The peg-out’s inputs, checked one by one against the BLAKE2b chain, September 7, 2026.
On September 6 at 14:28 UTC, on the SHA-256 chain, the Liquid federation’s multisig paid 4,019.44 BTC in one transaction with 83 inputs: 3,996 BTC to a fresh address, 16.8 BTC back to itself as change, and two smaller payments. Blockstream says the L-BTC that was redeemed had been minted through a bug in Elements, that no key was compromised, and that the sidechain is paused. The recipient left an OP_RETURN reading “we are whitehats. contact us on chain.” That is their story and it is still moving. This post is about one narrower question: what does that transaction look like from the chain we mine?
What our node says
We fetched the raw transaction from mempool.space and handed it to our Bitcoin Knots node, which follows the BLAKE2b chain, with testmempoolaccept. The call checks a transaction against the node’s rules and its view of the coins without broadcasting anything. The answer:
allowed: false
reject-reason: missing-inputsMissing inputs means the coins the transaction tries to spend are not in this chain’s set of unspent coins. Our first reading of that, and the first tweet, said the coins had already been spent here. That was wrong, and we deleted the tweet and reposted it. A coin that is not in the unspent set was either spent or never existed, and the node’s answer does not say which. So we checked the 83 inputs one at a time on mempool.guide, which indexes this chain.
Seventy-five of them were never created on this chain. The transactions that made them exist only on the SHA-256 chain. Eight exist here and are unspent, 340.7 BTC between them, created at heights 960,545 to 963,279. Five of those predate the proof-of-work change at 961,640, so both chains inherited them. Three were made after it, by federation transactions that were mined on both chains. The last of those landed here at 963,279 on September 1. After that, none of the federation’s transactions crossed.
The federation’s address on this chain still shows about 4,633 BTC unspent on mempool.guide. On this side, nothing moved.
Why nothing crossed
A transaction that spends coins from before the change is valid on both chains unless it opts out. That is the replay everyone was warned about. What the warning leaves out is that the transaction has to reach both chains first, and it does not do that by itself.
A Knots node on this chain fills its outbound connections with BLAKE2b peers and drops SHA-256 peers at their first header past the change. After that it hears nothing from the other network. Jason Sopko measured it on September 6 with a tool that peers with SHA-256 nodes and hands every transaction they announce to a Knots node: 12,157 transactions in ten minutes, 193 accepted, about one in fifty, the rest rejected as data traffic, missing inputs or policy. Nobody was running such a bridge for the federation’s traffic after September 1, so its later transactions were mined only over there. Each one spent change from the one before, so once the histories parted, every later federation transaction spent coins this chain had never seen. The drain is the last link of that chain.
What it means for you
If you spent pre-change coins on the SHA-256 chain since August 30 without the opt-in signature hash, the same transaction may or may not have landed here. It depends on whether anyone relayed it, not on anything you did. One Knots Discord user thought a public Electrum server had stolen his coins on September 4; his own spend had crossed and been mined here. The reverse also happens: a spend that never crossed leaves the coins sitting on this chain, spendable later.
The way to know is the way we knew: ask a node that follows this chain. gettxout on the output you care about says whether it is unspent here, and mempool.guide shows the transaction if it ever arrived. To repeat the check on the Liquid transaction:
curl -s https://mempool.space/api/tx/8db751a650ae2f12006b7e8c69a75e4df360e8afd6b9e05ae0b9fa6458a7b140/hex > tx.hex
bitcoin-cli testmempoolaccept "[\"$(cat tx.hex)\"]"What we are not saying
Nothing here is about the bug, who did it, or whether the 4,633 BTC on this side can be reached. That last one is a question for the federation’s signers, and it is their problem. Every number above is dated, and every one of them can be checked against a node you run yourself.
Saved you an hour? Tip the node and the clankers.