Date: Thu, 14 May 2020 04:02:07 +0000
To: Antoine Riard <antoine.riard@gmail.com>
From: ZmnSCPxj <ZmnSCPxj@protonmail.com>
Reply-To: ZmnSCPxj <ZmnSCPxj@protonmail.com>
Message-ID: <uOUyhfZ-Ti4E4sQn_Cap6Em_pqVc-p2INXoBLIEKsiOWpWKT-WNeqUge902E-HU0wWWWo4onr8UQTNKg5YmVkUWfrlNVJkkMSWYCnoj2WVY=@protonmail.com>
In-Reply-To: <CALZpt+G8SzeX4U-VBhEZqQ0ApwAs_jKkKe7aeZEQZ5KcJaMjCg@mail.gmail.com>
References: <CALZpt+GBPbf+Pgctm5NViNons50aQs1RPQkEo3FW5RM4fL9ztA@mail.gmail.com>
 <202005051300.38836.luke@dashjr.org>
 <CAH5Bsr27rN1SE166ON_q49=MNti0v7Vyn6s6T5R3=LC69K2QdQ@mail.gmail.com>
 <6883e35a-e584-523f-d6f9-cf9ce2cca66d@riseup.net>
 <CALZpt+G8SzeX4U-VBhEZqQ0ApwAs_jKkKe7aeZEQZ5KcJaMjCg@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>,
 "lightning-dev\\\\\\\\@lists.linuxfoundation.org"
 <lightning-dev@lists.linuxfoundation.org>
Subject: Re: [bitcoin-dev] [Lightning-dev] On the scalability issues of
	onboarding millions of LN mobile clients
Precedence: list

Good morning Antoine,


> While approaching this question, I think you should consider economic wei=
ght of nodes in evaluating miner consensus-hijack success. Even if you expe=
ct a disproportionate ratio of full-nodes-vs-SPV, they may not have the sam=
e =C2=A0economic weight at all, therefore even if miners are able to lure a=
 majority of SPV clients they may not be able to stir economic nodes. SPV c=
lients users will now have an incentive to cancel their hijacked history to=
 stay on the most economic meaningful chain. And it's already assumed, that=
 if you run a bitcoin business or LN routing node, you do want to run your =
own full-node.

One hope I have for Lightning is that it will replace centralized custodial=
 services, because:

* Lightning gains some of the scalability advantage of centralized custodia=
l services, because you can now transfer to any Lightning client without to=
uching the blockchain, for much reduced transfer fees.
* At the same time, it retains your-keys-your-coins noncustodiality, becaus=
e every update of a Lightning channel requires your keys to sign off on it.

If most Lightning clients are SPV, then if we compare these two worlds:

* There are a few highly-important centralized custodial services with sign=
ificant economic weight running fullnodes (i.e. now).
* There are no highly-important centralized custodial services, and most ev=
eryone uses Lightning, but with SPV (i.e. a Lightning future).

Then the distribution of economic weight would be different between these t=
wo worlds.
It may even be possible, that the Lightning future with massive SPV might e=
nd up with more economic weight in SPV nodes, than in the world without Lig=
htning and dependent on centralized custodial services to scale.


It is also entirely possible that custodial services for Lightning will ari=
se anyway and my hope is already dashed, come on universe, work harder will=
 you, would you really disappoint some randomly-generated Internet person l=
ike that.


>
> I agree it may be hard to evaluate economic-weight-to-chain-backend segme=
nts, specially with offchain you disentangle an onchain output value from i=
ts real payment traffic. To strengthen SPV, you may implement forks detecti=
on and fallback to some backup node(s) which would serve as an authoritativ=
e source to arbiter between branches. Such backup node(s) must be picked up=
 manually at client initialization, before any risk of conflict to avoid Re=
ddit-style of hijack during contentious period or other massive social engi=
neering. You don't want autopilot-style of recommendations for picking up a=
 backup nodes and avoid cenralization of backups, but somehow a uniform dis=
tribution. A backup node may be a private one, it won't serve you any data =
beyond headers, and therefore you preserve public nodes bandwidth, which IM=
O is the real bottleneck. I concede it won't work well if you have a ratio =
of 1000-SPV for 1-full-node and people are not effectively able to pickup a=
 backup among their social environment.
> What do you think about this model ?

Money makes the world go round, so such backup servers that are publicly-fa=
cing rather than privately-owned should be somehow incentivized to do so, o=
r else they would not exist in the first place.
Of course, a free market tends towards monopoly, because any entity that ha=
ppens to have even a slight advantage at the business will have more money =
to use towards business reinvestment and increase its advantage further, un=
til they beat the competition to dust, anyone who has won a 4X game knows t=
o search for and stack those little advantages until you snowball and conqu=
er the world/galaxy/petri dish which is why the endgame of 4X games is so b=
oring compared to the start, we have seen this happen in mining and exchang=
es and so on, and this works against your desire to have a uniform distribu=
tion.

If everyone runs such a privately-owned server, on the other hand, this is =
not so different from having a Lightning node you run at your home that has=
 a fullnode as well and which you access via a remote control mobile device=
, and it is the inconvenience of having such a server at your home that pre=
vents this in the first place.

Regards,
ZmnSCPxj