The problems addressed here derive from a confluence of factors: an increasing social dependence on information-based infrastructures, an increasing complexity of those infrastructures that makes it difficult to anticipate all their failure modes, a growing number of people versed in information technology who can harm information systems, and tools readily available to assist in carrying out malicious acts. These have resulted in alarming rates of growth of system malfunction and system intrusion.
Conclusions
1. The Internet operates by means of a voluntary but structured process that provides the capability, in principle, to respond to chang- ing technology and user needs, including enhanced security. But since the process is driven by its developers and users, the incorporation of security features in the Internet is not assured, however valuable that may be from a public policy standpoint.
2. There are a number of existing techniques computer and infor- mation system owners and operators can utilize for their protection. These include firewalls, virus protection software, one-time pass- words, encryption, and virtual private networks. They also include instituting and enforcing security policies, real-time and off-line au- diting of system operation and use, penetration testing, and imple- menting data and system backup practices at all levels. But enhanced security brings associated financial and operational costs that can result in lesser levels of security than are technically feasible.
3. Intrusion detection systems are available today and are increas- ingly being deployed, but at the same time the number of system attacks, already large, continues to grow. Furthermore, tracking in- truders is difficult, slow, and uncertain. Hence the current state of affairs, where attack is relatively easy and defense is difficult or absent, leaves the balance very much on the side of the attacker.
4. Ongoing security R&D is pointing to ways of protecting infor- mation systems such that detecting, tracking, and identifying intruders will be possible with greater ease and certainty through collective actions by users and service providers. These include deploying new Internet protocols, level-of-service agreements making security a con- tractual requirement, the automation of advanced intrusion-detection systems for warning and tracking, the integration of security tools to provide more complete capabilities to meet wider ranges of needs, the creation of global incident response capabilities, third-party clearing- houses for secure and anonymous communications among incident responders, the use of digital objects to better define ownership of and appropriate uses of information, and increased capabilities for net- work traffic analysis.
5. Advances in intruder detection and tracking will aid in deter- ring attacks by increasing the risk of being caught. They can also be expected to reduce intrusions on user privacy implicit in current track- ing techniques.
6. International agreements, both informal and formal, will be needed if information infrastructure users are to receive greater pro- tection than they can reasonably provide for themselves. These include extending intrusion detection to operate across larger domains, de- velopment of new Internet protocols, coordinating international re- sponses to global incidents, shared R&D to keep pace with evolving international threats, and collecting and providing attack information to users in a timely manner to allow them to provide for adaptive defenses. The Draft Convention presented in this volume illustrates the kinds of steps that can assist in achieving these capabilities.
7. National policies that encourage the introduction of informa-
179 Current and Future Technical Capabilities
tion technology into critical infrastructures, thereby allowing systems of unlimited degrees of complexity and vulnerability to be constructed without corresponding increases in system security, should be exam- ined. It is possible that nations can encourage the evolution of their infrastructure systems in ways that will make them more robust as well as more capable.
8. Today’s information infrastructure, which has provided such dramatic improvements in access to information, can usefully be re- examined from the point of view of system architecture. What is needed is to overlay on the current information transfer network an assurance network that makes possible the definition and enforcement of standards of behavior among its users. This assurance network will involve both technical facilities to assist in protecting user rights as well as provisions for allowing operators of the assurance network to establish and maintain concomitant trust relationships that will be necessary for international cooperation.
Short-Term Prospects for Enhanced Security Are Encouraging
Protective actions will take time to implement. A central question then becomes one of relative rates. Since available defensive technologies have not been universally deployed, users can do a great deal in the short term to reduce their vulnerabilities. The pace at which short- term enhancements in system security can be made will depend on several elements: the acceptance by users and system operators that increased spending on security is needed, the deployment of available technologies by both individuals and organizations, improvements in current security products to make them easier to integrate, and the availability of new and more powerful security products and services. Looked at from this perspective, the picture over the next several years merits some degree of optimism because there is so much “low-hang- ing fruit.” The combination of defensive technology and operational
process redesign can accomplish a great deal in comparatively short times.
Long-Term Prospects Are Less Certain
More difficult to assess is how much society as a whole is willing to change in more fundamental ways. Will infrastructure operators re- alize that their rush to adopt information technology risks system failures that can only be addressed at the level of system architecture? Will utility regulators recognize that security must be on their agenda and that, without private sector initiatives, a more aggressive public posture may be called for? Will we recognize that deregulation without consideration of the architectural issues can have severe unintended consequences? Will law enforcement agencies increase their levels of investigatory and enforcement capabilities, and will legislators appro- priate the required resources? And will the nations of the world agree that the protection of the information commons is a shared responsi- bility?
The highly dynamic nature of information technology is a further complication in the long-term outlook for protection of infra- structures. New technology creates new vulnerabilities, and it in- creases the power of attackers as well as that of defenders. System and network security is not a problem that can be solved once and for all; it has the measure-countermeasure and offense-defense nature of mil- itary competition. From this perspective there is less reason to be sanguine.
A threshold issue for considering fundamental long-term changes in information systems will be that of weighing the cost of ignoring cyber attacks against the cost of actions to reduce the frequency and severity of their failures. There is no simple or obvious answer to this question. Because no fatal infrastructure failures have so far been induced by cyber attack, our only evidence of catastrophic failure is indirect. The rates of attack, and computer crime of many forms, are increasing, in some cases doubling annually. Such strong exponential
181 Current and Future Technical Capabilities
increases can rapidly dominate the balance. Unless it can be shown that these exponential growth rates will saturate at some comfortably low level, policymakers in both public and private sectors would be well advised to adopt conservative positions. It would seem prudent to invest now to hedge future downside risks.