Re: Special Topic Call, DID WG

On 10/5/20 9:51 AM, Brent Zundel wrote:
> We are holding an additional call on *Thursday, October 8, 2020 at 12 PM
> ET.*

Oof, regrets.

I just realized that a number of us have a mandatory DHS SVIP call at
the same exact time, which is really unfortunate. All of the Editors
save for one won't be able to make the meeting.

I expect none of us caught this because the event wasn't in our calendar
at the regular time because of the shift due to W3C TPAC last week. :(

In an attempt to try to summarize what we discussed on the Editors call
and to provide some structure for the meeting:

* It is now clear that the last real Face-to-Face meeting, where we
  decided to create the DID Spec Registries, led to a number of
  miscommunications on the purpose of the registry.

* One of these miscommunications was a presumed goal by some that
  the registry would be used in the lossless conversion of
  representations to other representations. Namely, that it is
  possible to re-create @context from the registry. We have new
  information, now that the registry has been created, that this
  is mathematically impossible under certain conditions, which
  means that the registry is not capable of doing what some of
  us had originally intended it to do.

* There is also a miscommunication over what a "property" is in the
  data model. Namely, whether or not it includes "processing
  directives". Some think Abstract Data Model properties are
  solely about the DID Subject. Others think that Abstract Data Model
  properties are anything that is expressed in the DID Document (both
  registered and unknown/unregistered properties).

Thus the discussion today should probably revolve around these items:

1. Come to consensus on the revised purpose of the registry now that it
   can be proven that it can't do what some in the group wanted it to do
   (e.g., it is mathematically impossible to use it to construct certain
   properties like @context).

2. Come to consensus on whether properties are solely about the DID
   Subject, or if they can be about other things (e.g., the proof
   property).

3. Come to consensus on whether preserve-by-default applies to all
   properties in the abstract data model.

4. Come to consensus on whether implementers are allowed to "clean up"
   the abstract data model before an application uses it to "perform
   further processing higher up the stack".

Here are some proposals to help get feedback from the group:

PROPOSAL: The DID Spec Registries MUST contain a section on
Representations to enable future representations to be registered in an
extensible manner. The DID Core specification MUST specify how this
extensibility mechanism works as well as the requirements on
representation specifications.

PROPOSAL: The DID Spec Registries MAY be used by DID Methods and
Representations to inject or modify properties in the Abstract Data
Model, even though this might not always be possible (e.g., it is not
safe to generate and inject @context).

PROPOSAL: All properties in the Abstract Data Model MUST be preserved if
the property name and property value are types that the Abstract Data
Model supports.

PROPOSAL: The Abstract Data Model MAY be modified by software
implementations after the consumption process to sanitize or otherwise
protect implementations from properties that could lead to privacy or
security failures.

If we can get a +1 to some variation of the proposals above, we would
have a concrete solution that could be applied to the specification.

-- manu

-- 
Manu Sporny - https://www.linkedin.com/in/manusporny/
Founder/CEO - Digital Bazaar, Inc.
blog: Veres One Decentralized Identifier Blockchain Launches
https://tinyurl.com/veres-one-launches

Received on Thursday, 29 October 2020 12:37:22 UTC