This week marked the beginning of the trail of software designer Ross William Ulbricht, accused of being the progenitor and operator of the online black market “Silk Road.” Silk Road was arguably the most prominent of several anonymous online black markets, all facilitated by Tor. Tor, short for “The Onion Router,” is a free online software package that enables online anonymity and is renowned for being one of the most difficult to wiretap methods of communication on the internet. Through a system known as onion routing, Tor allows people from anywhere in the world (even in internet restrictive countries like China) to communicate completely anonymously, and has served as a bastion for both unhindered free speech and difficult to track illegal activity.
The Matryoshka Network
The details are complicated, but the basic idea of Tor is that rather than communicating directly with a website, you communicate through a series of intermediary “nodes” on the Tor network, each with limited knowledge of your overall communication. For comparison, imagine that traditional internet communication worked by sending letters. Your computer sends out a letter with a “to” address, a “from” address, and the content. Everyone who passes that letter along sees the same information: they know where the letter is going, who sent it, and that it has content. The content might be like a postcard, which anyone could read (unencrypted data), or like a sealed letter, which could only be opened by the recipient (encrypted data).
Tor, by contrast, works like a Russian nesting doll: each person who handles the doll has the ability to open only one layer of the nesting doll, which tells them where to send the next, smaller doll. Tor does this several times, until you reach the final doll, which will be read by an “exit node,” which then communicates with the website. The website only sees this exit node, and each previous level of the doll is encrypted, meaning that the individual nodes that pass the various dolls have limited knowledge of each other, protecting the user even if one of the nodes or dolls is compromised. In cryptology, this is referred to as being forward secret. When finished, each little doll jumps back into its slightly bigger doll, until it eventually reaches the original sender. (Tor uses the metaphor of peeling an onion, hence the name, but I like the Russian dolls better.)
Tor is noteworthy for its surprising resilience in an insecure world. As recently as 2012, the NSA identified Tor as causing them “major” problems, reflecting their inability to effectively monitor communications on the platform. When combined with several other security measures, Tor was deemed “catastrophic,” and represented one of the few platforms that could not be consistently cracked by even the highest level cryptologists with ample time. Which is not to say that Tor is impenetrable, the NSA has identified several potential weaknesses in Tor’s structure they may exploit (and the NSA may have developed more effective methods of surveillance since this document was written), but overall the service provides a consistently secure and anonymous online environment.
Since Tor is one of the few seemingly secure online anonymity platforms, its history of use typifies both the best and the worst of internet anonymity. The software is frequently used by people living under repressive regimes, by corporate and government whistleblowers, journalists, and even by those wishing to avoid cyberstalking. Yet it also facilitates the recent emergence of online black markets, dealing in drugs, sexual trafficking, money laundering, and countless other illegal activities. These online black markets are proliferated through Tor’s “hidden services” (websites only accessible through Tor) and are financed with Bitcoins, a difficult to track digital currency. Given Tor’s resilient anonymity, policing these sites has been a recurrent law enforcement challenge.
One of the difficulties in rooting out these black markets is that the continued existence of Tor allows anyone with the source code to repost the entire website after the initial host was arrested. After the initial takedown of Silk Road in 2013, another anonymous user created Silk Road 2.0, and also publicly released an encrypted file with the website’s source code. This ensured that if Silk Road 2.0 was taken down, someone else could create a Silk Road 3.0 (and indeed Silk Road 2.0 was eventually taken down, although other black markets still exist.) This distributed, decentralized marketplace of illegal activity makes individual sting operations relatively ineffective, and has led many to suggest more comprehensive regulations targeted at Tor itself.
How to Regulate Tor
The most commonly discussed legal concerns for Tor are probably DMCA challenges. These are basically copyright infringement actions that require service providers to remove copyright infringement occurring with their services. Yet most analysts agree the users who host Tor nodes and exit nodes fall within the safe harbor provision, meaning that someone who simply hosts a Tor node and doesn’t monitor the traffic has no obligation to police the content of the traffic it parses. This is the stance the Electronic Frontier Foundation espouses, which they back up by hosting a Tor exit node of their own. And while it has never been formally challenged in court, there seems to be little doubt that most users hosting Tor nodes are immune from liability for the copyright infringement of the content they pass along.
So perhaps the better question is: is there a way to regulate Tor that would allow government officials to target these black markets without compromising the integrity or purpose of the system? I’m tempted to say no. The most frequently discussed target for regulation would be the Tor exit nodes; these are the nodes that actually communicate with the Tor user’s desired website, meaning they have the actual data from the original user. This would theoretically allow for exit node hosts to monitor the content they process, and screen out requests from illegal sites or for illegal activity. Yet since exit nodes are located across the globe, it is unlikely that US regulations would have jurisdiction over all or even most of them, meaning ensuring compliance would be nearly impossible.
And even for domestically operated nodes, there is substantial pushback against government mandated “backdoors.” As technology increasingly generates methods of communication that cannot be wiretapped or devices that cannot be searched, the government is increasingly attempting to require backdoors to be implemented in the design process. Basically, if someone develops a safe that no one else can open, the government wants to ensure that they are given a skeleton key. These movements have been met with stern criticism by both privacy and security experts, claiming (correctly) that the creation of government backdoors also creates a security vulnerability for criminals to exploit, and that this is a fundamental violation of civil liberties. I would take issue with the later portion of that argument, as it’s difficult to claim you have a protected right to purchase something, or that these merchants have a protected right to sell it. (The government’s power to regulate commerce is quite broad.) Yet however applicable these arguments may be to commercially produced products, Tor is freely available software, which makes government mandated alterations of the source code both practically infeasible and probably unconstitutional, albeit for different reasons.
A Right to Anonymity
Although we have no broad Constitutional right to anonymity, there are several potential Constitutional challenges to any attempt to ban Tor or Tor node operators. To begin, the First Amendment seems to provide a broad protection for the dissemination of information, so long as that information isn’t subject to intellectual property laws. This has been seen in cases involving the publication of source code for encryption, and suggests that the continued availability of Tor’s source code would likely be protected, notwithstanding potential challenges to its use. This can also be seen in the cases involving Napster, LimeWire, and other P2P sharing services, where the use of the service for transferring copyrighted material was illegal, but the availability of the software was not. This First Amendment right is not unlimited, however, as any sufficiently dangerous information might satisfy the high requirements of strict scrutiny and therefore be banned. But it is unlikely that Tor, which merely facilitates anonymity, would come anywhere close to that high bar. (Exactly what would constitute sufficiently dangerous information is an interesting question; possibly code to 3D print a nuclear weapon?) So while it may be illegal to use Tor to violate copyright law, this cannot be extended to banning Tor itself, as this would be an overinclusive restriction on free speech.
There are also likely 4th Amendment challenges to any persistent government monitoring of Tor nodes. The current system of large scale metadata collection works because the NSA can request metadata in bulk from Telecom providers. For several reasons I’ve discussed previously, there is likely no 4th Amendment challenge on behalf of the individual in these circumstances. But Tor is fundamentally different because the individual nodes are controlled by the users. The Tor node operators would be able to raise a 4th Amendment challenge to the search of their own server, and they would most likely not be subject to subpoena requirements of the larger telecommunications companies. This would probably make any widespread or preemptive searches by the government unconstitutional.
Rather ironically, however, the 4th Amendment may not protect those hosting websites for illegal activity through Tor’s hidden services. Since the websites are hosted anonymously, any government hacking of the website would place the owner in a lose-lose situation, as they would have to claim ownership of the website in order to assert a 4th Amendment challenge to the search. Such a claim would be essentially admitting a large portion of the prosecution’s case, and would make the illegally searched material unnecessary. This was the analysis of the district court judge in the Silk Road case, although the defense preserved their right to appeal.
Furthermore, since Tor’s users are anonymous, the laws regulating when surveillance activity requires a warrant are less clear. Traditionally, the 4th Amendment applies different standards for American citizens and non-American citizens, which are further separated into those located domestically and those located abroad. This is a complicated issue, but the important aspect is that the government may seek warrantless wiretaps without a court order when it is conducting surveillance on a non-American citizen located abroad. But since law enforcement and national security agencies do not know whether a Tor user is located within the United States or abroad, or indeed whether they are a US citizen, this analysis is likely to skew in favor of a greater number of warrantless wiretaps. So while the 4th Amendment may provide some protections to Tor users from broad government surveillance, it may also open them up to greater surveillance on an individualized basis.
Personally, I think the better target for regulation would be Bitcoin, the digital currency used by these Tor black markets to finance their transactions. Bitcoin was originally almost completely unregulated, making it ideal for criminals. Yet as Bitcoin increases in prominence, it has come under the scrutiny of the SEC and IRS, and more substantial financial regulations are already in progress. While I am by no means an expert in financial regulation, I know that Bitcoin was utilized in this case primarily because it allowed criminals to maintain a larger degree of anonymity than traditional financial exchanges, and increasing regulation will also likely entail recording requirements that will remove this anonymity. Although we may have a First Amendment right to use anonymous communication systems like Tor, we are unlikely to have a right to engage in anonymous financial transactions, or to hold accounts without the IRS’s knowledge. By regulating Bitcoin, this would provide sufficient measures for the government to police illegal commerce without necessitating a shutdown or direct oversight of Tor itself.
Tor in many ways exemplifies the unique problems that technology brings to law. Tor empowers us to do something that has previously never been possible: truly untraceable communication. (I should note that many contest the possibility of completely unbeatable security measures. While I am not convinced this is the case, it suffices to imagine defenses so powerful that government intervention is made utterly impracticable.) One of the trends throughout history is that our personal security could never be entirely ensured through private measures, and we therefore ceded some portion of our security to the sovereign. Our imperfect locks and safes were protected by the threat of government enforcement, and the imperfections in the criminals’ locks and safes facilitated that enforcement. With technology creating potentially uncrackable safes and unpickable locks, we may gain a greater degree of personal security, but we are also potentially losing some of the security granted by the sovereign. If the sovereign were no longer in a position to guarantee our security, the resulting system would favor the wealthy and tech-savvy at the expense of everyone else. I’m not convinced this will ultimately be a major issue, but it certainly raises interesting questions.
And more broadly, does the emergence of new technology ever create new rights? While Tor may be a mostly innocuous example, how would we treat the creation of artificial intelligence? Or the digitization of our own consciousness? For those who enjoy speculating about what the future may hold, the answer may entail some wonderful and worrying repercussions.
But in the meantime, if you find yourself needing anonymity, Tor looks like a pretty good option, and should be here to stay.