🥷

The Patoshi Pattern

6 min read

history educational
Share: X Bluesky Mastodon Reddit HN

In 2013, an Argentinian researcher named Sergio Demian Lerner published an analysis claiming that a single mining entity had produced roughly 40% of all blocks in Bitcoin's first fifteen months (blocks 0 to 50,000), accumulating about 1 million BTC (980,000 by his 2013 count; his 2019 recount put it near 1.1 million). The fingerprint, he argued, was hidden in the coinbase scriptSig of every block that single miner produced. Almost certainly Satoshi Nakamoto. Almost certainly never spent.

The pattern, called Patoshi after the name Lerner gave the miner in an April 2019 post, lives on chain in plain sight. Anyone with a node and an afternoon can verify it.

What's a coinbase scriptSig?

A coinbase transaction is the special tx that mints the block subsidy. Its single "input" doesn't reference a prior output (there's no prior tx for the first transaction in a block); instead, the input has a scriptSig field that miners can fill with arbitrary bytes. Modern miners stuff in pool tags and ASCII messages. Early miners were less self-aware; they often left the bytes in their default form, which exposed implementation details.

Pull the genesis block's coinbase to see this in action:

Open in playground → see the famous Times-headline scriptSig

The full scriptSig: 04ffff001d0104455468652054696d6573.... The leading bytes are two small pushes, the difficulty bits (ffff001d) and the number 4 (the slot later blocks use for the ExtraNonce; block heights only became a required part of the coinbase when BIP34 was enforced in March 2013), then the actual data payload.

What Lerner found

Lerner pulled the coinbase scriptSig of every block from 1 to ~50,000 and looked at one specific field: the ExtraNonce. The ExtraNonce is a counter the miner increments when the regular nonce range is exhausted; it's part of the data the SHA256 hash function operates on.

The reference Bitcoin Core software increments ExtraNonce in a specific way. But Lerner noticed that about 40% of all early blocks had ExtraNonce values drawn from a dramatically different distribution. The pattern: in the original client the ExtraNonce was a running counter that kept climbing from one block to the next and only started over when the software restarted. Plot it against block height and each miner's blocks trace their own rising line. One miner's lines were far steeper and denser than anyone else's, climbing in long, steady runs that no other machine on the network matched.

The other roughly 60% of blocks showed normal incremented ExtraNonces, suggesting many different miners with normal hashpower distributions.

See it for yourself

Pull a few early blocks and look at the coinbase scriptSig bytes. The ExtraNonce sits in a predictable position. Block 1:

Open in playground → first non-genesis block
Open in playground → full transactions inside, view the coinbase scriptSig

Now block 100:

Open in playground → block 100, January 11 2009

And block 50,000, shortly before the Patoshi pattern stops:

Open in playground → late Patoshi era, April 10 2010

If you fetch each block at verbose=2 and decode the coinbase scriptSigs, you can manually trace the ExtraNonce pattern Lerner used. The original analysis and its later updates are public (see Sources below), but the data they rely on is right there on chain.

The Patoshi pattern stops

Around block 54,000, the pattern abruptly disappears. After that point, no blocks show the steep-ExtraNonce signature. The mainstream interpretation: Satoshi stopped mining around May 2010, possibly to avoid a 51% problem as more miners came online, possibly because they were ready to step away from active maintenance.

About a year later, in April 2011, Satoshi sent his last known emails and disappeared. The roughly 1.1 million BTC mined under the Patoshi fingerprint have not moved in 16 years. Today they're worth tens of billions of dollars.

Why it matters

The Patoshi analysis is the closest thing we have to an objective measurement of Satoshi's holdings. It's not perfect (other miners may have shared the same early Bitcoin Core version and produced similar patterns), but the cluster is dramatically isolated from the rest of the network's behavior. Most analysts accept that the patterns identify a single dominant early miner, and that miner is overwhelmingly likely to be Satoshi.

These coins represent a structural supply overhang: roughly 5% of the eventual 21M cap, sitting in addresses with no known private-key holder and no movement for over a decade. Every Bitcoin holder is implicitly betting that they'll never move, or that the market can absorb them if they do.

You can check on Satoshi's coins any time by querying balances at the addresses publicly attributed to the Patoshi cluster. They've been monitored continuously by analysts for years. So far, nothing.

Sources

Corrected October 2, 2026. An earlier version said the Patoshi miner produced roughly 22% of early blocks and that its ExtraNonce values stayed unusually small; Lerner's figure is about 40% of the first 50,000 blocks, and the fingerprint is an ExtraNonce that climbs unusually fast. It also misstated several figures and dates: the holdings were given as 750,000 to 1,100,000 BTC (Lerner's counts are 980,000 and about 1.1 million), block 100 as January 9, 2009 (it is January 11), block 50,000 as October 2009 and after the pattern's end (it is April 10, 2010, shortly before the end), the end of the pattern as late 2010 (it is about May 2010), and Satoshi's April 2011 emails as public messages sent a few months later (they were private emails sent about a year later). The genesis coinbase example used a getrawtransaction call that Bitcoin Core refuses for the genesis block and now uses getblock, and the genesis scriptSig's leading bytes were described as a height encoding when they are the difficulty bits and the number 4. Sources added at the suggestion of ABCbits on BitcoinTalk.

Read more: The Genesis Block, The Halvings, or The Inflation Bug.