[BIP-0054] 64-Byte Transactions and Potential Legitimate Uses

[jeremy]

For fun, let's start with a pop-quiz:

Select all that apply: There can exist a transaction of ___ bytes serialized size that BIP-0054's 64-byte restriction invalidates:

A) 64 Bytes
B) 0 Bytes
C) 1.5MB
D) 32 Bytes

E) 5MB


The answer is A, 64 Bytes, and -- perhaps surprisingly -- C, 1.5MB. 

Why is this the case?

BIP-0054 uses the term 64-byte transaction, but defines it as follows:

> Transactions whose witness-stripped serialized size is exactly 64 bytes are invalid.

In a [personally run] straw-poll of devs at a recent conference, no-one knew this precise edge condition or that the transactions could have a meaningful witness. For clarity, the restriction on bytes is on INVALID_TX_NONWITNESS_SIZE, not on the size with Witness.

Therefore, it is more accurate to refer to this in all sentences throughout the BIP as:

> transactions with exactly 64 bytes of non-witness data,

due to the propensity for confusion.

BIP-0054 also makes a comment that the transactions it invalidates are essentially useless:

> 64-byte transactions can only contain a scriptPubKey that lets anyone spend the funds, or one that burns them.

This is not strictly correct. Here are a few examples of current and future uses for 64-byte transactions:

Current Uses:

- A transaction that donates to a future miner from a segwit (any version) output via a spend to something like <512> OP_CSV (-> push2 bytes 512 csv -> 0x02 0x00 0x02 0xb2)

- That same output which is used as a connector output for things that should be claimed by a miner at a future time

- Pay-to-Anchor / ephemeral anchor outputs -- while typically p2a is for txns you want to add a subsidy ability, a 64-byte txn could be used to shim a keyed anchor to a p2a output after a certain delay.

Future Uses:

- Future work which might use output scripts for e.g. Transaction Sponsor encodings

- Future covenants work which encodes time-of-creation run scripts that e.g. quine an input; possibly in conjunction with sponsors

- Future where we have expensive reusable PQ or Contract public keys that are posted once and referred to by index

While, in a sense, current uses are much more concerning than future uses, with introspection opcodes, it might create substantive additional complexity to ensure that there is always a valid way to add a padding byte without upsetting a state machine.

As there are now documented use cases for 64-byte transactions that this proposal makes more difficult to do, I recommend replacing the text in the BIP that says

> 64-byte transactions can only contain a scriptPubKey that lets anyone spend the funds, or one that burns them.

With something like:

> There are documented use cases for 64-byte transactions that this proposal makes more difficult to cleanly do, but we do not believe these use cases will ever be valuable or worth protecting.

Or a more accurate reflection of the BIP-0054 authors' opinion.

Jeremy

[er...@voskuil.org]

Thanks Jeremey for this additional information. This exclusion is one of the reasons I originally pushed back, but I wasn't personally aware of any current use cases.

I would also suggest that the Rational section text in this area, while referencing my critiques in a footnote, doesn't capture the essence of them in the paragraph. It points out that I pushed back on importance, but excludes the reasons, which I consider essential in terms of making an informed decision. There is a referenced thread on Delving, and a related discussion on bitcoin-dev. I won't recount the details here, but I think the paragraph could more fairly represent the discussion, including the fact that the technical aspects were eventually agreed.

The TLDR is that:

(1) Merkle root malleation affects validation optimizations, not validation inherently.
(2) both forms of malleation can be mitigated by a node with no material performance hit (we do this).
(3) the material impact is to SPV wallets, as they must obtain the coinbase to mitigate.

This reference:

"It was suggested that the known vulnerabilities could instead be mitigated by committing to the Merkle tree depth in the header's version field"

Was added to the discussion by me, but is not the essence of my critique. It pertains to #3 and is not necessary for a node to mitigate malleation.

My pushback was that we are trading optimization implementation details for a consensus rule, and that the rule could create unforeseen problems by otherwise arbitrarily restricting the tx domain (which you have now pointed out below). I did not assume that everyone would see this modest SPV wallet benefit as worth the tradeoff. I am not personally taking a stand on that question, but I do think it could be presented more clearly.

Best,
Eric

> --
> 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
> <mailto:bitcoindev+...@googlegroups.com> .
> To view this discussion visit
> https://groups.google.com/d/msgid/bitcoindev/123e5545-2eda-4eca-9532-
> 4f4cea2b83ecn%40googlegroups.com
> <https://groups.google.com/d/msgid/bitcoindev/123e5545-2eda-4eca-
> 9532-
> 4f4cea2b83ecn%40googlegroups.com?utm_medium=email&utm_source=foo
> ter> .

[Anthony Towns]

On Fri, May 01, 2026 at 02:14:35PM -0700, jeremy wrote:
> *> 64-byte transactions can only contain a scriptPubKey that lets anyone
> spend the funds, or one that burns them.*

> This is not strictly correct. Here are a few examples of current and future
> uses for 64-byte transactions:
>

> *Current Uses:*

> - A transaction that donates to a future miner from a segwit (any version)
> output via a spend to something like <512> OP_CSV (-> push2 bytes 512 csv
> -> 0x02 0x00 0x02 0xb2)
> - That same output which is used as a connector output for things that
> should be claimed by a miner at a future time

This is an anyone can spend transaction. "Donation to a miner" and
"anyone can spend" are synonyms.

The script "OP_NOP <512> OP_CSV" would avoid the 64-byte special case
if anyone were to want to do this. Note that these are not standard
transactions and will generally not relay today.

> - Pay-to-Anchor / ephemeral anchor outputs -- while typically p2a is for
> txns you want to add a subsidy ability, a 64-byte txn could be used to shim
> a keyed anchor to a p2a output after a certain delay.

This only applies if you're donating an entire output to the miner, in
order to accelerate / increase the effective feerate of the transaction
that created that output. You would get a higher feerate by RBFing
the original transaction to delete the output -- that is, rather than
generating the transaction as "I'll have a feerate of X or X+Y, by
signing two transactions, one creating an output, and one spending that
output to miner fees (possibly with a timelock)", you do "I'll have a
feerate of X or X+Y+Z, by signing two transactions, one at at feerate X,
one at feerate X+Y+Z (possibly with a timelock)".

If there's some case where the two transaction approach does become
desirable, despite it's inefficiency, introducing an alternative 5-byte
P2A scriptPubKey would only be a policy change.

Cheers,
aj

[Chris Stewart]

I do concede your point that witness stripping isn't well understood by the general community. Here is a writeup I did for readers of the mailing list that may not understand what Jeremy is talking about the characteristics of a 64 byte transaction[0]. There is a big distinction between pre-segwit 64 byte transactions and segwit 64 byte transactions.

> A transaction that donates to a future miner from a segwit (any version) output via a spend to something like <512> OP_CSV (-> push2 bytes 512 csv -> 0x02 0x00 0x02 0xb2)

I find this confusing. Can you give 2 hex encoded transactions (funding transaction, spending transaction) that do this? regtest is fine, i can just decode locally.

Splitting hairs semantically if understand the transactions you are proposing above, "donating" to a "future miner" i.e. sending money to anyone (not a specific someone) in the future that can mine a block - doesn't seem much different to me than "anyone can spend the funds". I'll await your transactions though in the case that I am missing something.

[0] - https://delvingbitcoin.org/t/great-consensus-cleanup-revival/710/73?u=chris_stewart_5

--
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/123e5545-2eda-4eca-9532-4f4cea2b83ecn%40googlegroups.com.

[jeremy]

Chris,

These examples should work and could be adapted to a different funding input (I stubbed out the actual signature with just a preimage reveal for simplicity).

funding witnessScript: a820ff76f4f923cd475f27097b3822640a9b441894f3d4f742fe70e745e986da388e87

funding scriptPubKey: 002080c153db1ea3461a2b70126f853189fc34d7d9926f083dfb2bffae8ae2bb1130

funding txid: 166ef40e58f1a62e01d610023ca7935eef372cf3c1239d81430dffbd1fe8fe9e

funding tx hex: 02000000010000000000000000000000000000000000000000000000000000000000000000ffffffff020101ffffffff01a08601000000000022002080c153db1ea3461a2b70126f853189fc34d7d9926f083dfb2bffae8ae2bb113000000000

182 |

spending output scriptPubKey: 020002b2

spending stripped size: 64

spending full size: 114

spending txid: 44c752f499326a59a78259b455875d296108184b7f25b9a0b41bc018af7aedd8

spending wtxid: 1e6eb4de698f079cbaa0fdf47d7ab5e1bb4939fae42652f295dd05d15930074d

spending tx stripped hex: 02000000019efee81fbdff0d43819d23c1f32c37ef5e93a73c0210d6012ea6f1580ef46e160000000000ffffffff01905f01000000000004020002b200000000

spending tx full hex: 020000000001019efee81fbdff0d43819d23c1f32c37ef5e93a73c0210d6012ea6f1580ef46e160000000000ffffffff01905f01000000000004020002b2020a62697035342064656d6f23a820ff76f4f923cd475f27097b3822640a9b441894f3d4f742fe70e745e986da388e8700000000

$ decodrawtransaction $(cat funding_tx.hex)

{
  "txid": "166ef40e58f1a62e01d610023ca7935eef372cf3c1239d81430dffbd1fe8fe9e",
  "hash": "166ef40e58f1a62e01d610023ca7935eef372cf3c1239d81430dffbd1fe8fe9e",
  "version": 2,
  "size": 96,
  "vsize": 96,
  "weight": 384,
  "locktime": 0,
  "vin": [
    {
      "coinbase": "0101",
      "sequence": 4294967295
    }
  ],
  "vout": [
    {
      "value": 0.00100000,
      "n": 0,
      "scriptPubKey": {
        "asm": "0 80c153db1ea3461a2b70126f853189fc34d7d9926f083dfb2bffae8ae2bb1130",
        "desc": "addr(bcrt1qsrq48kc75drp52mszfhc2vvfls6d0kvjduyrm7etl7hg4c4mzycqmw604m)#2rrnsnlc",
        "hex": "002080c153db1ea3461a2b70126f853189fc34d7d9926f083dfb2bffae8ae2bb1130",
        "address": "bcrt1qsrq48kc75drp52mszfhc2vvfls6d0kvjduyrm7etl7hg4c4mzycqmw604m",
        "type": "witness_v0_scripthash"
      }
    }
  ]
}

$ decodrawtransaction $(cat spending_tx_stripped.hex)

{
  "txid": "44c752f499326a59a78259b455875d296108184b7f25b9a0b41bc018af7aedd8",
  "hash": "44c752f499326a59a78259b455875d296108184b7f25b9a0b41bc018af7aedd8",
  "version": 2,
  "size": 64,
  "vsize": 64,
  "weight": 256,
  "locktime": 0,
  "vin": [
    {
      "txid": "166ef40e58f1a62e01d610023ca7935eef372cf3c1239d81430dffbd1fe8fe9e",
      "vout": 0,
      "scriptSig": {
        "asm": "",
        "hex": ""
      },
      "sequence": 4294967295
    }
  ],
  "vout": [
    {
      "value": 0.00090000,
      "n": 0,
      "scriptPubKey": {
        "asm": "512 OP_CHECKSEQUENCEVERIFY",
        "desc": "raw(020002b2)#vceh5z0j",
        "hex": "020002b2",
        "type": "nonstandard"
      }
    }
  ]
}

$ decodrawtransaction $(cat spending_tx_full.hex)

{
  "txid": "44c752f499326a59a78259b455875d296108184b7f25b9a0b41bc018af7aedd8",
  "hash": "1e6eb4de698f079cbaa0fdf47d7ab5e1bb4939fae42652f295dd05d15930074d",
  "version": 2,
  "size": 114,
  "vsize": 77,
  "weight": 306,
  "locktime": 0,
  "vin": [
    {
      "txid": "166ef40e58f1a62e01d610023ca7935eef372cf3c1239d81430dffbd1fe8fe9e",
      "vout": 0,
      "scriptSig": {
        "asm": "",
        "hex": ""
      },
      "txinwitness": [
        "62697035342064656d6f",
        "a820ff76f4f923cd475f27097b3822640a9b441894f3d4f742fe70e745e986da388e87"
      ],
      "sequence": 4294967295
    }
  ],
  "vout": [
    {
      "value": 0.00090000,
      "n": 0,
      "scriptPubKey": {
        "asm": "512 OP_CHECKSEQUENCEVERIFY",
        "desc": "raw(020002b2)#vceh5z0j",
        "hex": "020002b2",
        "type": "nonstandard"
      }
    }
  ]
}

```python3

#!/usr/bin/env python3
from __future__ import annotations

import hashlib
import struct


OP_SHA256 = 0xA8
OP_EQUAL = 0x87
OP_CHECKSEQUENCEVERIFY = 0xB2


def sha256(b: bytes) -> bytes:
    return hashlib.sha256(b).digest()


def hash256(b: bytes) -> bytes:
    return hashlib.sha256(hashlib.sha256(b).digest()).digest()


def hash256_be_hex(b: bytes) -> str:
    return hash256(b)[::-1].hex()


def compact_size(n: int) -> bytes:
    if n < 0:
        raise ValueError("negative compact size")
    if n < 253:
        return bytes([n])
    if n <= 0xFFFF:
        return b"\xfd" + struct.pack("<H", n)
    if n <= 0xFFFFFFFF:
        return b"\xfe" + struct.pack("<I", n)
    return b"\xff" + struct.pack("<Q", n)


def i32(n: int) -> bytes:
    return struct.pack("<i", n)


def u32(n: int) -> bytes:
    return struct.pack("<I", n)


def u64(n: int) -> bytes:
    return struct.pack("<Q", n)


def ser_output(value_sat: int, script_pubkey: bytes) -> bytes:
    return u64(value_sat) + compact_size(len(script_pubkey)) + script_pubkey


def ser_input(prev_txid_be_hex: str, vout: int, script_sig: bytes = b"", sequence: int = 0xFFFFFFFF) -> bytes:
    # Transaction outpoints serialize txids little-endian.
    prev_txid_le = bytes.fromhex(prev_txid_be_hex)[::-1]
    return (
        prev_txid_le
        + u32(vout)
        + compact_size(len(script_sig))
        + script_sig
        + u32(sequence)
    )


# --------------------------------------------------------------------
# Funding transaction
# --------------------------------------------------------------------
#
# This is a coinbase-like funding transaction for a decode/test vector.
# It pays 100_000 sats to a native P2WSH output.
#
# witnessScript:
#
#   OP_SHA256 <sha256("bip54 demo")> OP_EQUAL
#
# The spending transaction must reveal the preimage "bip54 demo" in witness.

secret = b"bip54 demo"
secret_hash = sha256(secret)

witness_script = (
    bytes([OP_SHA256])
    + bytes([0x20])       # push 32 bytes
    + secret_hash
    + bytes([OP_EQUAL])
)

# Native v0 P2WSH scriptPubKey:
#
#   OP_0 <sha256(witness_script)>
#
funding_script_pubkey = b"\x00\x20" + sha256(witness_script)

coinbase_script_sig = b"\x01\x01"  # BIP34-style height=1 push; enough for a decode vector.

funding_tx = (
    i32(2)
    + compact_size(1)
    + bytes(32)                  # coinbase prevout txid = 0x00..00
    + u32(0xFFFFFFFF)            # coinbase prevout index
    + compact_size(len(coinbase_script_sig))
    + coinbase_script_sig
    + u32(0xFFFFFFFF)
    + compact_size(1)
    + ser_output(100_000, funding_script_pubkey)
    + u32(0)
)

funding_txid = hash256_be_hex(funding_tx)


# --------------------------------------------------------------------
# Spending transaction
# --------------------------------------------------------------------
#
# This spends funding_tx output 0.
#
# The sole output scriptPubKey is:
#
#   <512> OP_CHECKSEQUENCEVERIFY
#
# Minimal ScriptNum encoding for decimal 512 is little-endian 00 02,
# so the full script is:
#
#   02 00 02 b2
#
# This spending transaction has:
#
#   witness-stripped size = exactly 64 bytes
#   full witness serialization = larger than 64 bytes

csv512_script_pubkey = bytes([0x02, 0x00, 0x02, OP_CHECKSEQUENCEVERIFY])

spending_input = ser_input(
    prev_txid_be_hex=funding_txid,
    vout=0,
    script_sig=b"",
    sequence=0xFFFFFFFF,
)

spending_output = ser_output(90_000, csv512_script_pubkey)

spending_tx_stripped = (
    i32(2)
    + compact_size(1)
    + spending_input
    + compact_size(1)
    + spending_output
    + u32(0)
)

# P2WSH witness stack:
#
#   <secret> <witness_script>
#
spending_witness = (
    compact_size(2)
    + compact_size(len(secret))
    + secret
    + compact_size(len(witness_script))
    + witness_script
)

spending_tx_full = (
    i32(2)
    + b"\x00\x01"               # SegWit marker + flag
    + compact_size(1)
    + spending_input
    + compact_size(1)
    + spending_output
    + spending_witness
    + u32(0)
)

assert len(spending_tx_stripped) == 64
assert csv512_script_pubkey.hex() == "020002b2"

print("funding witnessScript:", witness_script.hex())
print("funding scriptPubKey: ", funding_script_pubkey.hex())
print("funding txid:         ", funding_txid)
print("funding tx hex:       ", funding_tx.hex())
print()
print("spending output scriptPubKey:", csv512_script_pubkey.hex())
print("spending stripped size:      ", len(spending_tx_stripped))
print("spending full size:          ", len(spending_tx_full))
print("spending txid:               ", hash256_be_hex(spending_tx_stripped))
print("spending wtxid:              ", hash256_be_hex(spending_tx_full))
print("spending tx stripped hex:    ", spending_tx_stripped.hex())
print("spending tx full hex:        ", spending_tx_full.hex())

```

[Antoine Poinsot]

Hi Jeremy,

Thanks for taking the time to lay out your objection here. I’m glad we now have
a clear statement of it.

As AJ points out, your examples are all anyone-can-spend, so BIP 54 is correct
here. But it's fair that it could be clearer with regard to witness-stripped
serialized size. Thanks for pointing this, i'll push an update.

What's more important is your claim that invalidating 64-byte transactions
could complicate working with advanced scripting features that may eventually
be added to Bitcoin. If that's the case i agree it should *at least* be
discussed in BIP 54's rationale section.

However, every time this was brought up, no one could come up with an example
that barely resembles today's Bitcoin. Of course one can come up with
hypothetical constructions of a very different Bitcoin in which a 64-byte
transaction could be useful. But that is not the goal. The goal is to find a
script for a 64-byte transaction that is *plausible* Bitcoin would adopt in the
future.

And as long as we are going to have a model where coins commit to their
spending conditions, a 64-byte transaction is always either burning the spent
coin's value or letting anyone spend it, regardless of the introspection or
covenanty features that get added to these spending conditions.

Therefore i do not currently see any plausible Bitcoin scripting upgrade that
not only could not be designed to accommodate the 64-byte witness-stripped size
exception, but also whose usage could even result in a 64-byte transaction in
the first place.

Antoine

On Friday, May 1st, 2026 at 5:15 PM, jeremy <jeremy....@gmail.com> wrote:

> For fun, let's start with a pop-quiz:
>
> Select all that apply: There can exist a transaction of ___ bytes serialized size that BIP-0054's 64-byte restriction invalidates:
>
> A) 64 Bytes
> B) 0 Bytes
> C) 1.5MB

> D) 32 BytesE) 5MB

[jeremy]

I haven't made an objection, I've raised an issue and made a request that the BIP be precise about what it is doing, which is what is generally expected of BIPs anyway.

You've extrapolated that this is an objection, but it's not. I, or someone else, may object later, but I at least wanted to do the groundwork of ensuring that we, the community at large, are on the same page and discussing the same set of issues. Irrespective of such a future discussion, I think these clarifications materially improve the BIP

On the similarity of anyone-can-spend-now and anyone-can-spend-later, it is not true that OP_TRUE,  52416 CSV, and 1 CSV are all equivalent.

They may be equivalent from the point of view of authorization "once spendable, no specific key is required"  but they are certainly not equivalent from a miner-incentive point of view.

A miner has incentive to include a transaction creating an OP_TRUE output and immediately (assuming sufficient value) spend it to themselves.

Inclusion of 1 CSV outputs in a block can incentivize a future miner to build on top of that block as "fee forwarding" (and, in a sense, a user may have incentive to generate 1 CSV, 2 CSV, ... outputs if they are doing a large transaction with a large fee during a low fee period where they want to reduce fee-sniping reorg instability).

A protocol can use delays like 52416 CSV for punishments where collusion with miners is undesirable, since the time of punishment and time of redeem are far spread.

I agree these can all be called anyone-can-spend in a narrow authorization sense. But they are not equivalent economically. The timing changes who can capture the value, when they can capture it, and what incentives are created for inclusion, reorgs, chain extension, and miner collusion.

Economic distinctions matter. We should not treat anyone-can-spend now and anyone-can-spend later as the same thing.

[Antoine Poinsot]

Hi Eric,

Fair enough. The BIP rationale should discuss explicitly your point that it introduces a "seam", instead of just mentioning it in a footnote. Also i clarified how the full node consensus failure point is not the motivation for 64-byte txs invalidations since it can be better addressed differently, as you previously pointed.

Changes here: https://github.com/bitcoin/bips/pull/2159. Also took Jeremy's feedback that "64-byte transactions" should spell out that it refers to witness-stripped serialized in more places than only the specifications.

Best,
Antoine

-------- Original Message --------

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/00a501dcd9b6%2459406560%240bc13020%24%40voskuil.org.