[EGreg]

One of my acquaintainces at Microsoft worked for years to get this through: https://identity.foundation/sidetree/spec/

DIDs built as a Sidetree on top of Bitcoin. At least they don’t require everything being on-chain, but securing with proof-of-work is unfortunate.

[de_keyboard]

PoW is the only proof mechanism that has seen real world use for a significant length of time.

[EGreg]

Sure, and in the 50s, vacuum tubes were the only mainstream computing mechanism that had seen real world use for a significant length of time. Same goes for all other outdated technologies.

When you have a distributed system, it really isn’t hard to “retain” information in a byzantine fault tolerant way. Proof of Work is mostly used to help with liveness, not persistence.

[southerntofu]

What kind of proof mechanism are you talking about? Do you mean PoW can prevent an attacker spawning many identities? That's certainly not the case, because bad actors have plenty of resources.

PGP signatures + Web of Trust have been in use for about two decades now and have proved robust over time (can't say the same of all PGP implementations unfortunately), but at the cost of revealing (parts of) the social graph which is an anti-feature for privacy.

I've heard the word Fog of Trust employed by GNU/Net project to refer to research projects on zero-knowledge proofs for a Web of Trust, so you can infer trust relationships without exposing the social graph. But i can't vet for the math behind that.

I'd be happy to have more alternative patterns explored, instead of insisting on the aspects that made Bitcoin a failure. Bitcoin was a revolutionary PoC for replacing centralized trust with trust in the majority of global computing resources allocated to the network. But that model has shown its limits and weaknesses (high economic/environmental cost, vulnerability to advanced actors like Bitmain, low transaction throughput) so we can research other approaches.

So far, the most advanced/consistent proposal i've read on decentralized identity, which is not based on monetary speculation or proof-of-work, is the GNU Name System [0], which features recursive resolution of zones (retro-compatible with DNS) via a global DHT, crypto-secure zone delegation (retrofittable into existing ICANN infrastructure), hyper-hyper local root (like /etc/hosts but with recursive resolution), query privacy (client requests don't leak), enumeration-proof zones (private zone entries).

Alongside the reclaim:ID [1] self-sovereign identity scheme, that looks like the most solid proposal for replacing both insecure DNS infrastructure and cracking the decentralized identity problem. Alongside the Taler [2] privacy-friendly payment platform, that looks like the most solid proposal for enabling fully-decentralized pseudonymous electronic transactions.

See also this somewhat recent article on the topic: https://gnunet.org/en/news/2021-05-DISSENS.html

[0] Some video presentations: https://gnunet.org/en/video.html

[1] https://reclaim.gnunet.org

[2] https://taler.net/en/

[geofft]

All the more reason to incentivize real-world use of other approaches. "This is how we've always done it" is poor reasoning, whether it's about blockchain or lead pipes or asbestos.