Wednesday, 29 March, Rooms G412-G413
Chairs: Spencer Dawkins and
Brian Rosen
Notes: Stuart Card
Area: ART
Responsible AD: Murray Kucherawy
BOF proponents: Dan Petrie and Thomas
Howe
The ritual capture of Notetakers:
Stu Card foolishly volunteered (Stu's words), for which Spencer is
extremely grateful (Spencer's words)!
The gentler recruiting of someone to watch Meetecho/Zulip chat for
questions:
Thomas McC-H
began by showing the tension between using AI/ML etc. and protecting
privacy, esp. across security boundaries. He suggested that digital
formats, esp. if standardized, could facilitate the use of software
tools for (hopefully objective) assessment of the nature and extent,
in conversations, of content elements that might require protection
(or deletion, whatever). He asserted that integrating the different
elements (of different data types) comprising conversations
facilitates both privacy protection and data analysis. He mentioned
the usefulness for staff (e.g. customer service) training. He
briefly alluded to the major constituents of a vCon representation.
He mentioned proposed charter) that benefit from cross-device and
cross-application capture and forwarding.
In chat, Hans-Jorg H.
wondered what the current gap in state of the art. how are things
done right now and what does this imply for this work
In chat, Pete R.
expressed concern that cross-device, cross-app, etc. sharing not
expose everything to intermediaries.
Dan P.
introduced the concept of "conversational data" and reiterated that
it involves not only the actual data of multiple types but also
meta-data for multiple purposes. He stated the proposed charter
goals of a vCon WG. He invited good concise summaries of other use
cases that might or might not be inserted into the charter. He
expounded on some but not all of the questions a WG might address,
especially media types & corresponding container types.
Mallory K.
asked why transport & storage are excluded from proposed scope as it
appears that some of the proposed principal objectives cannot be
addressed w/o getting into those aspects (esp. end-to-end encryption
thereof). She suggested that "data minimization" should be made
explicit.
Thomas
explained that vCon was focused more on knowing what you have
captured so you can figure out what you must protect, not on how you
might then protect it.
Jonathan R. (5-9)
suggested that single vendor cases don't need IETF standardization
but multi vendor cases (primarily involving data exchange) might
justify it, esp. to supporting logging and audit. He distinguished
between recording who transferred what to whom and protecting
(typically encrypting) the data in motion. He inquired about a use
case involving transfer of conversational data between an
organization that captured it and another organization that analyzes
it for training or workflow management purposes;
Spencer & Jonathan
appeared to converge on the need for use cases in the charter that
are more specific to enable clearer demarcation of the scope edges.
Cullen J.
spoke to (cryptographically protected) differentiated access and how
to express both requirements for it.
Spencer
thanked everyone who had asked questions, made comments or suggested
use cases and entreated everyone to follow up with written specific
inputs.
Sebastian B. (NYU Law)
asked about "conversations" as such (typically involving social
aspects) and mere collections of communications? He distinguished
between "privacy" and "data protection" and highlighted the aspect
of consent. He raised the issue of provenance, not only of complete
recordings but also of excerpts.
Joris B.
pointed us to JS-Contact (?).
Hans-Jorg H.
asked how is this done now (w/o vCon)?
Chris W.
suggested call center operators may want a narrower scope of
identity vs broader privacy.
Jaime J.
inquired if it is in scope to specify API components (rt,
media_type, tokens...) if any?
Jonathan R.
said the state of practice is [s]ftp transfer of WAV files w/names
based on timestamps, and too often, "[s]ftp" is "ftp".
Hans-Jorg H.
suggested that clarity could be improved by documenting the current
state of practice and the gaps therein that vCon proposes to
address.
Poll: does the problem need solving? 35Y+0N=35
Poll: is the IETF the right place to solve it? 24Y+6N=30
Thomas
replied that he has heard informally from other organizations
that they are interested but he is not currently formally proposing
it to any other SDOs.
Poll: is the proposed charter (as amended) headed in the right
direction? 16Y+6N=22
Poll: are the proposed deliverables (as amended) close enough to be
useful? 15Y+6N=21
Stu Card
belatedly expressed support for this work for a use case of air
traffic controllers, pilots onboard crewed aircraft, remote pilots
of uncrewed aircrart, etc., esp. for incidents necessitating
subsequent inquests.
Poll: please indicate if you are willing to review vCon documents?
9Y+11N=20
Poll: please indicate if you are willing to contribute to vCon
documents? 9Y
If you are willing to help lead the vCon WG
please email vcon-chairs@ietf.org
Onsite Meetecho tool (Blue sheet registration, Queuing for question and
answer session, polling) is here.
Full Meetecho tool is here.
Session materials are here.
Hedgedoc notepad for this session is here.
Zulip chat for this session is here (and also in Meetecho)
See in datatracker schedule here.