RPOW

On 15 August 2004, Finney invited the Cypherpunks mailing list to test RPOW, offering a server, source code and a web exchange interface. This was a working beta, not just a white paper. The original message was sent at 17:43:09 UTC; the archive's 16 August heading marks a subsequent extension of the invitation.

A user computes a Hashcash token and exchanges it for a server-signed RPOW token. Its recipient then exchanges it through the server for a fresh token of equal value. Each old token is redeemable once: reuse means passing value onward, not spending one token repeatedly. The server tracks redeemed tokens to prevent double-spending.

RPOW runs its critical program inside an IBM 4758 secure coprocessor. Remote attestation lets users check that the running program matches the published source. This reduces reliance on an operator's promises, but still requires trust in the hardware, IBM's certification and correct software. An operator can also shut the server down, stopping exchanges.

Our comparison: both use computational work. Bitcoin confirms history through a peer-to-peer proof-of-work chain, without RPOW's exchange server. Its whitepaper does not cite RPOW. Similarity alone does not establish direct adoption.

Finney presented RPOW as a non-commercial experiment, suggesting uses such as spam prevention and games. He explicitly rejected treating it as a stable store of value: faster computers make tokens of the same difficulty cheaper to produce. Read it as an experiment in exchangeable computational effort, not an established monetary market.

Hal Finney worked on PGP encryption software, participated in the Cypherpunks community and operated an anonymous remailer. In his 2013 retrospective, Bitcoin and me, he described testing Bitcoin early, reporting bugs to Satoshi and receiving a test payment of 10 bitcoins from him. This is a participant's account written after the events.