Re: Why we really can't use Facebook for technical discussion.

[Phillip Hallam-Baker <[e-mail]>]

On Sat, Jun 5, 2021 at 11:48 PM Dean Willis <[e-mail]>
wrote:

>
>
> -------- Original Message --------
> On Jun 5, 2021, 10:17 PM, Phillip Hallam-Baker <[e-mail]>
> wrote:
>
> Mathematical Mesh 3.0 Part VII: Mesh Callsign Service (ietf.org)
>
> ----
>
> This is exactly where I was trying to go with P2P-SIP/RELOAD but didn't
> get to, and it seems quite clever. The key is in using the distributed tree
> to store ephemeral but still relatively long-lived registrations, not
> massive or rapidly changing data.
>

Thanks,

The key is stripping out all the expensive parts of running DNS. Query is
pushed onto the Mesh Service Providers. So all the registry does is to
receive registration requests, validate the signatures on them and publish
them if correct. The registry log is broadcast in real time to the service
providers subscribing to it and is signed at regular intervals. Unlike the
DNS, there is no single point of failure, the most an attacker can hope for
is to take out one Service Provider, not 'the Internet' as would be the
case if .com went down. Even taking out the callsign registry merely
prevents updates reaching the service providers for a short while.

Cross notarization of the user and their Mesh Service Providers and the
Mesh Service Providers and the callsign registry is sufficient to mesh all
the signed notarized logs into a single entity. No party can defect
undetected without the collusion of every other party. So Alice would have
to be party to her own bamboozlement.

Once Alice has registered @alice, it is hers forever, it is bound to her
public key credential. If she misbehaves, curation services can mark her
account as being abusive. But they only have the power to describe, the
power to accept or reject her content lies with Alice herself.