User-defined transfer functions to improve pointing performance in graphical user interfaces

Pointing at a target is the most fundamental and frequent task in graphical user interfaces (GUIs). Pointing devices like the mouse is the most popular and cheapest input devices for current desktop computers and the touchpad or trackpad is the best match between performance and demand of pointing devices for laptop computers. Due to the widespread and frequent use of pointing devices, even a small improvement in pointing performance can have a large effect on a system's usability. The physical movement of a mouse or dragging motion of a finger on a touchpad is translated into the movement of a pointer on the graphical display through so-called transfer functions. Transfer functions are actually the only pointing facilitation technique available to all users in modern days operating systems. Despite the importance of transfer functions, very little is known about the nature of the optimal transfer functions. Pointer acceleration (PA) is the default behavior on the Microsoft Windows and Apple macOS operating systems. It dynamically manipulates the Control-Display (CD) gain between the input device and the pointer as a function of the device's velocity. This mechanism has been implemented in modern desktop user interfaces and increases the CD gain as the user's hand or finger velocity increases. Previous work showed that the default functions of Windows, Apple macOS and Xorg (the X.Org Foundation server) have shown better performance compared to a constant CD gain. Apple macOS has the improvised performance for small target widths but reduces performance for covering the long distances. Current knowledge on velocity based transfer functions relies on evaluations of basic functions and adapting the CD gain in discrete or continuous ways using low-order polynomials. The internal details and design rationales of the transfer functions that we all use are mostly unknown. The aim of this thesis is to gain a deeper understanding of the optimal transfer functions and to assess them based on natural interaction with the system using a user-driven approach. The implementation part of the thesis is divided into two desktop applications, one is for recording all raw mouse and pointer movement as well as recording contextual information to understand a transfer function. The other implementation part of the thesis is to enable the user to define their own transfer functions. Users can customize existing default transfer functions and use them to control the pointer. These two applications are used to conduct a study that collects user device information and as well as user-defined transfer functions. Finally, we identify interesting transfer function for touchpads and mice