The common-input-ownership heuristic

Graduate · enrolled learners

This lesson opens with Bitfinex hack, 2016, seized 2022.

What happened

In August 2016 119,754 bitcoin was stolen from the exchange Bitfinex in more than 2,000 unauthorized transactions, and most of it then sat still for years. Moving it was the problem, because on a transparent ledger every attempt to spend or consolidate reveals structure. Over five years about 25,000 of the coins were moved out through a complicated laundering process, through exchanges, darknet markets and chain hopping, that ended with some of the funds in financial accounts controlled by Ilya Lichtenstein and Heather Morgan, and investigators traced it there. The standard first tool for that kind of tracing is clustering: when a single transaction spends several inputs at once, those inputs are almost always controlled by one entity, since one party must hold all the keys to sign them, so each such transaction merges the addresses it spends into one cluster. In February 2022 the Department of Justice arrested the two and seized more than 94,000 bitcoin, valued at over 3.6 billion dollars, using private keys found, under search warrant, in files in Lichtenstein's own online account. No cryptography was broken; the ledger's structure carried the trace to the accounts, and the keys came from the suspects' own files.

The decision point

The Bitfinex thieves held anonymous-looking addresses and still lost, because addresses are not the unit that matters; entities are, and the chain quietly reveals which addresses belong to the same entity every time that entity spends. The common-input-ownership heuristic is the first and most powerful tool for that: inputs spent together in one transaction share an owner, so transactions weld addresses into clusters. This is the analyst's move from reading single addresses to reading actors. It is a heuristic, not a theorem, it can be defeated by techniques that deliberately break the shared-ownership assumption, but it is right often enough that patient tracing of this kind turned a field of unlabeled addresses into a case that put people in front of a judge. Learning to cluster is learning to see the entity behind the addresses.

Recorded loss

$3,600,000,000

Orientation and Year One are open: anyone can read them without an account. From Year Two onward the lessons are for enrolled learners, because progress through the later years only means anything if it is tracked against a record.

It is free. We do not sell the list and there is nothing to buy at the end of it.