Why Are There No Cyber Arms Control Agreements?
Erica D. Borghard and Shawn W. Lonergan are research fellows at the Army Cyber Institute at West Point. You can follow them @eborghard and @Shawn_Lonergan.
During the Cold War, when nuclear-armed superpowers faced concerns regarding crisis instability and escalation, they entered into arms controls agreements. Arms control regimes can alter the military incentives for the use of offensive technologies; limit the damage to states in the event these technologies are used; and generally contribute to stable interstate relations, even between adversaries. With the emergence of a militarized cyber domain that creates the conditions for misperceptions that could lead to inadvertent conflict, why are there no cyber arms control regimes?
Nonproliferation, Arms Control, and Disarmament
Traditional arms control regimes are inapplicable to cyberspace for four reasons: it is difficult to measure the relative strength of states in cyberspace; there is uncertainty regarding the military effects of cyber technology; the challenges of monitoring compliance; and difficulties with enforcement.
Arms control regimes require states to have a basic understating of each other’s relative strength so that an agreement can promote strategic stability. For conventional munitions, nuclear weapons, or even chemical ordinances, warheads or pounds of a virulent gas that a state possesses can be counted, allowing others to assess comparative strength. The same cannot be said for assessing relative strength in cyberspace for two reasons. First, how can a state possibly count virtual weapons that by definition cannot be destroyed and, in theory, could be continuously regenerated? Second, unlike nukes or tanks, some cyber weapons lack universal lethality—unique tools and accesses are often required to deliver effects against specific targeted systems.
In contrast, what may be more readily measurable in cyberspace is the technical skill of cyber threat actors—a qualitative, rather than quantitative measure of capability. However, using relative skill to drive arms control regimes may be impractical due to the difficulty of crafting or enforcing agreements that limit skill or access to technology, particularly when governments are not the sole proprietors.
Arms control regimes may also form if governments are able to make reasonable calculations regarding the likely military effect of technological changes. However, the rapid and unpredictable pace of technological innovation in the cyber domain complicates these assessments. At the tactical level, attack vectors and offensive capabilities are continuously evolving, in contrast to the nuclear arena where innovations had long development timelines and could often be observed. The lag time in nuclear innovation gave states breathing room to adjust arms control agreements or develop other means, such as tailored intelligence or their own complimentary programs, to mitigate the fears posed by technological advances. In cyberspace, the open-ended promise of innovation coupled with quickly changing tradecraft that can emerge with little to no warning challenges the creation of any agreement. A cyber arms control agreement runs the risk of being outdated or restrictive in some unanticipated way before the ink has even had time to dry.
Even if states are able to calculate relative capabilities and assess the military implications of a technological innovations, cyber arms control agreements are unlikely to form if governments cannot detect cheating. The verification problem contains two prongs: being able to ascertain the size of a state’s arsenal and monitoring it to ensure future compliance.
Nonproliferation, Arms Control, and Disarmament
Ascertaining compliance in the cyber domain would require participants to agree to intrusive access to government networks. Malicious software can be developed just about anywhere, meaning that any verification mechanism would require a government to open up all of its networks to inspection. It would be unfathomable for one state to allow another, or any outside actor, to have unfettered access to its networks. Such access would provide an external party with critical information about vulnerabilities and potential exploits, and potentially violate the agreement it is attempting to enforce.
A less invasive means to assess compliance would be to have monitoring occur through national technical means of intelligence. During the Cold War, spies analyzed imagery collected from satellites that monitored the nuclear posture of another state. The equivalent in cyberspace would be to use cyber espionage to collect information about another state’s internal networks. However, while satellite collection is entirely passive, gaining access to and potentially absconding with data from sensitive government networks is not. If a state observes a third party penetrating its networks, it may be unable to distinguish between routine espionage activity for the purposes of monitoring compliance, other legitimate espionage purposes unrelated to compliance, or preparation for an offensive operation. This could prompt the targeted state to respond in an escalatory fashion. Therefore, the inability to perceive intent could, again, undermine the very stability the arms control agreements were meant to create.
Finally, even if the preceding obstacles could be overcome, enforcement of any arms control agreement would be difficult to implement due to two factors: problems associated with attribution and divining a proportionate punishment. First, in the event of a violation, states would have to attribute it with a level of confidence that would justify a reciprocal response. While attribution capabilities have unquestionably improved over time, not all states have the same attribution capabilities or enough confidence in them to justify action. This is particularly relevant given that a state that detects a violation would need to convince other parties to the same treaty that a violation occurred.
Second, enforcing an arms control agreement requires proportionate responses to observed derogation. This is problematic for several reasons. There may be a significant time lag between when a derogation occurs and when it is actually observed. Thus, the deterrent effect of a response is likely to be diluted by the simple passage of time. Furthermore, resource and access constraints may limit the capabilities a state has at a given moment to respond, which means that it may not necessarily be effective. A potential alternative could be to use non-cyber elements of national power to punish a derogation of a cyber arms control agreement. However, crafting an effective response that relies on physical elements of power may be difficult to formulate if the violation only caused virtual damage.
This pessimism regarding the feasibility of cyber arms control agreements does not imply that there are no avenues for cooperation between cyber adversaries. For instance, confidence building measures, which are a step short of arms control, may go a long way toward facilitating cooperation and communication and, ultimately, stability between rivals in cyberspace.