On Thu, Dec 09, 2021 at 07:12:45AM -0500, Alex Schoof wrote:
> The multisig scheme is interesting. From my understanding of Single Use
> Seals, since seal n commits to seal n+1, for the on-chain aggregation seals
> you would want to pick some common aggregation service provider ahead of
> time and if that provider disappears, you’re stuck and cant close the next
> seal. If instead you say “this seal commits to three of the five of these
> next seals” then you mitigate both availability and censorship risk. Am I
> getting that right?

Re: "some common aggregation service provider", you might be misunderstanding
the protocol: since seals are trustless with regard to validity, I can validate
your seal, regardless of which aggregation service you use.

But other than that, I think we're on the same page!

A concrete example would be an exchange: they do a lot of transactions, so they
could choose to be their own aggregator, and wouldn't need any multisig at all
because they can trust themselves not to censor themselves. :) Meanwhile, one
of their customers might use 3-of-5 as you suggest, as they only do a few
transactions a month.

Interestingly, in some scenarios it might be worthwhile to both run your own
aggregator, and use multisig. Eg Alice could use a 2-of-3 with two third-party
aggregators, and her own aggregation chain. If both third-parties are up, she
does no on-chain transactions at all; if one third-party is down, she can use
her own, and the remaining third-party. Thus she would only do an on-chain
transaction to defeat censorship/failure.

-- 
https://petertodd.org 'peter'[:-1]@petertodd.org