Bitcoin Eater: The Burn Address Built Back to Front

A legacy Bitcoin address is a version byte plus a 160-bit hash plus a four-byte checksum, base58-encoded. Because that hash sits in the middle of the string, whoever chooses the hash chooses most of the visible characters. Ordinarily the hash is derived from a public key. In a constructed burn address the direction of travel is reversed: you pick the hash for how it reads, then compute the checksum so the result validates.

1BitcoinEaterAddressDontSendf59kuE is the best-known example on Bitcoin. Its hash160 is 759d6677091e973b9e9d99f19c68fbf43e3f05f9, chosen so that base58 renders it as a warning, with the trailing characters doing the arithmetic that makes the string a legal address. Wallets accept it. Nothing derived it from a key, so no key is known to exist.

The contrast worth drawing is with vanity addresses proper. Vanitygen, announced on the Bitcoin forum by samr7 on 4 July 2011, works the other way. In its author's description it uses the RNG to choose a private key and literally increments that key in a loop, searching for a match. A vanitygen address has an owner. The Eater does not. They look like the same trick and are opposites.

The first payment to the address landed in block 132,184 on 21 June 2011, about two weeks before vanitygen's announcement, and was 0.01 BTC. Thousands of transactions have followed, which is the odd part: the address spells out "Dont Send" and people keep sending. Community explanations range from experiment and joke to deliberate supply reduction to services burning as part of normal operation. Nobody has publicly claimed authorship.

Strictly, unspendability here is an assumption rather than a proof. Spending would require finding a script whose hash matches. The Bitcoin Wiki's own caveat is that this holds barring the cracking of RIPEMD160 and the finding of an alternative matching script, at which point, it notes, the cryptocurrency is in big trouble anyway. The same construction produced a small family of siblings: 1111111111111111111114oLvT2, whose hash160 is twenty zero bytes, and 1QLbz7JHiBTspS962RLKV8GndWFwi5j6Qr, whose hash160 is twenty 0xff bytes, at the opposite end of the range.

Eater addresses also carry a cost the ecosystem later moved to avoid. Coins paid to them stay in the UTXO set, because a node cannot prove a plausible-looking pay-to-hash output is unspendable. Bitcoin Core 0.9.0, released 19 March 2014, added OP_RETURN explicitly as a provably-prunable output, to avoid schemes that were storing arbitrary data as forever-unspendable outputs and bloating the UTXO database. The release notes were pointed about it: storing arbitrary data in the blockchain remained, in their words, a bad idea.

The practice never died. A 2025 study by Mohamed El Khatib and Arnaud Legout trained a classifier on 196,088 regular and 2,082 burn addresses and identified 7,905 burn addresses among 1,283,997,050, accounting for 3,197.61 BTC permanently removed, 0.016% of supply, roughly $295m at November 2024 prices, with more than 99% concentrated in just three addresses. Their catalogue of uses is a decent map of why anyone still does this: token creation via proof of burn, plain-text messages, and image storage through the OLGA encoding within the Stamps protocol, which builds on Counterparty and first appeared at block 833,000 in March 2024, a year after the first Bitcoin Stamp at block 779,652 on 7 March 2023.

The Eater address, then, is less a monument than a convention: an agreed-upon place to put coins you want visibly gone.