bitcoin09

README

Bitcoin 09 (BTC09)

The coin that you can mine like it's 2009.

This is Bitcoin, same idea top to bottom, changed in one place: the proof of work is Argon2id (memory-hard) instead of SHA-256. That means normal CPUs do the mining and ASICs/GPUs get no real advantage. Everything else follows Bitcoin.

	Bitcoin	Bitcoin 09
Supply cap	21,000,000	21,000,000
Block reward	50, halves every 210,000 blocks	50, halves every 210,000 blocks
Block time	10 minutes	10 minutes
Premine	none	none
Genesis reward	unspendable	unspendable
Proof of work	SHA-256d (ASICs)	Argon2id 64 MiB (CPUs)
Signatures	ECDSA/Schnorr	Ed25519

Read this first

BTC09 is worth nothing. Bitcoin was worth nothing in 2009 too, that's the whole idea here. No premine, no company, no allocation. The genesis reward is burned to an address nobody can spend from, so every coin that will ever exist gets mined by someone's CPU starting at block 1. Mine it because it's fun and you missed 2009. Don't put in money you care about, and don't expect any back.

Quick start

You need Go 1.25+.

git clone https://github.com/krutftw/bitcoin09
cd bitcoin09
go build ./cmd/btc09

Mine with all your cores:

./btc09 node -mine -seeds 82.22.32.82:9009

Your reward address prints at startup (wallet auto-created in ~/.btc09).

Wallet stuff:

./btc09 wallet list
./btc09 wallet new
./btc09 send -to ADDRESS -amount 1.5 -seeds 82.22.32.82:9009

Local sandbox with instant blocks:

./btc09 node -mine -network regtest -datadir /tmp/sandbox

Why change the proof of work

Bitcoin mining stopped being for normal computers around 2013 when ASICs took over. Today a laptop has a one in millions share against warehouse hardware.

Argon2id needs 64 MiB of memory per hash attempt. Memory bandwidth is the bottleneck, not raw logic speed, so a GPU's thousands of cores sit starved and an ASIC would basically be a stack of the same DRAM your laptop already has. Monero's RandomX has used the same reasoning for years and CPU mining is still viable there.

Nothing stays ASIC-proof forever, but memory-hard keeps the gap between a laptop and custom hardware small. That's what makes the 2009 experience possible again.

Consensus rules, short version

• 88 byte headers, block ids are SHA-256d, PoW check is Argon2id(header) <= target
• difficulty retargets every 144 blocks (about a day), max 4x move per step
• UTXO model, Ed25519 signatures, base58check addresses (version 0x09)
• coinbase pays subsidy + fees, spendable after 100 blocks
• heaviest total work wins, reorgs replay the UTXO set
• 1 MB max block

Mainnet genesis is pinned by a test: ba685f741a04ddad03d37500ff354ce3887e64dd9cb6154ae236952792e90c3f with the message "the coin that you can mine like it's 2009", nonce 20214.

Tests

go test ./...

Covers the emission schedule and 21M cap, rejection of inflated coinbases, signature checks (wrong key can't spend), double spend rejection, difficulty retargeting, reorgs, and the pinned genesis.

Status

v0.1.0. Working node, miner and wallet, simple JSON p2p. Stuff I want next, PRs welcome: peer banning and DoS hardening, headers-first sync, DNS seeds, a block explorer, a friendlier miner UI.

License

MIT, same as Bitcoin.