A Cryptography Q&A Is Unfolding on Bitcoin, One Block at a Time

Most of what gets written into Bitcoin is not a conversation. It is a payment, a bridge memo, a swap route, an ordinal, a token tag. Machines talking to machines, or people leaving a single mark and walking away. So when a scan of the last day of blocks surfaces something that reads like two people passing notes across the ledger — a question, a pause, an answer, a follow-up question — it is worth stopping on.

On August 23, 2026, over roughly four hours, an exchange about the internals of a Bitcoin witness script is being conducted entirely in OP_RETURN outputs. One participant asks. Blocks pass. Another answers. The asker sharpens the question. It is happening in public, on the most expensive message board ever built, and nobody involved appears to know or care that a census scanner is reading over their shoulder.

The thread, block by block

Here is the exchange exactly as it sits on-chain, in block order. Each line is a distinct OP_RETURN output in the block listed. Times are UTC, from the block timestamps.

Read it top to bottom and the rhythm is unmistakable. The question goes out in block 963,738 and is repeated a block later, the way you would resend a message you were not sure landed. About an hour passes. In block 963,744 the answer arrives, flat and complete: the witness script is a multisig. Half an hour after that, the asker comes back with the real question — not what is it but show me the parameters: how many keys, what threshold, and how are the keys derived from genesis? That question is posted twice, in 963,746 and again two hours later in 963,758, and so far it has drawn no reply.

We do not know who these parties are. The outputs carry no signature, no handle, no pool tag, no token marker — just the plain text, paid for at ordinary Bitcoin fee rates and made permanent. They could be two people. They could be one person talking to themselves, or a script, or a puzzle with a prize behind it. The chain does not say. What the chain does say is that the exchange is real, timestamped, and ordered, and that anyone who reads blocks can watch it develop.

Why the last question is the interesting one

The first round is small talk. Hash lock, multisig, or something else is a taxonomy question with three answers, and it got the boring one. But the follow-up is not small talk. How many keys, what threshold, and how are the keys derived from genesis is asking for the actual construction of a spending policy: the m and the n in an m-of-n multisig, and — the strange part — a claim that the public keys are derived from genesis.

Keys derived from genesis is not standard wallet behavior. Ordinary multisig keys come from independent seeds. "Derived from genesis" suggests something deterministic and shared: keys generated from a fixed starting point that any party could reconstruct, the way a puzzle or a proof-of-concept would build a wallet whose signers are defined by a rule rather than by secrets. That is the detail that turns this from idle chatter into something that might have a point — a challenge, a demonstration, a breadcrumb. And it is precisely the detail that is currently going unanswered.

We are not going to pretend to know the answer. We are flagging that the question is live, that it is well-formed, and that whoever asked it chose to ask it on Bitcoin rather than anywhere cheaper. That choice is the story.

How we caught it, and why we are watching

This surfaced from a chain-census pass: a scanner that walks recent blocks and pulls out the human-readable content — OP_RETURN text, witness inscriptions, coinbase messages — from the wall of machine noise around it. Most of what it finds is exactly that noise: USDT bridge routes, swap tags, RSK merge-mining headers. Occasionally it finds a love letter, a memorial, a manifesto. Rarely, it finds a conversation in progress.

The reason we care is not idle. PowForge exists because interaction on open networks has a cost, and Bitcoin is the clearest ledger of what people will pay to make a thing permanent. A two-party technical Q&A conducted in OP_RETURN, when the same exchange could happen for free in any chat app, is a small, concrete instance of the thing our whole thesis is about: some messages are worth paying blockspace to make unforgeable and unerasable. Watching one happen in real time is better data than any argument we could write.

So we are going to keep watching this one. The last question is open. If block 963,758's query ever gets its answer — a key count, a threshold, a genesis derivation rule — it will show up in a future scan, and we will follow it here. Near-live coverage of interesting block data is not a side effect of running a census scanner. It is the feature.

Update — the thread kept going

Appended 2026-08-24, from a follow-up census scan · blocks 963,829–963,888

When we published, block 963,758's question — how many keys, what threshold, how are they derived from genesis — was still hanging. It did not stay that way. Over the next day and a half the same thread kept advancing, one block at a time, and it turned out to be more disciplined than the first afternoon suggested. Here is what came after.

The answer we were waiting for lands in block 963,829: both keys use the same Genesis field, and there is no hash. That settles two things at once. There are two keys, and the derivation is not run through a hash function — the keys come straight off a field of the Genesis block. It is the first substantive answer since the multisig confirmation back in 963,744, and it confirms the suspicion from the original post: this is a deterministic, reconstructible key set, the shape of a puzzle rather than a personal wallet.

But the asker is not done, and this is where the exchange earns the word rigorous. In block 963,837 comes a sharper question: is the second key derived from the first, or both from genesis independently? It is the difference between a chain (key two is a function of key one) and two parallel derivations from the same source. Block 963,868 answers it cleanly: both keys are derived independently from Genesis. Not chained. Two separate derivations off the same Genesis field.

And then, in block 963,888, the thread drills one level deeper still: which genesis field — hash, merkle, nonce, time, headline, or pubkey? Independence is settled; now the question is which of the Genesis block's fields seeds the two keys. That question is still open as of block 963,906. The Genesis block has a handful of candidate fields, and the asker has just enumerated them. Whatever answers next names one.

What strikes us on the second pass is the shape of it. This is not two people trading trivia — it is a proper narrowing, each answer closing exactly one degree of freedom and each new question opening the next one down. Multisig, not hash lock. Two keys, same field, no hash. Independent, not chained. Which field. Somebody is either running a genuine specification handshake in the most expensive medium available, or building a puzzle whose solution is a wallet you can reconstruct from the Genesis block itself. Either way it is still going, and we are still reading.

Update — the answer was a refusal

Appended 2026-08-25, from a follow-up census scan · block 963,910

Block 963,888 had left one question hanging: which Genesis field — hash, merkle, nonce, time, headline, or pubkey — seeds the two keys. We expected the next line in the thread to name one. It did not. In block 963,910, about three and a half hours later, the reply came back:

That is not an answer. It is a refusal to give one — and it changes what the whole thread has been. Up to this point it read like a specification handshake: two parties nailing down the exact construction of a spending policy, one degree of freedom at a time. The last reply reframes it as a puzzle. The Genesis block's fields are all public and enumerable, the responder points out; deciding which one matters is the reader's work, not theirs. What looked like a Q&A was, at least in part, a set of clues being laid down in public, with the final step left deliberately blank.

So the storyline stays open, but for a different reason than before. We are no longer waiting for a participant to name a field. We are watching to see whether anyone — a party to the thread, or a stranger reading over the same shoulder we are — posts a candidate: a reconstructed key, a wallet, a solution. As of the tip at block 963,958, nobody has. The last word on-chain is the dare.