From: Boris Nagaev <bnagaev@gmail•com>
To: Bitcoin Development Mailing List <bitcoindev@googlegroups.com>
Subject: Re: [bitcoindev] Pre-emptive commit/reveal for quantum-safe migration (poison-pill)
Date: Thu, 5 Jun 2025 07:54:25 -0700 (PDT) [thread overview]
Message-ID: <2ed6d831-d87d-44f0-82b9-c8b9abaa8010n@googlegroups.com> (raw)
In-Reply-To: <ea71bbd5-1325-445b-977f-a52b8017eab4n@googlegroups.com>
[-- Attachment #1.1: Type: text/plain, Size: 11604 bytes --]
Hi Leo, hi list,
You raise a valid point that a wTXID coming from a quantum attacker could
be invalid, and the scheme must be robust against this. In this scenario,
the attacker is not motivated by profit but rather by the desire to harm
the legitimate owner.
I think we do not need to include a full transaction in the announcement to
address this.
I propose the following scheme. There is just one announcement type, which
was defined as a weak proof in my previous message. An announcement
includes the UTXO, wTXID, and a proof (tagged_hash(EC public key ||
wTXID)). This tuple can be stored in an OP_RETURN output. The rule of a
hypothetical soft fork would be: to spend an EC output, there must be an
announcement with the corresponding wTXID and a valid proof published at
least X blocks earlier.
This design prevents an attacker from maliciously blocking someone else's
coin. There is no "earliest announcement wins" rule. Any valid announcement
wins once it is sufficiently buried. If an attacker publishes an invalid
announcement, the legitimate owner can simply publish their own and wait X
blocks before sweeping the funds. If they have already published an
announcement (as they should), they can use it without delay.
The scheme also preserves Satoshi's privacy: anyone can publish an
announcement for Satoshi's coins, but without the actual spending
transaction, nobody can tell whether the announcement is valid.
A downside of the scheme is that all announcements must be stored by nodes
until the corresponding UTXO is spent, and there can be multiple
announcements for a single coin. Another downside is that announcements
cannot be packed into a Merkle root, because the UTXO must be visible for
Satoshi to identify his coins, and he must be able to verify the proof.
Perhaps there are ways to improve the scheme to address these efficiency
issues.
Best,
Boris
On Thursday, June 5, 2025 at 9:55:34 AM UTC-3 Leo Wandersleb wrote:
Hi Boris, hi list,
I think the weak announcement is a bad idea once EC crypto is broken to the
point where an attacker can break the key before the transaction gets mined
but the strong announcements should still hold as they have less urgency.
If the attacker sees the transaction in a strong announcement with a full
transaction, he cannot win even if he gets into a block first, as the
strong announcement proves a prior commitment to that transaction and would
win even if it gets mined only some blocks later.
A scheme where the announcement does not contain the full transaction is
problematic as the transaction might then turn out to not be valid. Then
nodes would wait for the "winning" wtxid blocking the UTXO forever.
So the scheme is:
After activation at block height X:
1. **Vulnerable UTXOs cannot be spent directly** - they require a prior
announcement
2. **Commitment** to a wTXID that spends the vulnerable UTXO. Multiple
wTXIDs can be stored in a hash tree in an OP_RETURN
3. **Reveal** full transaction with proof of prior commitment but not as a
normal transaction yet
4. **Counter Reveal**: For 144 blocks, others can reveal older commitments.
This protects exposed pubkeys!
5. **After 144 blocks**: The UTXO can be spent according to the strongest
announcement (oldest commitment of valid transaction wins).
As (5) is just the normal transaction, the scheme is a soft fork and
compatible with pre-recorded transactions where the keys were lost. It
would at least double the on-chain costs for these vulnerable UTXOs as they
would have to store the full transaction twice. We can make the
announcements prunable again though.
Best,
Leo
On Wednesday, 4 June 2025 at 20:40:32 UTC+2 Boris Nagaev wrote:
Hi Leo,
I think it is possible to provide privacy for Satoshi and also reduce the
size of a weak announcement (strong announcements can already be small:
just a txid or a Merkle root of many txids).
Importantly, we cannot include the whole signed transaction in the weak
announcement. Doing so would leak the EC public key immediately, allowing
an attacker to create their own valid weak announcement. We must avoid
revealing the public key until the actual spending transaction is broadcast.
We need a scheme where the EC public key is not leaked in a weak
announcement, but the legitimate owner can verify it, while no one else
can. Also, once the EC public key is revealed, anyone should be able to
verify a past weak announcement (to validate the transaction when it is
broadcast). This reduces to the following requirement: we need a proof of
knowledge of the EC public key that can be verified if the public key is
known and provides no information otherwise.
I think this is called a zero-knowledge proof. One simple approach could be
to apply a tagged hash function to the concatenation of the EC public key
and the future wTXID, and include this in the weak announcement. The
structure would be:
- UTXO (previous TXID and output index)
- future spending wTXID
- proof := tagged_hash(EC public key || wTXID)
The wTXID is included in the concatenation to bind the proof to a
particular future transaction. Otherwise, someone could copy a weak
announcement and substitute their own wTXID.
Satoshi could publish a strong announcement now and then monitor all weak
announcements involving his UTXOs. If someone publishes a weak announcement
for one of his coins, he could verify the "proof" field. If it is valid, it
would mean someone has cracked his key with a quantum computer, and he
would need to use his strong announcement immediately to reclaim the funds
before the attacker does.
Best,
Boris
On Wednesday, June 4, 2025 at 2:40:53 PM UTC-3 Leo Wandersleb wrote:
Hi Boris,
the announcements, weak and strong would have to not be transactions yet to
be compatible with legacy nodes and thus keep it a soft-fork. They could be
OP_RETURN data. Only after the 144 blocks, the upgraded full nodes would
allow the inclusion of the actual transaction. This would mean the
transaction would be both in full in the OP_RETURN strong announcement and
without the witness part later, so it would be a bit expensive this way but
maybe we can do better?
A node that gets updated would have to re-index all the blockchain to find
announcements if we don't introduce a time frame for actually using the
announcements. We could also say that any announcement has to be used
within another 1000 blocks. Then the upgrading node would have to re-index
the last 1000 blocks.
The legitimate owner of a UTXO might wait for an attack for privacy
reasons. My proposal would allow Satoshi himself to make all his UTXOs
quantum safe without any of us learning about him being active. He could
add one 64B OP_RETURN in 2027 and when QC becomes an issue, we would learn
about him having been active in 2027 in 2040 when actually somebody tried
to attack and not in 2027 when people started to panic because of imminent
quantum breakthroughs.
Hmm ... a problem is the weak announcement doesn't require keys, so anybody
could provoke Satoshi to come forward. Maybe we have to add key ownership
as a requirement for the "weak" announcement, too. So it should also
contain a serialized transaction.
Best,
Leo
On Wednesday, 4 June 2025 at 04:15:59 UTC+2 Nagaev Boris wrote:
Hi Leo,
Thanks for the clarifications, much appreciated!
I have a couple of questions:
1. How is a weak announcement stored in the blockchain and in the UTXO set?
I assume it must be a transaction, correct? And it should somehow mark
the UTXO as planned to be spent for 144 blocks?
How would older (non-upgraded) nodes interpret a transaction
containing a weak announcement? Would they just skip over it without
any special processing?
If so, is there a problem for nodes that upgrade after the fork: would
they have to reprocess all blocks since the fork to find and index all
missed weak announcements?
2. In the case of reclaiming a UTXO after a weak announcement by an
attacker: why would the legitimate owner wait for a weak announcement
at all?
If the EC public key was already leaked, it seems they should publish
a strong announcement themselves rather than wait. If the EC public
key wasn't leaked, there's nothing to worry about even if someone
publishes a weak announcement: they are most likely bluffing, since
they wouldn't have the actual public key.
Best,
Boris
On Tue, Jun 3, 2025 at 3:29 PM Leo Wandersleb <lwand...@gmail•com> wrote:
>
> Hi conduition,
>
> Thanks for your careful analysis - excellent catches.
>
> You're absolutely right about the txid vulnerability. The commitment must
be to the complete transaction including witness data (wTXID or equivalent)
to prevent an attacker from pre-committing to unsigned transactions. This
is essential - otherwise an attacker could indeed enumerate the UTXO set
and create commitments without knowing the private keys.
>
> Regarding updates: You're correct that frequent updates would be needed
as wallets receive new UTXOs. However, I don't see this as a major issue -
users could batch their commitments periodically (say, monthly) rather than
after every transaction. The scheme is particularly important for existing
UTXOs that already have exposed pubkeys (old P2PK, reused addresses, etc.).
For new UTXOs, wallets should ideally migrate to quantum-safe addresses
once available. OpenTimestamps aggregation would indeed help with scaling
and provide plausible deniability about the number of UTXOs being
protected.
>
> The time delay serves a different purpose than you might expect. It's not
about preventing commitment forgery after pubkey exposure, but rather about
allowing priority based on commitment age when multiple parties claim the
same UTXO:
>
> 1. Weak announcement starts the 144-block window
> 2. During this window, anyone with a strong commitment can reveal it
> 3. The oldest valid commitment wins
>
> This creates the "poison pill" effect: an attacker might crack a key and
try to spend a UTXO, but if the original owner has an older commitment,
they can reclaim it during the window. The uncertainty about which UTXOs
have poison pills makes attacking large "lost" UTXOs risky - hence less
disruptive to the network.
>
> The delay essentially allows a "commitment priority contest" where age
determines the winner, protecting users who prepared early while still
allowing these users to not move their funds.
>
> Best,
>
> Leo
>
> --
> You received this message because you are subscribed to the Google Groups
"Bitcoin Development Mailing List" group.
> To unsubscribe from this group and stop receiving emails from it, send an
email to bitcoindev+...@googlegroups•com.
> To view this discussion visit
https://groups.google.com/d/msgid/bitcoindev/5e393f57-ac87-40fd-93ef-e1006accdb55n%40googlegroups.com.
--
Best regards,
Boris Nagaev
--
You received this message because you are subscribed to the Google Groups "Bitcoin Development Mailing List" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bitcoindev+unsubscribe@googlegroups•com.
To view this discussion visit https://groups.google.com/d/msgid/bitcoindev/2ed6d831-d87d-44f0-82b9-c8b9abaa8010n%40googlegroups.com.
[-- Attachment #1.2: Type: text/html, Size: 13421 bytes --]
next prev parent reply other threads:[~2025-06-05 14:59 UTC|newest]
Thread overview: 14+ messages / expand[flat|nested] mbox.gz Atom feed top
2025-06-02 21:06 Leo Wandersleb
2025-06-02 23:11 ` Nagaev Boris
2025-06-03 4:19 ` Leo Wandersleb
2025-06-03 11:51 ` Leo Wandersleb
2025-06-03 15:15 ` 'conduition' via Bitcoin Development Mailing List
2025-06-03 17:26 ` Leo Wandersleb
2025-06-03 19:49 ` Tim Ruffing
2025-06-04 17:14 ` Leo Wandersleb
2025-06-03 21:49 ` Nagaev Boris
2025-06-04 17:39 ` Leo Wandersleb
2025-06-04 18:38 ` Boris Nagaev
2025-06-05 8:18 ` Leo Wandersleb
2025-06-05 14:54 ` Boris Nagaev [this message]
2025-06-05 15:01 ` 'conduition' via Bitcoin Development Mailing List
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=2ed6d831-d87d-44f0-82b9-c8b9abaa8010n@googlegroups.com \
--to=bnagaev@gmail$(echo .)com \
--cc=bitcoindev@googlegroups.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox