Date: Mon, 21 Sep 2020 01:14:29 +0000
To: Tom Trevethan <tom@commerceblock.com>
From: ZmnSCPxj <ZmnSCPxj@protonmail.com>
Reply-To: ZmnSCPxj <ZmnSCPxj@protonmail.com>
Message-ID: <TC2UtcaDKCIkjtn41sIutqApciyqaGCD3SBSEWLBk4OT12siSkWIsp2LMmlmA0CLzUNuG1W8w-hB4Pq3ko1uGwbxpjxezFWKPq4C7OLUQo8=@protonmail.com>
In-Reply-To: <CAJvkSse5GDxzrDc0OeNSoHV90tsSvr-oCgY_P+p4O4T6PDUNpA@mail.gmail.com>
References: <CAJvkSseOx8mwYowLEnfKVEUuM5xsiYkbLdAvtaYxPpu4jY9pCg@mail.gmail.com>
 <2Abk4Gqv5hJyGtA8Gg1WCP5RNuyUFkmRn1uUKp_mdUaXrRTz4SDBTPi0MGU7D5jj36VSzrqsIiO5lMR4gGRApRX2jyp8vXDeHBnFt-6ca-g=@protonmail.com>
 <CAJvkSse5GDxzrDc0OeNSoHV90tsSvr-oCgY_P+p4O4T6PDUNpA@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
Cc: Bitcoin Protocol Discussion <bitcoin-dev@lists.linuxfoundation.org>
Subject: Re: [bitcoin-dev] Statechain coinswap: assigning blame for failure
	in a two-stage transfer protocol.
Precedence: list

Good morning Tom,

> Hi ZmnSCPxj,
>
> Thanks for the reply.=C2=A0
>
> > Okay, I suppose this is much too high-level a view, and I have no idea =
what you mean by "statecoin" exactly.
>
> Sorry, most of the protocol details are in the links, but terminology sho=
uld be made clearer. A=C2=A0"statecoin" is a UTXO that is a 2-of-2 between =
the owner and SE (the tr*sted signing server) i.e. can be transferred off-c=
hain.=C2=A0
>
> Also, should have been clear that `addr1` is the 'statecoin=C2=A0address'=
 which is different from the on-chain address (the shared public key the bi=
tcoin is paid to). The on-chain address does not change, whereas the=C2=
=A0'statecoin=C2=A0address' changes with each new owner and is used to auth=
enticate owners to the SE and act as proof of ownership on the=C2=A0statech=
ain - it is not related to the onchain address/pubkey and controlled by the=
 owner only.=C2=A0
>
> > So it seems to me that this requires tr\*st that the coordinator is not=
 going to collude with other participants.
>
> This is correct. The SE also must be trusted to not actively defraud user=
s. The main advantage of this scheme is that assuming the SE can be trusted=
, it is strictly non-custodial.=C2=A0
>
> > This is strictly worse than say Wasabi, where the coordinator colluding=
 with other participants only allows the coordinator to break privacy, not =
outright steal funds.
> > It seems to me that the trust-minimized CoinSwap plan by belcher_ is su=
perior to this, with reduced scope for theft.
>
> This is true if the overriding aim is trust minimisation, but not if the =
aim is speed and cost while staying=C2=A0non-custodial. Off-chain SE transa=
ctions are near instant and orders of magnitude cheaper than on-chain. Prob=
ably best thought of as a non-custodial centralised mixer.=C2=A0


I think the entire point of non-custodiality ***is*** trust minimization.

The main objection against custodiality is that someone else can prevent yo=
u from spending the coin.
If I have to tr\*st the SE to not steal the funds, is it *really* non-custo=
dial, when after a swap, a corrupted SE can, in collusion with other partic=
ipants, take control of the coin and prevent me from spending it as I wish?

So I think touting "non-custodial" is relatively pointless if tr\*st is not=
 minimized.

(I am aware there is an update mechanism, either Decker-Russell-Osuntokun o=
r Decker-Wattenhofer, that is anchored off he onchain transaction output, b=
ut anyone who can recover the raw keys for signing the funding transaction =
output --- such as a previous participant and a corrupt SE --- can very eas=
ily bypass the mechanism.)

For example, in my previous description of [implementing investment aggrega=
tion](https://lists.linuxfoundation.org/pipermail/bitcoin-dev/2020-July/018=
055.html), while I admit you need tr\*st in the business owners who you are=
 investing in, it does not require tr\*st in the aggregator, due to the n-o=
f-n, which cannot be reconstructed by the aggregator and all other particip=
ants without you.

Regards,
ZmnSCPxj