HCI & Security Systems Workshop Summary

Workshop Goals

This workshop sought to understand the roles and demands placed on users of security systems, and explored design solutions that can assist in making security systems usable and effective. In addition to examining end-users, this workshop also examined the issues faced by security system developers and operators. The goal of the workshop was to build a network of interested people, share research activities and results, discuss high priority areas for research and development, and explore opportunities for collaboration.

Key Points From Papers

Process

• Personnel costs and other business costs of security need greater attention
• Users’ security needs often differ from corporate security needs
• User vigilance differs between corporate and personal contexts
• Security and UI design must be integral in application development
• Assumptions underlying security policies should be empirically tested

User Perspectives

• Users view potential threats very differently from security experts
• Explanations at decision points are overwhelming to users and should be staged
• Complexities of security & privacy must be translated to user terms (e.g., privacy policies)
• Technologies like PKI are a poor fit to mental models of typical users
• User judgements of trustworthiness are often based on factors unrelated to security

Design

• People are central to security
• Radical adjustment of underlying technology may be necessary to meet user needs
• Secure interaction design can be at least partially formalized

Applications

• Credential recovery is a pressing challenge for existing systems
• Visualization can be used effectively for security tools (e.g., network monitoring)

Research Themes

"Transparency"

• Technologists use 'transparency' to mean the hiding of security mechanisms
• Users use ‘transparency’ to mean making security mechanisms visible
• Both have their place, but the technologist view is currently over- represented

Authentication

• Systems support delegation badly, so users are forced to share passwords
• Biometrics are not secret, and people generally don't understand that
• People don't understand the value and potential liability associated with biometrics
• Graphical passwords are promising, but may not scale across multiple systems
• Poor password practices have been quantified empirically

Operations

• System administrators face significant usability issues
• The issues are generally different from those faced by end users
• Administrators are often resistant to change, sometimes for reasons beyond their control

Design

• Security experts tend to prefer clearly defined objectives
• Problems examined by security experts are often not the security problems end users experience
• An important distinction is between education (presenting information) and training (forming habits through repeated action)
• Security is often perceived to be at odds with performance and ease of communication, and finding ways to reconcile the two is an important goal
• Do security solutions belong as tools, appliances, or built-in to the operating system?

Meta-Themes

Integrate Security & Usability

• Communication between security and HCI experts is essential
  • e.g., co-authoring papers
  • e.g., integrated design/review teams
• There is often a fundamental misalignment of research goals which should be discussed openly
• It must be an equal partnership, not a patronizing one
• The security community is currently fairly receptive to guidance in HCI matters

A Different Message for Each Community

• We need rich and compelling stories to foster understanding of HCI issues within the security community
• HCI people entering field should anticipate the need to develop this communication

Guiding Principles

• Security is for people to control, not to control people
• We need to fully characterize target users in specific contexts, not treat them as uniform and generic
• We should apply user-centered design to security problems to demonstrate
  • that it is possible
  • that doing so can improve upon what would otherwise be achieved
• HCI methodology compliments traditional security design

Unique Challenges

• Success in security is invisible
  • it's not clear when you're finished
  • the results are rarely gratifying in themselves
• Security is almost always a secondary task or goal
• Security designs must withstand active adversaries
• Security people like mathematical proofs, but there are none to be found in this area

Possible Follow-Up

• CHI panel
• Workshop and/or panel at a security conference
• Special issue of a journal or magazine