[bitcoin-dev] [BIP Proposal] Simple Proof-of-Reserves Transactions

From: Steven Roose <stevenroose@gmail•com>
To: bitcoin-dev@lists•linuxfoundation.org
Subject: [bitcoin-dev] [BIP Proposal] Simple Proof-of-Reserves Transactions
Date: Tue, 29 Jan 2019 22:03:04 +0000	[thread overview]
Message-ID: <CAChG=YV2em+6c9P4DEUB1=+ZSEA6S0j9HDuWoKBeRVMmtzaMNg@mail.gmail.com> (raw)

[-- Attachment #1: Type: text/plain, Size: 10638 bytes --]

Been working on a proof-of-reserves tool POC for a while, trying to
formalize the formats so that wallets can integrate more easily.

<pre>
  BIP: ?
  Layer: Applications
  Title: Simple Proof-of-Reserves Transactions
  Author: Steven Roose <steven@stevenroose•org>
  Comments-Summary: No comments yet.
  Comments-URI: tbd
  Status: Draft
  Type: Standards Track
  Created: 2019-01-28
  License: CC0-1.0
</pre>

==Abstract==

This BIP describes a simple way to construct proof-of-reserves transactions.
This proposal formalizes a standard format for constructing such proofs,
easing
their construction with existing wallet infrastructure and enabling general
proof-verification software.  It relies on existing standards such as
regular
Bitcoin transaction serialization/validation and the BIP 174 PSBT format.
The proposal also includes the description of a PSBT extension for a better
user experience.

==Copyright==

This BIP is licensed under the Creative Commons CC0 1.0 Universal license.

==Motivation==

From the very early days in the history of Bitcoin, there have been
companies
managing bitcoins for their users.  These users give up control over their
coins
in return for a certain service.  Inevitably, there have been many cases of
companies losing their users' bitcoins without timely disclosing such
events to
the public.  Proofs of Reserves are a way for companies managing large
amounts
of bitcoins to prove ownership over a given amount of funds.  The regular
proof
of control helps to ensure that no significant loss has occurred.

While the term proof-of-reserves is not new by any means, the procedure is
not
very common among high-value custodian companies.  One of the reasons for
this
is that every company that wants to perform a proof-of-reserves has to
construct
its own way to do so.  Accordingly, their users have to understand the
construction of the proof in order to be able to verify it.  This raises
the bar
of entry both for custodians and for users.

===What this BIP is not doing===

The proof-of-reserve construction described in this document has some known
shortcomings, mostly with regards to its privacy properties.  While there
exists
research about improved proof-of-reserves mechanisms that have much better
privacy properties<ref>Dagher, Gaby G., Benedikt Bünz, Joseph Bonneau,
Jeremy
Clark, and Dan Boneh. "Provisions: Privacy-preserving proofs of solvency for
Bitcoin exchanges." (2015).</ref>, this BIP intentionally only formalizes
the de-facto existing method.

==Specification==

Our specification consists of two parts:
# the format for the actual proofs
# a file format used to package a set of proofs and relevant metadata

The final construction should have the following properties:
* flexible proof construction to support complex wallet infrastructures
* easy integration with existing wallet solutions (both hardware and
software wallets)
* support for verification via a standard procedure, regardless of
publisher of the proof
* proof prevents reuse of proofs by other parties by commiting to a message
* allow validating that the issuer had the funds under his control at a
certain block, regardless of what happened after that block

===Proof Format===

To allow for maximal compatibility with existing systems, proofs are
formatted as regular Bitcoin
transactions.  However, one small adaptation to the transaction is made
that has two functions:
# make the transaction unspendable to avoid putting funds at risk
# link the proof to the issuer of the proof to prevent copying proofs from
other custodians

The resulting construction is a Bitcoin transaction with the following
characteristics:

* The first input (the "commitment input")
** MUST have the txid part of the previous outpoint set to the SHA-256 hash
of the commitment message prefixed with "Proof-of-Reserves: "<ref>If the
message is "Some Message", the txid part should be
<tt>SHA-256("Proof-of-Reserves: Some Message")</tt> with the string encoded
as UTF-8.</ref> and index 0.
* The remaining inputs
** MUST have signatures that commit to the commitment input (e.g. using
<tt>SIGHASH_ALL</tt>).
* The transaction MUST have a single output that is the exact sum of all
the inputs, assuming the commitment input to have 0 value; this means the
transaction has no miner fee.

The existence of the first input (which is just a commitment hash) ensures
that this transaction is invalid and can never be confirmed.

===Proof File Format===

In theory, the first part of the specification would be sufficient as a
minimum
viable standard.  However, there are a number of motivations to extend the
standard with an extra layer of metadata:

# constructing and combining multiple proofs
#:Having thousands of UTXOs spread across different offline and online
wallets could make it difficult to construct a single proof transaction
with all UTXOs.  Allowing multiple proof transactions with the same
commitment message and block number gives extra flexibility to custodians
with complex wallet infrastructure without making the combined proof less
secure.
# metadata for verification
#:Not all systems that will be used for verification have access to a full
index of all transactions.  However, proofs should be easily verifiable
even after some of the UTXOs used in the proof are no longer unspent.
Metadata present in the proof allows for relatively efficient verification
of proofs even if no transaction index is available.
# potential future improvements
#:The extensible metadata format allows for amending the standard in the
future.  One potential improvement would be having UTXO set commitments.
These would allow the proofs-of-reserves to come with accompanying
proofs-of-inclusion of all used UTXOs in the UTXO set at the block of proof
constsruction (making validation even more efficient).

The proposed proof-file format provides a standard way of combining multiple
proofs and associated metadata.  The specification of the format is in the
Protocol Buffers<ref>https://github.com/protocolbuffers/protobuf/</ref>
format.

<pre>
syntax = "proto3";
import "google/protobuf/any.proto";

message OutputMeta {
// Identify the outpoint.
bytes txid = 1;
uint32 vout = 2;

// The block hash of the block where this output was created.
bytes block_hash = 3;
}

message FinalProof {
// The proof transaction.  Should be able to be parsed like a regular
// Bitcoin transaction.
bytes proof_tx = 1;

// The metadata of the ouputs used in the proof transaction.
repeated OutputMeta output_metadata = 2;
}

message ProofOfReserves {
// A version number for this format to enable extending it with
// additional fields.
uint32 version = 1;

// The network magic for the network in which the proofs are valid.
// 0xD9B4BEF9 for mainnet, 0x0709110B for testnet
//TODO consider BIP44 coin type ids instead:
// https://github.com/satoshilabs/slips/blob/master/slip-0044.md
uint32 network_magic = 2;

// The commitment message for this proof-of-reserves.
// This message is global for all the proofs.
string message = 3;

// The block at which this proof is supposed to be validated.
// Verification should take into account unspentness of outputs at this
// block height.
bytes block_hash = 4;

// The set of final proof transactions with their output metadata.
repeated FinalProof final_proofs = 5;

// Reserved field that can potentially be used by proof-construction tools.
// It can be ignored for verification.
repeated google.protobuf.Any pending_proofs = 6;
}
</pre>

The last field, <tt>pending_proofs</tt>, leaves open some space in the same
file that can be used by proof-construction tools.  This allows them to
construct different proofs incrementally without having to switch between
file
formats.

===PSBT (BIP 174) extension===

The "commitment input" detailed in the proof format section does not spend
an
existing UTXO and thus shouldn't be signed (empty <tt>scriptSig</tt> and
witness).  This can cause some problems when signing this type of
transactions.
For example, hardware wallets often require the signer to provide
information
about all inputs of transactions they are signing, such as the previous
output
or previous transaction; this data obviously doesn't exist for the
commitment
inputs.

For most existing devices, it's possible to circumvent these requirements by
providing dummy data or by instructing the device to ignore this specific
input.  However, there is still a UX problem.  Because the hardware wallet
device doesn't recognize the transaction as a proof-of-reserves transaction
it
will think it is signing a regular transaction that is spending all the
money
in the UTXOs.  Most devices will ask for confirmation with a message along
the
lines of "Are you sure you want to send XXX BTC to address [...]?".  This is
not the best user experience.

An addition to the BIP 174 PSBT format could help signing devices to
recognize proof-of-reserve transactions.
The following field is added to the BIP 174 <tt>INPUT</tt> map:

* Type: Proof-of-reserves commitment <tt>PSBT_IN_POR_COMMITMENT = 0x09</tt>
** Key: None. The key must only contain the 1 byte type.
*** <tt>{0x09}</tt>
** Value: The UTF-8 encoded commitment message string for the
proof-of-reserves.
*** <tt>{porCommitment}</tt>

Wallets processing an input that has this field set
* MUST make sure the txid of the previous outpoint is set to the SHA-256
hash of the prefixed commitment message string, as detailed above;
* MUST assume the input value to be 0 (without requiring the previous
output or transaction to be provided);
* SHOULD display the commitment message to ask the user for confirmation
before signing any inputs;
* SHOULD only provide signatures with a signature hash that commits to this
input;
* SHOULD accept an empty <tt>scriptSig</tt> for this input (as if the
<tt>scriptPubKey</tt> was <tt>OP_TRUE</tt>).

==Implementations==

A proof-of-concept implementation of the PSBT extension in the
[https://github.com/rust-bitcoin/rust-bitcoin rust-bitcoin] project can be
found in the <tt>psbt-por</tt> branch here:
https://github.com/stevenroose/rust-bitcoin/tree/psbt-por

A work-in-progress implementation of a tool that produces and verifies
proofs
in the described format can be found here:
https://github.com/stevenroose/reserves

== Footnotes ==

<references />

PR: https://github.com/bitcoin/bips/pull/756

[-- Attachment #2: Type: text/html, Size: 14648 bytes --]