctly according to the spec and also that
you have met your terms of reference and requirements. Also independent
labs. There are plenty of resources, and this is not rocket science.

Slava Borilin @ mic: At Spirit DSP have expertise in design and we are
engaged on the Internet application side. A working group can serve
most of all as an analyst to specify the right tasks for developing
the kind of codec that is needed for Internet applications. The major
value is identifying the technical and business requirements and
developing technical solutions to meet those requirements. The working
group might even end up with a decision that G.722 is the best choice.

Monty Montgomery @ mic: Two hours ago the argument was that we didn't
have the competence. That has shifted to "you are competent to design a
codec but you don't have the competence to test". In my experience,
testers outnumber implementers 10 to 1. This is irrelevant.

Adam Roach @ mic for Eric Rescorla in the Jabber chatroom: I'm not
concerned about Joel's point. We have similar problems in security,
where the ability to evaluate cryptographic techniques outside the
working group is extremely limited. But at least the performance of
codecs is measurable, whereas with security even that isn't true.

Adam Roach @ mic: At least codecs can be evaluated directly by everyone
in the IETF, since we all have working ears.

Stefan Bruhn @ mic: Do we have expertise? Depends on the companies
sending their experts to this group. Experts are sent to ITU-T and MPEG
and 3GPP, are you sure they will be sent here? For experts here, why
aren't you at MPEP and ITU-T and 3GPP? Are you afraid of facing
competition?

Derek McDonald @ mic: It seems that we do have the expertise. But the
IETF retains the right to fail. It isn't like we've lost a war if the
working group fails. Don't let FUD block us.

Mikael Abrahamsson @ mic: The Internet should demand things from the
codec rather than the codec demanding things from Internet. We know IP,
we know how the Internet works.

Alan Duric @ mic: I fully agree, and if we brought this to ITU-T or
ETSI, yes these proposals were brought forward 8 years ago and shot
down without consideration.

Julien Meuric @ mic: People are interested, but do you have any proof
that the technical skill is present here?

Roni Even @ mic: We are trying to define the whole process, need
expertise in defining the whole process. But really need a beauty
contest. Either approve all the specs that are submitted or you need
to select a few and define how to do that.

Xavier Marjou @ mic: IETF documents are good because they are so
readable, but a codec RFC would be unreadable. Can't write formulas in
an Internet-Draft.

Jean-Francois Mule @ mic: Responding to request for evidence, Broadcom
is willing to bring expertise and competence. Dr. Chen and his group
have worked for many years in ITU-T.

Slava Borilin @ mic: Spirit DSP has developed more than 10 proprietary
codecs standardized under various bodies like radio standards. We are
not going to ITU-T because there are no stakeholders interested in the
result (freely distributable, high quality). In IETF there are.

Monty Montgomery @ mic: First, we will mix inputs from different
contributors. Second, Roni is right that a lot of characterization,
testing, and documentation needs to be done. We're aware of that and
we have a lot of work to do.

Ye-Kui Wang @ mic: Who in IESG is capable of reviewing audio codecs?

Ted Hardie @ mic: Please stop talking about stakeholders -- not a useful
construct in the IETF. The decision is whether to charter work that
would benefit the Internet. Serious question is whether this is the best
place and whether this work benefits the Internet. Stakeholders are
users, operators, vendors, etc. Please try to take a whole Internet view
rather than a particular stakeholder view or we will ask ourselves the
wrong question.

Cullen Jennings @ mic, as AD who will be asked to evaluate codecs: Do
we have competence on this topic? The IESG relies a lot on expert
advisers across areas. We try to assess whether things cause harm. We
have to listen to other people and try to evaluate whether claims of
brokenness are legitimate or not. Often our expertise lies in finding
the right people to provide input and in listening to them in order to
judge which input is reliable. For example, will have to review IDNAbis
(internationalized domain names) but there are no linguistics experts
on the IESG, so we will have to rely on other people and work with them
and try to integrate their feedback. Another factor is that IESG members
are selected based on skills needed to do the work and review it. This
codec work is really not different from other work and the important
thing is that there's expertise to get the right comments so the IESG
can make good decisions.

[12:50, T-10 minutes]

QUESTION from Chairs: Who here would call themselves codec experts and
would be willing to develop codecs?

Outcome: 10-12 real hands and 5+ virtual hands from the Jabber room.

QUESTION from Chairs: Who is willing to participate in any way listed on
slide 1 (guidelines, process, requirements, codec proposals, commenting,
evaluation)?

Outcome: 40-60 real hands and virtual hands.

Lars Eggert @ mic: Of the codec experts, would you still participate
if a proposal you contributed is ruled out by the working group?

Outcome: Same as for the first question.

HUM REQUEST from Chairs: Is or is not the technical scope of the work
well defined?

[one hum for yes, one hum for no]

Outcome: Inconclusive.

HUM REQUEST from Chairs: Do we or do we not have enough expertise to
form a working group that can meet the goals?

[one hum for yes, one hum for no]

Outcome: Inconclusive.

Ted Hardie @ mic: Large grey area here, what if you have no idea if
there's enough expertise to complete the work because you don't know if
it's well-scoped in the first place? Perhaps a separate hum is in order.

Cullen Jennings @ mic: Could you type the question in a slide? My
experience from other BoFs is that the question can get very confusing.

[12:58 conclave at front of room among Chairs, AD, BoF Sponsor, etc.]

HUM REQUEST from Chairs: Do we or do we not have enough expertise to
properly scope the work?

[one hum for yes, one hum for no]

Outcome: Rough consensus in favor.

[conclave continues]

Cullen Jennings @ mic as Area Director:

- - I think we've identified that it's prematuve to ask whether to form a
  working group
- - We have the information we need from the BoF and will go forward with
  discussions on the mailing list

[meeting adjourned at 13:01]

========================================================================
6. Gauging Likelihood of Success

- - not discussed -

========================================================================
7. Call for Consensus on Whether to Form a WG

- - not discussed -

========================================================================
8. Charter Discussion

- - not discussed -

========================================================================
9. Conclusions and Next Steps

- - not discussed -

END



-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.8 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/

iEYEARECAAYFAkqhIcIACgkQNL8k5A2w/vwxjgCfTyywrg5qUVPmgI1pn0QpB26Q
X50AnR3WlAadkA6A6yfpAmGBM2vlUFp/
=Sebj
-----END PGP SIGNATURE-----