The virtual currency Bitcoin has soared in value against the U.S. dollar in recent months, topping out a staggering $913 USD to 1 Bitcoin (or BTC) as of late Tuesday. The currency had many ups and downs since it was launched in January 2009. But its main attraction, all along, has been anonymity.
Unlike any other online payment system, Bitcoin transactions – like cash transactions – cannot be traced back to specific individuals. Also like cash, they cannot be reversed. Both those factors give Bitcoin users the confidence that their online purchasing activity – whether computer hardware or contraband will remain private.
But a group of researchers at two U.S. universities have released a paper that suggests reports of Bitcoin’s anonymity may (to paraphrase Twain) “be greatly exaggerated.” Specifically: the researchers found that, by culling a variety of open source data using public data from the Bitcoin Peer to Peer network and from their own Bitcoin transactions and public Internet postings, they were able to “identify major institutions” engaged in Bitcoin transactions “and the interactions between them.”
The experiment, though small, suggests that a large slice of the public keys used in Bitcoin transactions – around 14 percent – can be linked back to larger, institutional players, including banks, Bitcoin (or BTC) exchanges or large vendors like the now defunct Silk Road. That centralization makes the Bitcoin network susceptible to surveillance by law enforcement or governments that have the computing power and determination to track down the individuals, groups and institutions at either end of specific exchanges.
The paper, “A Fistful of Bitcoins: Characterizing Payments Among Men with No Names” was presented at the IMC (Internet Measurement Conference) 2013 Conference in Barcelona, Spain in October and is reprinted in the December issue of ;login: logout a USENIX publication. It is based on research conducted at The University of California, San Diego and George Mason University. In it, the researchers, led by Sarah Meiklejohn of UCSD used a combination of strategies to “de-anonymize” the BitCoin network.
Aspects of the work have been noted before in news reports, including work that Meiklejohn did with Brian Krebs of Krebsonsecurity tracking an online purchase of heroin in Krebs name. However, Meiklejohn has recently expanded her analysis of Bitcoin protocol and its potential weaknesses.
In the first part, the researchers set out to “tag” (or uniquely identify) as many Bitcoin holders as possible. They did this by crawling exchanges and discussion groups where Bitcoin owners published their public key(s). The researchers also engaged with the Bitcoin network in various roles: as part of a shared mining pool and in actual transactions with merchants via Bitcoin portals like CoinDL, BitPay and BitMint. Whenever possible, they recorded the public keys associated with various actors and institutions.
Next, the researchers developed a novel, two-stage heuristic for clustering discrete, but related Bitcoin transactions. First, they noted different public keys that were inputs for the same transaction, inferring a shared owner for the two, discrete public keys. Just noting that information allowed them to boil down a block of some 12 million unique PKI keys into around 5 million Bitcoin ‘actors’ (individuals and institutions). By knowing their own public keys and where they spent their Bitcoins, they could trace transactions and group public keys associated with BTC.
In the second part, the researchers studied “change addresses” – new private key/public key key pairs that are created by special Bitcoin change applications as a way to return change to the buyer in a BTC transaction. As with other kinds of Bitcoin transactions, the researchers found that they could track transactions that result in change being created.
By mapping unique Bitcoin change addresses, the researchers were able to further break down their cluster of around 5.5m distinct clusters of public keys to 3,384,179 clusters. Then, after combining the fruits of their open source research and “tagging,” the researchers positively identified 2,197 of the clusters, linked to over 1.8 million BitCoin addresses.
The underlying Bitcoin architecture provides for anonymity. However, current Bitcoin marketplace tends to undermine that – opening a (widening) gap between the amount of anonymity to be had, in theory, and in practice. In particular, all BitCoin transactions all are associated with a public key, while all Bitcoin transactions are recorded in transaction block chains. Public tools allow anyone to download and extracting transaction information from Bitcoin block chains, revealing associations between public keys that are subject to high-power data analytics.
“I think (the research) is just the tip of the iceberg,” said Bruce Potter, a noted security expert who founded The Shmoo Group and Ponte Technologies. Bitcoin’s anonymity only works only as long as other parties can’t associate a public key with a real world identity, Potter notes. But once that link is made, the cryptographic underpinnings of the Bitcoin network work against anonymity, providing a “cryptographically bound, (and) public record” of that relationship – effectively forever.
“That’s really a core part of the protocol that will always be working against those trying to do illicit or untraceable things with BTC,” Potter said.
Steps to inject anonymity into the current system will just create an arms race, Potter warns. “The exchanges are an example. If you can use the exchanges to launder money off the block chain, then you stand a chance of getting away free and clear. However, the exchanges are clearly where the governments will put their effort to gain access and regulate the audit trail (just like any other financial exchange).” For exchanges to survive, they will need to become more- not less transparent, Potter said – in other words: more like traditional currency exchanges.
Bitcoin has become the most prominent among a range of alternative, ‘non-fiat’ currencies. The U.S. Government Accountability Office (GAO) issued a report in June that found many Bitcoin denominated activities, from mining to online purchases were likely subject to U.S. taxes. Two hearings on Capital Hill last week were overwhelmingly positive, with the Obama Administration saying that it saw legitimate uses for the online currency that rose to fame as the preferred payment method for shadowy online marketplaces like The Silk Road.
But Bitcoin’s supposed secrecy may finally be seen as an accident of history – the byproduct of the virtual currency’s early days. Like most other “Big Data” problems, the more economic activity around Bitcoin, the more data there is to correlate and the more transparency, overall, Potter said.
“I think whatever anonymity that’s been enjoyed on the network to date has been the result of the relatively low volume of transactions and actors on the network,” Potter wrote. “Now that’s its growing exponentially, that anonymity will start disappearing, in my opinion.”