Attribution’s impossible---means the plan’s norm can’t be enforced
Kenneth Lieberthal 12, director of the John L. Thornton China Center and senior fellow in Foreign Policy and Global Economy and Development at Brookings; and Peter W. Singer, director of the 21st Century Defense Initiative and a senior fellow in Foreign Policy at Brookings, February 2012, “Cybersecurity and U.S.-China Relations,” https://www.brookings.edu/wp-content/uploads/2016/06/0223_cybersecurity_china_us_lieberthal_singer_pdf_english.pdf
Beyond just the issue of terminology, however, there are also other dimensions in the cyber arena that are especially consequential to U.S.-China relations and must inform any process aimed at engagement. Put simply, in this dynamic arena, it is necessary to be realistic about what types of activities can potentially be brought under control through agreed-upon actions and what things are effectively beyond the capacity of institutions to control. It is important to appreciate why this is so and, in that context, to focus on those areas that are still potentially subject to meaningful controls. Perhaps the most difficult is the problem of attribution.41
The ability to capture the operations of another computer and use it to launch activities that its owner does not intend and might even be unaware of (described inside the field as to “pwn,” a computer hacker term meaning to “own”) is very well developed and widely used. This often takes the form of creating botnets that link unrelated computers and enable the controller to leverage their combined computing and communications capabilities for a particular purpose. The resulting network of secretly linked devices can easily grow to extraordinary dimensions. For example, three not terribly sophisticated Spaniards allegedly created a global botnet that included over twelve million computers.42 In other cases, a controller may seek to capture and leverage only one or a small number of computers. In this case, the major purpose is more likely to be to conceal the controller’s identity.
Three key features of this capability to capture and utilize other computers are particularly important. First, there are no geographical limits. For example, a pernicious actor in Brazil can compromise computers in Russia and South Africa to launch attacks on systems in China (one can substitute almost any countries for the ones named here). Second, the owner of a captured computer typically has no idea that the computer is being used by a remote actor for pernicious purposes. And third, when some pernicious activity is perpetrated, sophisticated analysis can typically, at best, identify the computer being used to launch the attack. It is far more difficult to determine whether that computer is, in turn, being remotely controlled and, if so, by whom. Equally, even if it is not being remotely accessed, in many situations (such as a computer being used at a university or an internet café), it will be difficult to determine who the individual behind the computer is, or even what his nationality is or what organization he actually represents. Such information would be crucial in a crisis but is rarely available in a timely manner. In short, it is typically not feasible to identify the underlying command structure behind an attack with complete certainty. Instead, painstaking forensics may be a key driver to understanding the full chain of events.
It does not take much imagination to see how pernicious the resulting problems can be. Since many are inclined to assume that the Chinese government is behind most insidious activities that are launched by computers located in China, for example, bad actors elsewhere may be inclined to capture Chinese computers to use in their activities, and vice versa for the U.S. This same logic, though, also enables Chinese bad actors to deny responsibility, arguing that activities launched from China almost certainly are being perpetrated by others who want to take advantage of the widespread suspicions of China. And the same type of misdirection can be argued regarding computers physically located inside the U.S.
The issue of attribution is further complicated by the difficulty of establishing complicity. An effort might be attributed to an actor emanating from a certain geographic locale, but it is even more difficult to establish a formal role of a government as perpetrator or sanctioner of the operation. That is, the situation often parallels that of maritime security in centuries past, where the lines between what was criminal piracy and what was state-sanctioned privateering often proved fuzzy.43 There have been frequent reports of “patriotic hacker” communities and other non-state groups, including student and even cyber criminal groups that have been mobilized by their governments for such purposes of deniable, but directed, attack.44 Thus, in certain cases where the local state government may be needed to investigate and then prosecute those behind questionable cyber activity, the victims perceive that some elements of the state are actually a willing accomplice or planner of the attack, and thus unlikely to assist. An example would be the DDoS attacks on Estonia. Many believed Russian security services actually instigated them and thus that Russia was certainly not interested in finding and stopping the perpetrators.45
Attribution is further complicated by the fact it is quite difficult initially to determine even whether a packet is intended to be “hostile” or not (as one expert put it, “Packets are not like ICBMs”). A DNS query, for instance, may be a legitimate look up attempt or it may be an attempted penetration. And, even in the latter case, it may be a legitimate DNS lookup or it may be part of a large scale DDoS attack. On top of this, once malware enters into a system, it does not necessarily bear a telltale sign of where it was designed and sometimes even what its actual intent may be. Unlike fissionable materials, for example, where each nuclear reactor has a distinctive “signature” that typically permits tracking the origins of the material, even when malware is uncovered it usually does not point a finger at a particular culprit.
The long-running effort to figure out the origin and intent of the Conficker worm illustrates this problem. The worm was one of the most effective in recent history, assembling a global botnet of some 7 million compromised computers in networks that ranged from the British Parliament to Southwest Airlines to the French and German military to scores of computers in China (one security expert called it “the Holy Grail of a botnet”). Yet, even as the worm spread, investigators could only sift through the data for clues to its origin, which still left them uncertain. In one breakthrough, they found certain programming linked to a Ukrainian language keyboard. But, even then, they could not conclude whether this meant the malware was designed in Ukraine or whether the clue was a clever attempt at misdirection.46