Software design beyond usable security

长期以来，计算机安全在系统设计和管理中一直很重要，但电子商务和社会计算意味着它现在是每个人都关心的问题。系统和网络基础设施的安全设计仍然是必不可少的，但现在也需要设计使每个人都可以通过越来越重要的软件来确保和控制自己的安全和隐私：密码系统和管理器、防火墙、反病毒监视器、网站有效性指示器等等。“可用安全”领域的出现，其重点是使安全软件更易于使用。虽然易用性很重要，但越来越多的人意识到问题变得更加复杂。这个研究项目将探索交互设计原则，使安全软件不仅易于使用，而且导致更安全的结果。除了适当的用户之外，攻击者也可以使用安全软件，并且设计必须在不给攻击者任何优势的情况下帮助适当的用户。密码系统的设计可能性提供了几个例子：小的密码空间保证了用户容易的内存负载，但允许对攻击者进行详尽的搜索；用户选择允许用户记住密码，但导致密码被攻击者猜到；密码提示可以帮助用户回忆起他们的密码，但也可以帮助攻击者。即使是诸如故障锁定之类的实用方法也会产生不良后果，例如拒绝服务攻击，并且仍然允许尝试访问所有帐户的广泛攻击。我们最近的研究已经确定了解决这些问题的几种策略，这些策略提出了新的设计原则。特别是，我们设计了一些技术来影响用户选择更强但更容易记住的密码，并找到了向合适的用户提供隐蔽反馈的方法，这些反馈对攻击者来说没有意义。提出的研究计划的目标是探索这些技术，开发一般设计原则，并解决更广泛的安全软件。

Computer security has long been important in system design and administration, but e-commerce and social computing mean it is now a matter of concern to everyone. Design for security in system and network infrastructure is still essential, but there is now also a need for design so everyone can ensure and control their own security and privacy via increasingly critical software: password systems and managers, firewalls, anti-virus monitors, website validity indicators, and so on. The field of "usable security" has emerged with a focus on making security software easier to use. While ease of use is important, there is a growing realization that the issues turn out to be more complex. This research project will explore interaction design principles for making security software not only easy to use, but also leading to more secure results. As well as the appropriate users, security software is available to attackers, and the design must assist the appropriate users while not giving any advantage to attackers. The design possibilities for password systems offer several examples: small password spaces assure an easy memory load for users, but allow exhaustive search for attackers; user choice allows passwords memorable to users, but leads to passwords guessable by attackers; password cues can help users recall their passwords, but can also assist attackers. Even pragmatic approaches such as lockout on failure have undesirable consequences, such as denial of service attacks, and still allow broad attacks attempting to gain access to all accounts. Our recent research has identified several strategies for addressing such issues that suggests new design principles. In particular, we have devised techniques to influence users to select stronger but memorable passwords, and found ways to provide covert feedback to appropriate users that is not meaningful to attackers. The objective of the proposed research program is to explore such techniques, develop general design principles, and address a broader range of security software.