Carbon Dating the Internet Archive with OpenTimestamps
Peter Todd's primary-source account (May 2017) of timestamping about 750 million Internet Archive files with OpenTimestamps: SHA1 digests read from the archive's own metadata, aggregated through a calendar server into a Merkle tree, and committed in a single Bitcoin transaction.
Overview
This is Peter Todd’s primary-source announcement, dated 25 May 2017, of using OpenTimestamps to timestamp the entire Internet Archive: about 750 million files across thousands of collections (books, movies, songs, programs, and legal documents). It is the authoritative first-hand account of the effort by the creator of OpenTimestamps; later coverage, such as the Bitcoin Magazine explainer cited below, is secondary reporting on it. OpenTimestamps is an open, free standard for proving that data existed at or before a point in time (proof of existence) by committing hashes to the Bitcoin blockchain.
Key points
- The effort timestamped roughly 750 million files by reading the SHA1 digests the Internet Archive already exposes in its item metadata through its API, so no file content had to be downloaded or uploaded.
- Those hundreds of millions of digests were aggregated through the OpenTimestamps calendar server, which keeps “an append-only journal of commitments”, into a single Merkle tree.
- The whole batch was committed to Bitcoin in exactly one transaction (
8465d34e...), funded by 0.218 BTC of donations from dozens of contributors, in deliberate contrast to schemes that waste thousands of separate transactions. - Riccardo Casatta and Luca Vaccaro of Eternity Wall built the searchable database interface over the timestamps, and Igor Barinov contributed to the UI design.
- Verification is trustless and offline-capable: a proof traces back to the Bitcoin blockchain, “a widely witnessed data structure” whose timestamps “can’t be backdated”, so anyone can confirm a file existed by the committed time without trusting the calendar or the archive.
- Caveats the author states: SHA1 is broken for collision resistance but “good enough for timestamps” because only preimage resistance matters here; the first pass missed an entire year and parts of two others because of incomplete directory copying; and some items lacked the metadata needed to be covered. The ~750 million figure counts archive items, not Wayback Machine web snapshots, which number over half a trillion objects.
Relevance to Truestamp
This effort is a direct antecedent of Truestamp’s model: hash data, aggregate many hashes into a Merkle tree, and commit a single root to a public blockchain so existence can be verified independently later. Committing 750 million files in one Bitcoin transaction is the clearest illustration of batched aggregation, the same amortization Truestamp uses when it folds many items into one block commitment. It also shows the boundary Truestamp is careful to name: a timestamp proves a submission window, that data existed by a given time, not its authorship or original creation.
Citations
- Carbon Dating the Internet Archive with OpenTimestamps. Peter Todd, 25 May 2017. Primary-source announcement of the Internet Archive timestamping effort.
- OpenTimestamps Has Timestamped The Entire Internet Archive: Here’s How. Bitcoin Magazine, technical section. Secondary explainer covering the same effort.