Re: DIDs and httpRange-14

On 4/11/19 7:39 AM, Markus Sabadello wrote:
> But wouldn't the WebID approach mean that for example we would have to 
> write VCs like this?
> {
>    "id": "",
>    "type": ["VerifiableCredential"],
>    "issuer": "*did:btcr:xkrn-xzcr-qqlv-j6sl#me*",
>    "issuanceDate": "2010-01-01T19:73:24Z",
>    "credentialSubject": {
>      "id": "*did:sov:WRfXPg8dantKVubE3HX8pw#me*",
>      ....
>    }
> }
> That's not intuitive and would break existing implementations.
> I think most in the group would agree that the bare DID identifies 
> (denotes) Alice (the DID subject).
> My main question on this subject has been whether or not Alice's DID 
> Document should have its own URI / URL that is different from Alice's URI.

As I said in the Google Doc, I think this is a non-problem. Yes, make
them the same URI (i.e. change nothing we've already been doing). Until
someone presents a serious real world issue that cannot be reasonably
solved without adopting this hash approach, we should move on. I suspect
that the hash approach will likely increase implementation and cognitive
burden without practical benefit.

What if you want to make a statement about my identifier "foo#me" and
not about me? Do we need a new identifier for my identifier? Where will
it end? I think the only practical way it ends is by saying something
more about the thing you're talking about that will create the
appropriate distinction. For example, include information like `"type":
"DidDocument"` or `"type": "Person"` if you feel that's necessary to
avoid confusion when making statements. I assert that, in the ecosystem
we are trying to build, people will be using DIDs to refer to the DID
subject nearly 100% of the time.

There's a reason this issue has taken many groups down a DoS rabbit
hole: it is really more about philosophy than a practical concern. Any
abstraction is necessarily not the thing to which it refers. But since
we can only conceive of abstractions anyway, let's stop worrying about
it so much and get things done.

> Markus
> On 4/11/19 11:04 AM, Kjetil Kjernsmo wrote:
>> On onsdag 10. april 2019 17.32.34 CEST Markus Sabadello wrote:
>>> I wouldn't call this a high priority issue, but there have been
>>> recurring questions on whether a DID is a URL or "only" a URI, what a
>>> DID identifies (Alice? Or Alice's DID Document?), and what it means
>>> exactly to resolve/dereference a DID (URL).
>> Nice writeup! Indeed, my vote would be to take WebID's example and use hash
>> URIs for Alice.
>> Kjetil

Dave Longley
Digital Bazaar, Inc.

Received on Thursday, 11 April 2019 15:30:46 UTC