Article Preview
TopIntroduction
People use computers and mobile devices every day. They often unnecessarily disclose identity information and thus put their privacy at risk. As a result, identity information is collected and even maliciously used (Trend Micro, 2017). When computer warnings are displayed, most of the time, people ignore them (Akhawe et al., 2013). Some people do not even notice security warnings. On the other hand, our survey showed that people claim that they care about their personal information and keep it private (Zhu et al., 2012). This contradiction may be explained by the ineffectiveness of warnings (Kolimi et al., 2012).
Researchers have been studying the effectiveness of warnings for decades using questionnaires and interviews to analyze the effectiveness of warnings. Frameworks, such as the Communication–Human Information Processing (C-HIP) model (Wogalter, 2006), the sequential model of human information processing (Lehto, 2006), and the performance model (Lehto, 1991), have been used to analyze the information processing of warnings (Wogalter, 2006). In recent years, warnings have been integrated into web or mobile applications to improve users' attention to warnings (Balebako et al., 2013; Bravo-lillo et al., 2013; Maurer et al., 2011; Raja et al., 2011). Warnings designs were typically separated into two categories - active warnings and passive warnings (Egelman et al., 2008). Active warnings benefit from attracting users' attention by requiring users to complete some action, such as clicking a link or a close button. Passive warnings do not interrupt users’ workflow. They pop up and fade out after several seconds (Windows 10 action center warnings). Researchers have used eye trackers to study the effectiveness of warnings for more than 20 years (Anderson et al., 2016; Krugman et al., 1994; Vance et al., 2018).
However, previous warning research has limitations. First, eye trackers were used to investigate users' attention to warning messages to get quantitative information, such as the duration of eye gaze fixation on a warning message (Anderson et al., 2016; Vance et al., 2018). Users' attention and eye gaze information have not been used as input for displaying warnings. In other words, eye gaze was not used to control warnings' display and fade-out actions. Secondly, active warnings have a problem in that it abruptly interrupts users’ workflow and forced attention switches. Too many active warnings make users feel annoyed. In contrast, the passive warnings do not interrupt users’ workflow and make the attention transition more smoothly. It has drawbacks: they may not attract users’ attention and tend to be ignored.
Our warning is motivated by protecting users from non-essential disclosure of identification information. The novel design is motivated by keeping the benefit of passive warning and optimizing it. So that it can attract users' attention, not be ignored, and users won’t be forced to click on any button. User’s workflow would be more smooth and more effective.