The Anchored Ledger

Begin with a deceptively simple goal. I want to be able to prove, years from now, that a particular fact existed at a particular moment and has not changed since. That is the whole of it. Yet almost every system we rely on fails this test under pressure. A database can be edited by anyone with the right credentials, and the edit can be made to look original. A log file can be truncated, and a fresh log written over it with no seam to show the join. A cloud archive depends on

Here is the part that surprises people. To anchor an entire chain's worth of state into Bitcoin, you do not write the state. You write a fingerprint of it. Pantheon takes its current state, the set of balances, the contents of its application chains, the audit records sealed since the last anchor, and reduces all of it to a single cryptographic commitment, typically the root of a Merkle tree. That root is a short fixed-length string. Whether it summarises a thousand records o

Precision about claims is where most blockchain writing falls apart, so let me be exact about what an anchor demonstrates. After a Pantheon root is committed into Bitcoin and that Bitcoin transaction is buried under enough subsequent blocks, three things become provable to a sceptic. That the committed state existed no later than the timestamp of the Bitcoin block that contains it. That the state has not changed since, because any change would break the root. And that any spe