[Specific DID Method Schemes] Very, very confuding non-recommendation for central registries needs to be removed

[mwherman2000]

In https://w3c-ccg.github.io/did-spec/#specific-did-method-schemes, it states:

Since DIDs are intended for decentralized identity infrastructure, it is NOT RECOMMENDED to establish a registry of unique DID method names. Rather the uniqueness of DID method names should be established via human consensus, i.e., a specific DID scheme MUST use a method name that is unique among all DID method names known to the specification authors at the time of publication.

...and then in the next paragraph, directs the reader to an Appendix listing specific centralized registries...

A list of known DID method names and their associated specifications is provided in Appendix A. Registries

Question 1. This is confuding to the reader, why these sentences been left in the draft spec?

Question 2. What is the process for disambiguating/ensuring uniqueness of DID method names?

[TomCJones]

agree with @mwherman2000 on this

[mwherman2000]

In support of my opening comment, Registeries is also a formal W3C CCG work item category: https://github.com/w3c-ccg/community/blob/master/work_items.md#registries ...and within that, the DID Methods Registry is a specific work item.

[mwherman2000]

a217857 doesn't address the specific wording issue highlighted at the beginning of this issue. The change only moved the offending sentences to another part of the document. Please re-open this issue until it is resolved.

[rhiaro]

The sentence wasn't moved, but rephrased in place. The intention is to note that new DID methods go through a lightweight human-centered W3C process with a certain bar to entry (DID method should be spec'd) whilst recommending against any kind of software-driven first-come-first-served registry of method names, in order to prevent a landgrab for method names.

[mwherman2000]

Me bad. My original questions 1 and 2 were a distraction from the core issue: Why do we say:

it is NOT RECOMMENDED to establish a registry of unique DID method names.

and then in the next paragraph say:

A list of known DID method names and their associated specifications is provided in Appendix A. Registries

Either there is a need or requirement for central registries or this isn't. We need to remove the contradictory messages.

If by "it is NOT RECOMMENDED to establish a registry of unique DID method names", we really mean "it is NOT PREFERABLE OR DESIRABLE to establish a registry of unique DID method names" ...this is an editorial comment/viewpoint and should be removed.

[talltree]

On Sun, Mar 3, 2019 at 6:10 PM Michael Herman (Toronto) < ***@***.***> wrote: Me bad. My original questions 1 and 2 were a distraction from the core issue: Why do we say: it is NOT RECOMMENDED to establish a registry of unique DID method names. and then in the next paragraph say: A list of known DID method names and their associated specifications is provided in Appendix A. Registries Either there is a need or requirement for central registries or this isn't. We need to remove the contradictory messages. If by "it is NOT RECOMMENDED to establish a registry of unique DID method names", we really mean "it is NOT PREFERABLE OR DESIRABLE to establish a registry of unique DID method names" ...this is an editorial comment/viewpoint and should be removed.

Michael, it turns out that the question of "registry of unique DID method names" is a *critically* important question to some implementers and some governance authorities. This was already clear before the Rebooting the Web of Trust conference that just ended in Barcelona, but there were several very deep discussions about it there. I would like to: 1. Propose that we replace any/all text dealing with this subject in the current version of the spec with a new, short section entitled "Governance of DID Method Namespaces". 2. Volunteer to draft the text explaining why this subject is out of scope for the spec. How does that sound to you (or to anyone else on this thread)?

[mwherman2000]

1. Yes, that would address and close this issue.
2. Let's see that the new section reads like.

[talltree]

Agreed—the wording of the new section is critical. I plan to make a proposal shortly.

[peacekeeper]

Methods are now listed in the separate DID Method Registry. And @rhiaro has done a great job with #186 which has been merged. I think this issue can be closed. @mwherman2000 do you agree?

[rhiaro]

Closing because there has been no further comment regarding the changes made. If there are issues with the new sections Unique Method Names and Choosing DID resolvers please raise new issues, and if there are problems with copy in the DID Method Registry itself, please raise issues over there.