1. Introduction

Information sharing and persistent data access is increasingly critical to people’s work and personal lives. Yet, corporate security policies rarely comprehend dynamic user models and people’s informal and persistent practices around ad-hoc sharing. This paper reports the results of a field study focusing on people’s practices around access control, security and file sharing.

Our study sought to understand three areas: under what circumstances do people or companies share or restrict access to files, what tools or behavioral norms are being used to do so, and how are people’s experiences, problems and needs changing in regard to secure file sharing and access control, especially in the case of geographically dispersed colleagues, clients, friends and family members?

We identified a number of key problems users face in sharing data:

· Sharing with myself: users are their own most common sharing partner, effortfully moving data between their own machines, accounts and devices in order to ensure continued access.

· Transient data: users often need to hold data only briefly while transporting it from one place or another; and that data may linger, be lost and forgotten.

· Transient access: users need to access data for only short periods of time – they intend only one-time access, or to make data available in certain situations.

· Over sharing: users grant more access than necessary when it is difficult to limit who has access to content or how much to share with others, or when pressed for time to extract information from larger data sets

· Ad-hoc sharing: users often share content with groups of recipients they have not shared with before, and may not again.

· Impedance matching: users spend considerable time and effort tailoring content for sharing based on their understanding of recipient needs or the demands of the sharing mechanisms in use.

Based on these insights, we propose that the general nature of the problem faced by users is what we term ad-hoc guesting: where users need to share data securely with unplanned sets of people with whom they have not previously shared who may belong to another organization, thus cannot be “named” by traditional access control. These interactions are transitory and lightweight, often not worth the effort required to set up new sharing mechanisms or change administrative state.

2. Background and related work

Our research builds upon a growing body of literature on file-sharing and access control. Previous studies have focused on personal file sharing, specifically, in the domains of music [3,4] or photographs [1,2], or professional collaborations in corporations [6], where email is viewed as the preponderant medium for file sharing [6,7].

Ahern et al.[1] investigated sharing preferences for personal photos over a mobile phone photo sharing network, and discovered that access control mechanisms were too coarse for many users’ needs. They found that end users often overloaded access control mechanisms in order to get around usability issues, such as making all