Re: [Ietf-hub-boston] proof of work and consensus protocols

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

On Thu, Nov 7, 2019 at 10:12 AM Salz, Rich <[e-mail]> wrote:

> This came up during Phillip’s presentation, there’s apparently a new paper
> out:
>
>     https://twitter.com/IACR_News/status/1192387232855973889?s=19
>

Thanks.

I think I have come across a solution that actually solves the problems
raised at the hub meeting and it folds back on the central axiom I have
that each user should ultimately be their own root of trust.

So alice has a Mesh account which means that she has a DARE Catalog that
tracks her set of devices. That catalog is authenticated by means of a
Merkle Tree. So she creates an attestation at regular intervals and submits
it to her Mesh service which will include it in the service notary log and
ultimately mesh it in the internotary. And she will periodically accept the
attestation of her Mesh service and include it in her own chain.

So Alice will be her own ultimate root of trust.

Now we do have a Descartes daemon type situation. What if her Mesh service
defects?

But the whole point of this infrastructure is to allow Alice to
communicate. So can we somehow make it a condition of Alice being able to
communicate with Bob that they are going to discover a possible defection
by their respective Mesh services?


I think we can *provided that* the diameter of the inter-notarization graph
of the Mesh services is sufficiently bounded. Which is almost certainly
going to be the case as there will be at least one company offering an open
notarization service.