RE: Identity Hubs and Agents

Daniel, thank you (and to the collaborators) for putting this document together and for sharing it.  Some questions:


  1.  Can you clarify or add a working definition for “identity”? …esp. in relation of the term “overall identity” and “subset of overall identity” referred to in bullet 7.

  2.  Similarly, can you clarify/explicitly define “sensitive things” and “non-sensitive things” referred to in bullet 6?

  3.  Wrt to the use of the term “shard”, are two shards (in the context of this document) allowed to contain overlapping subsets of the same data?  I think this is necessary if I understand the document’s narrative.  However, shards in the context of database technology typically store mutually exclusive partitions of the overall dataset (https://en.wikipedia.org/wiki/Shard_(database_architecture)).  There are minor exceptions such as metadata that describes the overall database structure that may be found in multiple shards (see https://en.wikipedia.org/wiki/Shard_(database_architecture)#cite_note-1).
Proposition: shard is not a good term to use here.  Subgraph is better.  There may be something else that is better.

  4.  In 7, can you elaborate/clarify the document’s concept of “context”? …esp. in “context may be pairwise, n-wise, or anywise” and “subject is always contextualized”.

  5.  Feedback: In the fullness of time, DID (Digital Identifier) identifiers will be used to index and retrieve all imaginable data stored in unimaginably large universe of data providers – including but not limited to transactions stored distributed trusted journals (aka DLT-based blockchains), records stored in relational databases, rows of data stored in CSV files, photos stored in a file system or content management systems, documents stored in IPFS, etc. etc.

A [Universal] DID will become a form of universal key for addressing and accessing any type of data stored anywhere.  Universal DID Servers (UDIDServers) will become the universal mechanism for addressing all types of data.

[Universal] DID Methods (UDID Methods) will specify the requirements for the storage of DID Subject data that is governed by a particular UID Method.

Think of the UDIDServer architecture as you would a JDBC/ODBC/ADO.NET style of data architecture uniformly architected around the concept of [Universal] DIDs.

When viewed through this lens, the concepts in the document are likely to change fairly substantially.

Best regards,
Michael Herman
Independent Blockchain Futurist, Architect, and Developer


From: Daniel Hardman <daniel.hardman@evernym.com>
Sent: August 15, 2019 5:08 PM
To: Adrian Gropper <agropper@healthurl.com>
Cc: Manu Sporny <msporny@digitalbazaar.com>; W3C Credentials Community Group <public-credentials@w3.org>
Subject: Re: Identity Hubs and Agents

Regarding the issue of terminology around identity hubs and agents (raised by Adrian in the Google Doc, and also re-mentioned in the email thread), I have a doc I'd like to share.

Before anyone roll their eyes and suggests that I need to rein in my terminologically imperialist tendencies :-) again, please suspend judgment for a second. When Daniel B and Sam C and I were writing the "Rhythm" paper about how hubs and agents could harmonize, we identified terminology as one of our big inhibitors to progress, and we spent time coming up with a list of concepts that needed terms. We did not get to closure on what those terms should be, so unlike my earlier push on agent terminology, I don't actually have a strong opinion about what terms the community should choose for these concepts. But I do care about the list of concepts itself. I think it might provide a good (but incomplete) set of concepts for which terms might be needed, with respect to the narrow subtopic of this email thread.

I've attached the doc as a PDF. @Manu Sporny<mailto:msporny@digitalbazaar.com> does that satisfy the need for the doc to be in the permanent archive, or do I need to paste it inline in my email?