Research Abstracts - 2006

Facemail: Preventing Common Errors While Composing Emails

Eric W. Lieberman & Robert C. Miller

Introduction

Facemail is a project designed to investigate and prevent errors that users commonly make when writing email. We are modifying the mail composition interface of an existing email client to show the faces of the recipients in addition to the current display. We hope this addition will give the user more information about the email, and therefore he can make a better decision as to whether it is going to the intended recipients.

Problem Statement

Email is prevalent in all aspects of our life today, but in many ways it is still a flawed system. The most notable example is the problem of spam, and there is a wide body of research on detecting, filtering, and removing spam. There is also research dealing with email security, and how to make that security easier to use [1, 2]. However, there is a class of errors that users make that has not been well researched. In particular, there are three categories of errors this work examines:

1. Incorrect Recipients: Users often accidentally send emails to the wrong recipients. This situation can result from incorrectly clicking "Reply-to-all", mistyping an email addres, or having the wrong address entirely.
2. Missing Attachments: Users often forget to include attachments in their emails, leading to embarassing followup emails and delays in work. This occurs often because users are conditioned to type their email then click send, and they forget the extra, uncommon step of adding an attachment.
3. Obscured Secure Information: - Email is often copied back and forth between many parties using "Reply" and "Forward", and there is a risk that sensitive information could become buried and forgotten in the bottom of such an email.

Facemail is designed to examine these issues and come up with a user interface solution to this problem. We believe such a solution is necessary, because while most examples of these problems are simply inconvenient or embarassing, they can actually pose a large security risk when emailing sensitive information.

Approach

We propose a system that adds a display to the compose window of an email client that shows the actual faces of the recipients of the email that a user is composing. By adding this information, the user can quickly determine whether the email is going to the correct recipients, and he can correct any mistakes before the email is sent. This approach addresses primarily the first type of error discussed in the problem statement, which we believe to be the most important of the three.

There are several obstacles to implementing such a system, and they can be divided into three categories: modifying an existing email client, retrieving the face images, and finding email list information.

We chose to modify MIT's Webmail system using the Firefox extension Chickenfoot [3]. Chickenfoot allows a user to use scripts to modify a live webpage without necessarily looking at the underlying HTML. We use this capability to insert our display of images to the compose window of MIT's Webmail. Figure 1 shows such an insertion into the compose interface.

Figure 1: Inserting Faces into the Webmail Compose window.

To find the faces corresponding to each email address, we use many different search methods. We plan to search through websites such as Google Image Search, Flickr, or Facebook, and our design could also incorporate corporate directory servers or other "Face servers." The user will also be able to specify an image directly for a given email address.

For lists, we wish to provide the user with information about how many people are in the list, and ideally who exactly those people are. Figure 2 shows an example of a representation of a list of around 100 people. To determine the members of a list, we try many different methods. At MIT, we can use blanche or moira to find the membership of public lists. We can also use the webpages that Mailman provides for lists at CSAIL. At a company, a plugin could search a corporate database or other source for list information.

Figure 2: An example list image with around 100 people. The image contains a logo, the list name, and a representation of the faces of the list members.

User Study

In order to evaluate whether the idea of adding faces to email truly helps users make more informed decisions, I ran a user study where users were presented with scenarios in which they would send emails. The user would read the scenario, and then be flashed a quick view of an email interface that either contained or did not contain faces. We chose to flash the user the interface to simulate the brief glance that the user makes at the recipients field. Most users don't spend a lot of time looking at the To, Cc, or Bcc fields, and this lack of concentration causes many of the errors with incorrect recipients.

Duriong the flash, the interface showed either a "correct" email being sent to the right recipients, or an "incorrect" email being sent to the wrong recipients. The user then answered two questions about whether the email was going to the correct recipients. The results from my user study are shown in Figure 3, and we can see that when the user wasn't given enough time to fully process the interface, the faces provided much more information than simply the recipients lines.

Figure 3: Results from the user study.

Future Work

In the upcoming months we hope to add more features and plugins to our prototype, and then to evaluate its effectiveness. We plan to test the effectiveness of our "face gathering" techniques, as well as our ability to determine list membership. We also hope to have test users employ the Facemail system, and determine whether it is actually helping them to make fewer errors when composing emails. We also wish to explore user interface solutions to the problems of missing attachments and obscured secure information.

References:

[1] Alma Whitten and J. D. Tygar. Why Johnny Can't Encrypt: A Usability Evaluation of PGP 5.0. In Proceedings of the 8th USENIX Security Symposium. Washington, USA, August 1999.

[2] Simson L. Garfinkel and David Margrave and Jeffrey I. Schiller and Erik Nordlander and Robert C. Miller. How to make secure email easier to use. In CHI '05: Proceedings of the SIGCHI conference on Human factors in computing systems. pp 701--710, New York, New York, USA, April 2005.

[3] Michael Bolin and Matthew Webber, Philip Rha, Tom Wilson, and Robert C. Miller. Automation and Customization of Rendered Web Pages. In UIST 2005. pp 163--172. Seattle, Washington, USA, October 2005.