The ontology lays out how Chainalysis views the role of attribution to these clusters, presenting a two-tier structure; the first tier “defines the structural graph,” while the second assesses how confident the analysis is in that graph.
“What does it mean that these addresses belong together, right? It’s clearly because somebody believes that they are under the control of the same entity, right?” Illum said. “Maybe it’s an exchange, or maybe it’s a darknet market, or maybe it’s a mixer, or whatever. But what are the grounds to establish that these things actually belong together?”
Investigators likely won’t have private keys, which would be the easiest way to tell if a cluster of addresses is all being controlled by the same entity, so they would then have to look at onchain data.
Illum was also clear about the limitations of this type of analysis: While Chainalysis could conduct research into transactions and clusters, it cannot, on its own, identify the actual end user without additional information.
Chainalysis could track funds to a crypto exchange, for example, or another entity managing wallets on behalf of customers, but investigators might need to issue a subpoena to identify who the customer is.
In other words, who controls a wallet or what entity is associated with the wallet are separate questions from the actual tracing aspect.