[Lightning-dev] Proposal for "local" channel announcements.
ZmnSCPxj at protonmail.com
Sun Nov 4 15:04:14 UTC 2018
Good morning Rusty and all,
On reflection, it seems to me that non-public channels have the incentives very wrong.
1. Non-public channels are intended as a way to keep public maps small. So a node maintaining a non-public channel provides a service to the rest of the network by increasing number of participants without increasing map sizes of other nodes.
2. Users of non-public channels are not paid for the above service.
3. Users of non-public channels *pay* for their non-public channels by revealing to the other user of the channel that they are the only possible source/destination of payments.
Let me instead propose, a different mechanism (which is what actually initially occurred to me when I first saw "local" channel announcements on the list of topics for the upcoming summit).
1. On channel open, the initiator of the channel indicates a "local" or "global" channel. Current channels are "global". In the far future, non-public channels have been subjected to an Exterminatus order.
2. "Local" channels are only gossiped up to some small number of nodes away, say 3. This still reduces the sizes of maps, while still providing an increased anonymity set in the number of possible users of the local channel.
In my mind, something is wrong about non-public channels and their incentives. I suspect, some kind of "last mile" problem exists somehow.
Sent with ProtonMail Secure Email.
‐‐‐‐‐‐‐ Original Message ‐‐‐‐‐‐‐
On Sunday, November 4, 2018 12:21 PM, Rusty Russell <rusty at rustcorp.com.au> wrote:
> ZmnSCPxj ZmnSCPxj at protonmail.com writes:
> > Good morning Rusty,
> > To clarify, it seems the below:
> > 1. There is a "private" node, one whose channels are all non-published.
> > 2. There is a public node who knows that everything that passes through the channel with the "private" node comes only from the "private" node. It thus has an information advantage it might not have any incentive to sacrifice.
> This is true.
> > 3. This protocol is initiated by the public node, and if the public node does not initiate it, the "private" node can do nothing.
> > Is my understanding correct?
> More routes means more fees, though. Your peer can always offer
> substandard service, so I don't think this is worse.
More information about the Lightning-dev