The underlying VoodooIO implementation has the distinguishing characteristic of supporting fast control presence detection and unique identification. The physical design makes adding and removing control from the substrate material almost as easy as manipulating them. From an interaction perspective, the question is: can the ability to quickly detect control presence be useful as an interaction technique in its own right? Attachment/removal of controls is a very explicit action already charged with a certain amount of expressiveness, and suggests that detection of control presence could be used as an interaction technique akin to control manipulation. For example, a par- ticular control could be used to physically encapsulate application concepts (e.g. as in [48] and [70]): when a particular control is attached, its associated concept is activated; manipulating the control modifies parameters associated with the concept, and remov- ing the control deactivates it.
In order to test this concept, an experiment was designed which compared different ways to control a software drum machine. The drum machine involves a small set of percussion instruments: a snare, base drum, cymbal and shaker. Each of these can be
turned on and off separately, and adjusted in their speed. Figure 4.9 captures the exper- imental setup with a VoodooIO substrate overlaid a desk surface, a set of controls, and speakers for output.
Figure 7.22: Experimental setup: participants interact with a software percussion machine.
7.5.1 Experimental Setup
The application was used for a comparative study of three interface conditions, shown in Figure 7.16. In all conditions, the percussion instruments were associated with a dial to control their playback speed; each dial was clearly labelled with its cor- responding instrument. Across the conditions, the LED beacon on each of the dial controls was used to provide localised feedback on whether the instrument was “on” or “off.”
Figure 7.23: Experimental conditions: "Button," "Dial," and "Presence."
Each condition differed in mapping of “on/off ” to physical controls: In the first con- dition, this was mapped to an additional button control, in the second it was overlaid on the rotating dial (i.e. “off ” mapped to “zero speed”), and in the third it was mapped to control presence (i.e. instruments turned on by insertion of the associated control on the substrate). These conditions will be referred to as Button, Dial and Presence, respectively.
The study was conducted with 20 participants, including 10 men and 10 women, aged 16 to 48, mostly very familiar with computers, but with varying background in music. All users engaged in test of all three conditions but in varying order. For each condition they received a brief introduction and were then asked to perform a series of small tasks, such as creating a rhythm with two instruments at the same speed, fol- lowed by a rating of the interface on a Likert scale. After completion of all three con- ditions they were asked to rank the conditions for various criteria, and to comment more generally on what they liked or disliked with particular conditions. The test ses- sions were also video-taped and transcribed to collect anecdotal remarks.
7.5.2 Experiment Results
Figures 4.11 and 4.12 summarise the quantitative results. The Button condition, which is also the most common in comparable to regular audio hardware, and the Presence condition consistently rated higher than the Dial condition which can be at- tributed to the Dial prohibiting separate control of on/off and speed.
Figure 4.11. Likert scores for usability of the Button, Dial and Presence conditions.
It is noteworthy that device insertion and removal are evidently seen as equally us- able for direct control of an application parameter as more conventional means. The cumulative ranking scores provide more detail on relative preferences. Overall, the Button condition is seen as best suited for the task and in particular providing best control. The Presence condition is rated relatively low on task suitability, but valued as easiest to understand, most confident to use and causing least mistakes. This is also backed up by user comments hinting at the simplicity and transparency as particular advantages of this condition. One user drew a parallel to WYSIWYG, pointing out “what you see is what you hear”.
Figure 4.12. Ranking scores for the tested conditions.
7.5.3 Discussion
The results of the experiment are interesting: the fact that, in terms of usability, add- ing/removing controls to turn on/off application parameters was widely accepted as an interaction technique is a powerful result. This concurs with some of the findings from tangible user interface research, which establishes the practice of having tangible ob- jects represent abstract functionality or data [26]. In the case of VoodooIO, each "ob- ject" also comes with a dedicated control for the concept it represents, and which can intuitively be used to manipulate a parameter associated with that concept. The results suggest that users can equate the act of removing a control to disables its associated concept from the application.
The VoodooIO design provides a clear sign of when a control is "in play" or "out of play" – there is little ambiguity about this, as the two states of being attached or de- tached from the surface are explicit and mutually exclusive. This is validated by the user feedback, where "Presence" condition scored so highly in the "Most confident to use," "Caused least mistakes," and "Easiest to understand" categories. One user even commented on the pleasure she gained from attaching controls from the substrate, making the analogy to "walking on snow." However, the effort of taking controls in and out of the structure does require more effort than is necessary to push a button, and for this reason the "Button" condition probably won over, seemingly better suited for an action which users were asked to perform repeatedly over the course of the study.
To sum up the most interesting results, and their applicability to the design of flex- ible user interfaces: the VoodooIO interface allows users to attach and detach physical controls to and from substrate areas; the system is able to quickly detect changes in control presence, and the result of our study show that users can equate adding/remov- ing controls as an interaction technique akin to control manipulation. The possibilities for user interface design include the use of physical device attachment and detachment as expressions in the dialogue between user and application, dynamic mapping of vir- tual interface elements to physical ones, and ad hoc creation of composed devices. The following section explores these interaction techniques in more detail, by using control presence as a central concept in the design of interfaces for musical performance.