From joseph at lightning.network  Tue Jul 28 01:42:24 2015
From: joseph at lightning.network (Joseph Poon)
Date: Mon, 27 Jul 2015 18:42:24 -0700
Subject: [Lightning-dev] Single-funder anchor model?
In-Reply-To: <87egjupc5f.fsf@rustcorp.com.au>
References: <87egjupc5f.fsf@rustcorp.com.au>
Message-ID: <20150728014212.GA10070@lightning.network>

On Mon, Jul 27, 2015 at 11:58:12AM +0930, Rusty Russell wrote:
> The asymmetric risk is as much a feature as a problem: with the
> dual-anchor proposal, either side could abort with no penalty and make
> the other side wait for the escape timeout anyway.

Cool, I think this works.

Yeah non-cooperation risks in this model aren't that big of a deal,
because the channel is going to be closed out anyway if they're not
cooperative -- they could always refuse to route payments, and then
everyone's going to get their money back.

> We've handwaved over the incentives for channel creation so far; they're
> tied with routing, and not immediately clear to me.  But it doesn't seem
> unreasonable that if you connect to a hub, you front the funds.

Yeah, there's likely some weird asymmetric economic incentive things
going on. If both channels are established with the same person, it
should be fairly clean, though. I think there will probably be some
measure of trust/reputation involved with pre-payment of the time-value
for the channel (shorter with OP_CSV, but still non-zero).

OP_CSV still requires BIP 62, though. It's possible to construct a model
with OP_CLTV without BIP 62 (described in a post earlier today) using
single-funder with some OP_CLTV'd output which returns the full balance
to the original funder at a date very far in the future after the
expiration of all Commitments and its dependent outputs.

-- 
Joseph Poon