The story is always told the same way round: a man threw away a fortune on dinner. Open the thread and the sentence that stops you is his, four days in, asking whether ten thousand bitcoins was too little to get a pizza delivered.

Nobody would take it. That is the part worth sitting with.

So this article goes to the sources: the original posts in a web archive, and the payment itself in the blockchain, read twice on the same day.

The man in Jacksonville

It was just past midnight on 18 May 2010. A man in Jacksonville, Florida posted a message on a small forum where a few hundred people were arguing about a kind of money that had existed for barely a year.

I’ll pay 10,000 bitcoins for a couple of pizzas.. like maybe 2 large ones so I have some left over for the next day.

He went on about toppings for a while. Onions, peppers, sausage, mushrooms, nothing with fish on it. A plain cheese pizza would be fine too, he added, since it might be cheaper to make. What he really wanted was food arriving at the house without him having to organise any of it, the way a hotel sends up breakfast.

Someone asked which country he lived in. He answered with his city and his zip code, and that one line tells you the size of the room. Handing your postcode to strangers to arrange dinner was not a strange thing to do there.

It was not a strange time for it either. No shop anywhere accepted the coin. The places that would swap it for money were barely starting to exist. Using it meant finding a person on a forum and trusting them.

The forum itself was only months old. Look at the accounts that replied to him and their post counts run in tens and hundreds. Nobody in that thread imagined it would be quoted around the world later. To them it was a man trying to get dinner delivered.

And he was not a bystander. He worked on the software itself, and he wanted to know whether the thing he had been working on did anything outside the screen. What he was really testing was not the program. It was whether a stranger on the other side would play along.

Four days

Then nothing happened.

A day went by, then another. No pizza.

On the evening the offer went up, an account with ten posts to its name replied in passing. Ten thousand was quite a bit, it said, and you could sell them on a bitcoin market site for $41 right now. Then a small dig at the end: good luck getting your free pizza.

Hold on to that line. The $41 that follows this story everywhere today came from exactly there. It was not an official rate, nobody traded at it, and the person writing it was guessing in passing, four days before any pizza existed.

A reader in Europe offered to help but could not work out how to pay an American pizza place from across the Atlantic. Would they take a foreign card over the phone? The man answered patiently that plenty of places took orders online by then. The conversation stopped there.

Somebody else left a one-line joke about knowing any pizza-eating hairdressers. It was meant as a joke, and that was genuinely the mood of the thread. Nobody took the offer seriously.

Two more days passed. On a forum, four days is a long time. The thread would already have been sliding down the page.

On the evening of 21 May he came back to his own post and wrote this:

So nobody wants to buy me pizza? Is the bitcoin amount I'm offering too low?

Read it again. He was worried that ten thousand bitcoins were not enough to interest anybody in ordering him dinner. At the price on my screen as I write this, the offer he was apologising for was worth $762,034,300. It sat there for four days and nobody wanted it.

Two hours later he explained himself in a single line, and this is the part that usually gets cut out of the retellings:

I just think it would be interesting if I could say that I paid for a pizza in bitcoins

He was not trying to make money and he was not trying to lose any. He wanted to be able to say the sentence. Does this thing buy dinner or not. That is the kind of question you can only settle by doing it, and doing it required somebody else to say yes.

That evening

On the evening of 22 May he posted again.

I just want to report that I successfully traded 10,000 bitcoins for pizza.

There was a link to photographs. He thanked the person who had arranged it by their forum name. Three hours later one of the early developers replied briefly: congratulations, a milestone reached. The thread moved on to other things.

Nobody in that room thought the date would be remembered every year afterwards. A report that two pizzas had arrived, and one line of congratulation. There was no anniversary, no name for it, no hashtag. All of that got attached over the following sixteen years.

The man himself said nothing more about it in the thread. He had done the thing, he had reported it, and that was that.

Meanwhile, somewhere nobody was looking, one thing had happened. The payment went into the ledger at 2010-05-22 18:16:31 UTC, in block 57,043. Since then 910,360 blocks have been stacked on top of it, and the record has stayed underneath exactly as written.

That is the half we can check rather than repeat.

Sixteen years later, I opened the transaction

So I opened it. You do not need an account and you do not need to own anything. Paste the transaction id into a block explorer and you get the same screen from anywhere in the world. I read it twice on the same day to be sure the values repeated. Here is the transaction.

What comes up is unremarkable. A time, a block number, a list of what went in and a list of what came out. It is simpler than a bank statement.

A block is a bundle of transactions wrapped up together. A new bundle closes roughly every ten minutes, and once it closes the contents do not change. That is why this statement has not moved by a character in sixteen years.

Start with what went out. There is a single output: exactly 10,000 coins, to one address. No change coming back, no second recipient, nothing split. A round number leaving a wallet in one piece.

It was the input side that kept me there. The payment was assembled out of 131 separate pieces, and every one of them came from the same address.

Coins here are not a balance in an account. A wallet holds a pile of separate chunks, each one left over from some earlier payment. To send an amount you gather up enough chunks to cover it. It is closer to paying with the notes and coins in your pocket than to a bank transfer.

So the shape of the input side shows you what was in that pocket. The largest piece was 3,753.88 coins. After it a run of big round pieces. And at the end 111 pieces of 0.01 coins each, which come to 1.11 coins between them.

Those hundred and eleven scraps are what I keep coming back to. Somebody went through the corners of a wallet to land on a round number. He did that to order dinner.

It also made the earlier part of the story read differently. The picture usually handed down is a man tossing away ten thousand coins as though they were nothing. The shape in the ledger is not that. It is somebody counting out pieces to reach a number he had promised.

And there is one more thing nobody mentions. Add up the inputs and you get 10,000.99 coins. The output is 10,000. The difference, 0.99 coins, was left on the table for whoever mined the block.

That is how fees work here. There is no field where you type one. Whatever you put in and do not send out becomes the fee, and in 2010 the software was relaxed about leftovers because they were worth nothing to anybody. At the price when I checked, that leftover is $75,441.

The other thing worth knowing is that the fee has almost nothing to do with the amount being sent. It is priced by the room the transaction takes up in a block. This one ran to 23,620 bytes, which is bulky, because describing 131 inputs takes space. The same rule is still running today, and it is why network fees behave the way they do.

One last thing was waiting at the end. The ten thousand coins landed in that address and then left again in the very next block, about 9.6 minutes after they arrived. They went out in two pieces, 5,777 coins and 4,223 coins, and one of the addresses they went to emptied itself completely the same evening.

It does not tell you who anyone was. An address is not a person, and the ledger only shows that coins moved between addresses. The timing still says something. Nobody treated these coins as a trophy. They arrived, they were broken into working pieces within ten minutes, and they carried on moving.

The wallet that paid puts the scale in perspective. Over its lifetime that address took in 81,432 coins across 3,355 transactions, and today it holds dust. This was not somebody’s life savings carefully set aside. It was a wallet in heavy use.

If you want to see it for yourself, now is as good a time as any. Open a block explorer, click the transaction linked above, then click the receiving address. You will get the screen I got. Any explorer will do, since they are all reading the same ledger.

The address never died

I assumed the address that received the coins was a dead end. I opened it to confirm that and move on.

It was not.

Money has arrived there 15 separate times since that day in 2010. Not much money: 0.00379983 coins in total. But it keeps arriving, spread across 7 different years.

The dates are where it turns. 4 of those arrivals landed on 22 May, the anniversary of the payment.

Two of them came in on the same day less than an hour apart, and they are the two largest gifts in the list. Whether those two knew each other, or each remembered the date on their own, is not written in the ledger. What is written is that on the same afternoon two people thought of the same address.

Look at the satoshi column and 6 of those payments are for exactly 546 satoshis.

A satoshi is the smallest unit the system can express, one hundred millionth of a coin. 546 is not a number anybody picks by accident. It is the floor below which the network treats an output as dust and refuses to pass it along. Six different people sent the smallest amount it is physically possible to send.

Nobody is enriching anybody. They are signing a guest book. Somebody worked out the smallest gesture the network permits and made it, at the address where the first real purchase landed. There is no plaque anywhere. The address is the plaque, and anyone in the world can add a name to it.

Read the table by year and something else appears. The arrivals in 2015, 2017, 2018 and 2020 fall on scattered dates: April, July, October, February. People knew about the address back then and sent to it, but nobody was watching the calendar. Arrivals on 22 May itself begin in 2023, and since then they turn up on the day every year. Nobody announced it. The dates collected themselves.

I have to be honest about one thing here, because it would be easy to make the story prettier than the evidence allows. Not every arrival is a tribute. Some people send tiny amounts to well-known addresses in bulk, hundreds at a time, for reasons that have nothing to do with sentiment. Counting those as pilgrims would be inflating my own story.

So I drew a line I can state plainly. A transaction paying ten or more addresses at once counts as a broadcast rather than a gift.

By that rule 4 of the 15 arrivals were broadcasts. Two of them paid more than a hundred addresses in a single transaction, one of them 201. The remaining 11 look deliberate, and they come to 0.0037732 coins between them.

It is a guess dressed up as arithmetic and it has soft edges. Somebody sincere might use a wallet that batches payments, and a bulk sender aiming at one address at a time would walk through my filter. The chain records amounts and addresses, never motives. What survives is the calendar. A dust broadcast does not know what day it is.

What hardened in the retelling

This story is repeated thousands of times a year, and each version copies the one before it. Facts harden while they travel. Reading the original, I found several that had set.

The one that set hardest is the $41. As you saw, it came from an account with ten posts, guessing in passing, four days before any pizza arrived.

I went looking for the actual May 2010 rate as well. A hand-run service published a price for the coin back then, and its old pages are preserved in a web archive. The explanatory text is still there and so are the instructions for trading with the owner. The rate table itself did not survive in the copy I could read. So this article does not say what the coins were worth then.

The pizza chain is the same kind of problem. Plenty of articles name a brand, and a brand does appear in the thread. The context is different: the buyer was explaining that many pizza places took orders online by then, and named one his household used. That is not a receipt. I do not know what arrived at that door.

The addresses get swapped around too. He carried a donation address in his signature under every post, and that address is not the one that funded the payment. Compare them and the histories differ. The mistake is easy to make, because the signature address is the one printed inside the quoted text, so it is the one people copy across.

You will also read that the ten thousand coins are still sitting in that address. They are not. They left ten minutes after arriving. What is there today is the small change people have sent since.

And then there is the framing that calls this the worst trade in history. That is the one I like least, and not only because it is unkind. It reverses the order of events.

The coins did not have a real price that he squandered. He went to find out whether they could have a price at all, and it took him four days of failing to give them away. Calling the result a blunder requires you to already know the answer to the question he was asking.

What it comes to, and what it leaves

Now for the part with a shelf life. Everything above is fixed in the ledger and will read the same in ten years. The numbers below were true at the moment I checked them and are already slightly out of date.

Prices came from 4 public sources at the same moment, on 2026-09-17 11:38 UTC, and the ledger was queried twice the same day so the values could be compared.

I took the price from 4 public sources at the same moment. Two of them quote real dollars and two quote a dollar-linked token, and the four disagreed with each other by 0.096 percent. There is no single official price for this asset, which is a thing worth understanding on its own.

I am not going to do the arithmetic about what those coins would be worth if he had kept them. It is one multiplication and I have just handed you both numbers.

What that multiplication hides is that the coins only came to be worth anything later because things like this happened first. In May 2010 there was no price worth the name, because there was nothing to buy. A currency that cannot buy dinner is a spreadsheet. Somebody had to go first, and going first meant handing over an amount that looks absurd once the thing has a price.

The four days of silence measure that better than any chart from the period. Ten thousand coins could not get a pizza delivered. That is what the asset was worth in practice, and nothing states it as plainly as a man asking a forum whether he ought to offer more.

Reading a story is a poor reason to change what you do with your money. Three things in this one are practical, though.

You will end up dealing in satoshis whether or not you notice. One coin is a hundred million of them, and the 546 satoshi floor in this article is a real limit. You walk into it when a tiny leftover refuses to move.

Fees are charged by size rather than amount. A payment stitched together from many small pieces costs more to send than one clean one, which is also why moving coins between exchanges costs what it does.

And the address is the whole of the instruction. The 2010 payment went where the address said, in ten minutes, with nobody to call. That is why the wrong address or the wrong network is the mistake worth checking twice.

One more thing. Nothing in this article was seen with special access. The transaction, the address history, sixteen years of arrival dates are all open. Nobody has to let you in. That is probably half the reason the story stays alive: everyone could just copy the retelling, and yet the original is right there to check.

I used to file this under expensive mistakes. After reading the thread and the ledger I file it under experiments, and it changed the question I ask about any coin I come across.

Not what it is worth. What it can do, and who accepted it doing that recently. In May 2010 the honest answer for bitcoin was almost nothing, and one man proved it could do exactly one thing, once, with a stranger, for dinner. Everything after that is a footnote.

If you are starting out, the dull parts matter more than the story does: how to actually buy the thing, where it lives once you own it, and when a dedicated device is worth the trouble.

Affiliate disclosure: some links are partner links. We may earn a commission at no extra cost to you. This is not investment advice.

Frequently asked questions

Next: what a network fee is actually paying for