Patoshi: The Nonce Fingerprint Across Bitcoin's First 50,000 Blocks

On 17 April 2013 the researcher Sergio Demian Lerner published "The Well Deserved Fortune of Satoshi Nakamoto, Bitcoin creator, Visionary and Genius" on his Bitslog blog. The method was simple. Bitcoin v0.1 incremented an ExtraNonce counter inside the coinbase script when the 32-bit header nonce overflowed, and did not reset it on finding a block — only on restart, when it went back to 1. Plotted against block height, one miner's ExtraNonce values trace a straight line whose gradient encodes its hash rate. He graphed "block 0 upto block 36288" and found a dominant, remarkably regular slope, restarting "approximately every 100 hours, probably to backup the wallet". His estimate was deliberately loose: "I estimate at eyesight that Satoshi fortune is around 1M Bitcoins." He was explicit about the limits: "I can't assure with 100% certainty that the all the black dots are owned by Satoshi, but almost all are owned by a single entity."

By 2019 Lerner had a sharper distinguisher, published in "The Return of the Deniers and the Revenge of Patoshi" on 16 April. The least significant byte of the header nonce, which ordinarily sweeps 0–255, was in these blocks confined to "[0..9] U [19..58]". Across the first 50,000 blocks — block 50,000 is timestamped 10 April 2010 — he counted "about 22k blocks in total", and wrote that "the Patoshi blocks account for 43% of all the blocks in the first 50k". He argued for a larger total: "I have three new strong arguments that Satoshi mined close to 1.1M coins." A timestamp statistic did the heavy lifting for the single-entity case: "There are no time inversions between Patoshi blocks. Zero," against 224 among non-Patoshi blocks and 398 across the boundary between the two sets. He also reported that 99.9% of Patoshi blocks were unspent at the time of writing, against roughly 10% of non-Patoshi blocks.

The size of the set has never been settled. BitMEX Research, publishing on 19 August 2018, reconstructed the ExtraNonce slopes independently and concluded the dominant 2009 miner more likely produced "600,000 to 700,000 bitcoin". Their caveat was pointed: "Although there is strong evidence of a dominant miner in 2009, we think the evidence is far less robust than many have assumed." They found the pre-August 2009 slopes consistent — "an increase in the value of the ExtraNonce by 4 per block found" — but that afterwards the gradient "varies considerably (from 1.1 nonces per block to 10 nonces per block)", weakening attribution in the later period. Whale Alert, publishing on 20 July 2020, went the other way, attributing "22,503 out of the first 54,316 blocks" and 1,125,150 BTC, and identifying 18 spent blocks totalling 907 BTC.

The hardware question splits the researchers just as sharply. Whale Alert reasoned that "Patoshi was comprised of at least 48 computers, with one machine for coordination". Lerner reached the opposite conclusion in "The Patoshi Mining Machine" (22 August 2020). Re-mining every Patoshi block in the first 13.2K blocks — 10,025 of them — his simulation turned up 1,075 additional solutions, and the five scanned subranges [0..9], [19..28], [29..38], [39..48], [49..58] mapped onto parallel threads. "Patoshi was simply multi-threading in a high-end CPU," he concluded, of roughly Intel Core i7-965 Extreme class. After the first 18,000 blocks, he found, "Patoshi starts reducing his hashrate by removing parts of the nonce range scanned."

Block 54,316, the endpoint Whale Alert uses, is timestamped 3 May 2010 at 13:17:07 UTC, at difficulty 11.46. Lerner's own closing note remains the honest one: "Because the relation is based on unsigned emails and not math, we'll never be sure."