The First Transaction Between Two People, and Why It Is Verifiable
Most of this history rests on archives and memory, and a small part of it rests on a chain that anyone can read.
Almost everything on this site is written from archives. A forum thread that somebody chose to keep hosting. A mailing list post that survived because a volunteer mirrored it. A page that is one hosting decision away from being gone.
This piece is about the exception, and about what the exception can and cannot do.
The sequence, in order
3 January 2009. The genesis block. Its coinbase output cannot be spent by any node running the reference software, which is a detail with consequences and belongs to its own piece.
9 January 2009, 03:54:39 UTC. Block 9 is mined. Its coinbase transaction
0437cd7f8525ceed2324359c2d0ba26006d92d856a9c20fa0241106ee5a597c9 has a single
output of 5,000,000,000 satoshis, paying directly to a raw public key. There is
no address in it. Addresses were a later convenience.
Blocks 1 through 169. Every one of them holds exactly one transaction, the coinbase that pays the miner. Nothing is being sent to anyone. The network is running and nobody is using it for its stated purpose.
I did not take that on trust. I asked a public index for the transaction count of every block from 0 to 175 and looked for any block with more than one. Exactly one came back.
12 January 2009, 03:30:25 UTC. Block 170. Two transactions. The second,
f4184fc596403b9d638783cf57adfe4c75c605f6356fbc91338530e9831e9e16, spends the
block 9 coinbase. Its structure:
input 1 the block 9 coinbase output 5,000,000,000 sat
output 1 a different public key 1,000,000,000 sat
output 2 the same key that funded it 4,000,000,000 sat
fee 0 sat
size 275 bytesBoth outputs are pay-to-pubkey with OP_CHECKSIG. The fee is zero, which is not
generosity, it is the absence of a fee market. Nine days after the network
started, someone sent ten coins to someone else and kept forty as change.
That is the first transaction on this chain that moves value between two keys rather than creating it. It is nine days late, which is itself the finding: the system spent its first week and a half doing nothing but proving it could keep running.
Why this counts as a different kind of source
A forum thread is a claim held by a host. The chain is a claim held by everyone who keeps a copy, and it has two properties no archive has.
It cannot be quietly edited. Every block commits to the one before it by hash, so changing a byte in block 170 changes its hash, which invalidates block 171, and so on to the tip. There is no version of the edit that leaves the rest intact. Compare that with a forum post, where editing is a button and the only trace is a line of small text that the software may or may not display, and which the record of these years depends on entirely.
It does not go offline when a company does. I fetched the transaction above from one public index. Anyone can fetch it from a different one, or from their own node, and get identical bytes, because the bytes are what the network agreed on rather than what a publisher decided to serve. When a hosting bill goes unpaid the archive dies and the chain does not.
Neither property is magic and both are frequently oversold. The chain is not tamper proof, it is tamper evident and expensive to rewrite, which is a different and weaker and more useful claim.
What the chain does not record
It records public keys, amounts in satoshis, and an ordering. That is the complete list, and every one of the things people most want from it is absent.
No identities. The two outputs above are public keys. A public key is not a person, an account, a jurisdiction or an intention.
No prices. There is no field for one. The chain does not know what a satoshi is, has never been told, and has no mechanism for finding out. Every price attached to an early transaction was attached from outside, later, by someone with a spreadsheet, and pricing was its own separate invention.
No agreements. A transaction does not say what it was for. It shows a movement and stops. Whether it settled a debt, tested software, paid for something or moved value between two wallets belonging to one person is not in the data.
No authoritative time. Block 170’s timestamp is a claim made by whoever mined it. The rules only require that it be later than the median of the eleven preceding blocks and not too far ahead of a validating node’s own clock. That is a bound, not a certification. It is good enough to sequence events to within hours and it is not a notary’s stamp.
Where the names come from, and it is not the chain
The recipient of that ten coin output is named constantly. The chain does not name him. Testimony does, and the testimony is public, voluntary and signed with its author’s own name.
On 19 March 2013, in a forum thread titled “Bitcoin and me (Hal Finney)”, Hal Finney wrote:
When Satoshi announced the first release of the software, I grabbed it right away. I think I was the first person besides Satoshi to run bitcoin. I mined block 70-something, and I was the recipient of the first bitcoin transaction, when Satoshi sent ten coins to me as a test.
The announcement he is describing has its own piece here. The four year gap between the event and this account of it is the ordinary condition of the subject: the transaction is dated to the second, and the only sentence naming who was on the other end of it was written in 2013.
Set that beside the chain. The chain records an output of exactly 1,000,000,000 satoshis, which is ten coins, in the first non-coinbase transaction ever confirmed. The testimony and the record agree on the one thing the record contains.
That agreement is worth something and it is worth being precise about what. It is not chain evidence of identity, because there is no such thing. It is a person saying who he was, four years later, in a way that could have been contradicted by anyone who knew better, about a transaction whose amount and position anybody can check. The habit worth building is to hold those two categories apart in your head and to say which one you are using in any given sentence. Chain evidence for the transaction. Testimony for the name. Never one wearing the other’s clothes.
The structure above is one input and two outputs, with one output returning to the key that funded it. That is change, and change is what most address clustering rests on: the assumption that the returning output belongs to the sender, and the related assumption that inputs spent together share an owner.
Both are heuristics. They are often right and they are guesses, and a guess that is usually right is still a guess about a named human being. This site does not treat clustering output as identification, and it is worth noticing that the very first transaction on the chain is already exhibiting the pattern the technique would later be built on.
What was expanding, what was contracting
What expanded, from nothing, is the set of claims about this history that a stranger can check without asking anyone’s permission. Before 3 January 2009 there were no such claims in this subject at all. Every assertion about digital cash rested on somebody’s server, somebody’s memory, or somebody’s paper. From block 0 there is a growing body of statements that are true in the same way for everyone who looks.
What contracted, slowly and not by anyone’s decision, is the share of this history that lives only in archives. It never reached zero and it never will, because the chain does not record why. It fell, and the decline is why a piece about 2009 can be written with more confidence than a piece about 1997.
Who could tell at the time is the honest answer to the honest question: nobody, and they were not trying to. The nine people, or however many it was, running that software in January 2009 were debugging. Finney’s own account of the days after is “mostly me reporting bugs and him fixing them”. Nobody sent that transaction in order to create an unforgeable record of having sent it. The archival property was a by-product of a mechanism built for a different reason, which is the most common way durable infrastructure gets made.