MIME-Version: 1.0
Sender: gmaxwell@gmail.com
From: Gregory Maxwell <greg@xiph.org>
Date: Thu, 18 Jan 2018 19:30:09 +0000
Message-ID: <CAAS2fgQa8hXO1VsXUnUGnfu17dM0B-Jtfa7TNW=O3M8Mt=t2eQ@mail.gmail.com>
To: Bitcoin Dev <bitcoin-dev@lists.linuxfoundation.org>
Content-Type: text/plain; charset="UTF-8"
Subject: [bitcoin-dev] ScriptPubkey consensus translation
Precedence: list

A common question when discussing newer more efficient pubkey types--
like signature aggregation or even just segwit-- is "will this thing
make the spending of already existing outputs more efficient", which
unfortunately gets an answer of No because the redemption instructions
for existing outputs have already been set, and don't incorporate
these new features.

This is good news in that no one ends up being forced to expose their
own funds to new cryptosystems whos security they may not trust.  When
sigagg is deployed, for example, any cryptographic risk in it is borne
by people who opted into using it.

Lets imagine though that segwit-with-sigagg has been long deployed,
widely used, and is more or less universally accepted as at least as
good as an old P2PKH.

In that case, it might be plausible to include in a hardfork a
consensus rule that lets someone spend scriptPubkey's matching
specific templates as though they were an alternative template.  So
then an idiomatic P2PKH or perhaps even a P2SH-multisig could be spent
as though it used the analogous p2w-sigagg script.

The main limitation is that there is some risk of breaking the
security assumptions of some complicated external protocol e.g. that
assumed that having a schnorr oracle for a key wouldn't let you spend
coins connected to that key.  This seems like a pretty contrived
concern to me however, and it's one that can largely be addressed by
ample communication in advance.  (E.g. discouraging the creation of
excessively fragile things like that, and finding out if any exist so
they can be worked around).

Am I missing any other arguments?