W3C home > Mailing lists > Public > public-credentials@w3.org > April 2018

RE: "Decentralized Identifiers": Bitcoin Cargo-Culture and Land Grabbing for the Top Level Names

From: Reto Gmür <reto@factsmission.com>
Date: Tue, 24 Apr 2018 15:09:09 +0000
To: Manu Sporny <msporny@digitalbazaar.com>, "public-credentials@w3.org" <public-credentials@w3.org>
Message-ID: <AM0P191MB0451427C5B201BC6CEB40C31B6880@AM0P191MB0451.EURP191.PROD.OUTLOOK.COM>
Hi Manu, all,

“Money goes to your bank account” really doesn’t sound like something that happens with a decentralized approach. The money probably originates from the bank accounts of those self-registering an identifier. How does money goes from those that self-sovereignly register from those who offer resources (without some centralized authority)? And who makes sure that there's an incentive to store identity documents for a long time?

And if DIDs can equally well be used for centralized systems, what's all the talk about decentralization good for? Do "those of us in this community" really believe that simply adding some wishes to the URI spec will make them become true? If this would work, why not enhance the HTTP (now: Human Tranquil Togetherness Protocol) spec to say that the shared information should be free from hate-speech and promote truth and world-peace? 

We have decentralized URIs schemes like dat, ipfs. HTTP can be used with .onion hostnames. On the other hand there are approaches to have URIs identifying persons that can be dereferenced to a document (WebID). 

DID introduces a super-scheme that can be used for centralized as well as for decentralized systems. It is probably meant to identify persons and not inanimate objects or concepts. I still haven't seen an argument why such a scheme is needed.

As for the non-binding wishes associated with the scheme I think there are two very different approaches that can legitimately claim to be decentralized.

One approach is to have the IDs base on the public key of a key-pair. This is how PGP identifiers work, any user can create such an identifier without the need to register with any authority. The DID spec seems to envisage such an approach when it says that globally unique IDs come at the cost of human memorability. To use types of decentralization described by Vitalik Buterin in https://medium.com/@VitalikButerin/the-meaning-of-decentralization-a0c92b76a274, such a system is architecturally, politically and logically decentralized.

A second approach is to base identifiers on a Blockchain. As described by Aaron Swarz and implemented in Namecoins such a system can "square Zooko's triangle" and allow for human memorable identifiers. The DID spec repeatedly mentions the Blockchain and distributed ledger so it seems that the authors have such a system in mind. Such a system is architecturally decentralized but is logically centralized ("there is one commonly agreed state and the system behaves like a single computer"). Depending on the blockchain it can be politically decentralized, but this typically requires PoW and is thus environmentally and ethically highly problematic. 

The most prominently proposed methods for DID are neither politically nor logically decentralized. In that they fall in the same category as the DNS system. The architecture is decentralized but they have an organization ruling the system and create a logical center of authority. 

In the spirit of decentralization I advocate the first approach. The DID spec however seems to imply logical centralization with the very concept of method names that would refer to some sort of commonly agreed state. The idea of being able to dereference to DID-Documents also bases on the idea of a logically centralized registry. That Zooko's triangle isn't squared seems to be the only reminiscence of full decentralization.

So to summarize: I can't see any good reason for DIDs as per the vision embodied in the current spec and dissemination activities. This is just a system where companies and organizations with any degree of centralization can grab a method name and assign names within it. The scheme only provides a marketing advantage to these provider. A truly decentralized and thus provider-free system for personal identifiers would in my opinion indeed be useful. While in such as system identifiers do not depend on an entry in a logically centralized registries, the PGP Keyserver system shows, that some degree of dereferenceability can nevertheless be afforded.


> -----Original Message-----
> From: Manu Sporny <msporny@digitalbazaar.com>
> Sent: Monday, April 9, 2018 3:00 PM
> To: public-credentials@w3.org
> Subject: Re: "Decentralized Identifiers": Bitcoin Cargo-Culture and Land
> Grabbing for the Top Level Names
> On 04/09/2018 03:56 AM, Dennis Yurkevich wrote:
> > What are the economic incentives for implementers to store DIDs
> > forever?
> Fundamentally, nothing is free and these decentralized systems need money to
> operate. In every one of these systems, nodes are compensated for
> participating in the network on an ongoing basis. Fundamentally, if you run a
> node, money goes into your bank account. That's the economic incentive.
> > What stops method implementers not creating centralised and non SSI
> > systems?
> Nothing can stop that. The only thing the standards provide for is that if such a
> system were to be built, it will be technically interoperable with the larger
> system (but would lack some of the decentralized characteristics that those of
> us in this community would like to see).
> Market forces may doom such a centralized DID Method, but that happens in
> the future... which is hard to predict. :)
> -- manu
> --
> Manu Sporny (skype: msporny, twitter: manusporny, G+: +Manu Sporny)
> Founder/CEO - Digital Bazaar, Inc.
> blog: Veres One Decentralized Identifier Blockchain Launches
> https://tinyurl.com/veres-one-launches

Received on Tuesday, 24 April 2018 15:09:40 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 24 March 2022 20:24:47 UTC