Chi2014 rebuttal

We thank reviewers for their appreciation and feedback. To the best of our knowledge our paper is the first to make a comprehensive analysis of transparent mobile interactions, which reviewers agree is timely (R4) and highly needed (AC,R1,R3). Reviewers see value in the paper's many ideas, and highlight its potential to inspire the HCI community (R2).

We agree with the reviewers that our paper can greatly benefit from focusing on the 1) novelty of the interactions and 2) the analysis of the multitasking and photo-capture experimental results. As suggested by R1/R4 we will trim the design study, software implementation details and color mixing discussion in order to make room for our more significant contributions.

The transparent mobile interactions are the result of analyzing and synthesizing the initial set (+65 interactions) into 10 interactions grouped in 4 categories. The rationale behind our groupings relates to the technical capabilities needed to implement each group of interactions (AC). Naturally, some tPad interactions relate to techniques explored with other devices (AC). However, our contribution includes new techniques (querying, tap'n flip, surface capture), showcases tPad apps that benefit from them, and delineates future lines of research along the capability-based categories we defined. We will add a discussion section focusing on three aspects. First, we highlight the new techniques of which we believe tap'n flip and surface capture have the greatest potential. Second, we show how some tPad interactions relate to previous work (via a table, AC), including DoubleFlip & Tilt+Touch-in-Motion. Third, we discuss the intrinsic affordances of transparent mobiles which make existing techniques valuable and of a different nature when used on tPads. For example, placing the tPad on top of an object allows shifting from PACER's "camera pointer" to direct touch, which leverages tangibility and proprioception for object selection.

Our work demonstrates that flipping is a valid alternative for mode-change & multitasking. When coupled with a modifier, e.g. tap'n flip, it outperforms other methods and users preferred it. We will expand the description to clarify our set-up and extend our discussion. The positive results of tap'n flip can be attributed to the time gains observed: copying+flipping+pasting are aggregated into a single motion (we will add graphical representations of the temporal model for each technique). These advantages are greater with no navigation on the flip side. Thus, a tPad app with a tool menu on the flip side has performance gains of an order of magnitude of 50%, as shown in Fig 10 (left, 1 application). Another tap'n flip advantage is its implicit nature, requiring low cognitive efforts as revealed by the user ratings. These results mean that users are able to stay focused on the task and not on the interface. Finally, these gains surfaced even with our 7 inch tPad (mid-size) which made tap'n flip cumbersome. With smaller devices this gesture can be even more efficient.

Our work shows tPad's surface capture interaction can perform better (quantitatively and qualitatively) than current photo-capture. With tPad, users captured images faster and more accurately. Users also preferred tPad's surface capture. We will expand the experimental description and extend the results discussion. The faster capture can be attributed to the time saved by resting the device on the object. In normal photo-taking users must aim the camera, adjust the focus and correct for involuntary movement. With tPad users simply slide the device over to the capture target, without adjustments for focus or involuntary movement. Not only is tPad ~50% faster (higher interaction throughput) but it also offers other benefits. For example, normal photo-taking requires holding the device in mid-air which requires more physical effort and fatigue. Finally, tPad interactions were often single handed while photo-taking normally required both hands.

Other Issues

• Our categories reflect the scalable nature of our work, by highlighting the value of transparent mobiles at different levels of sophistication (from simple transparency, to surface capture displays). For example, overlay interactions can be implemented in existing transparent mobiles, while the others need additional capabilities.
• We did not focus on active reading (it is only one of 9 apps built) as this is a very specific and heavily researched area. We broadened our view to general interaction concepts and techniques.
• Transparent displays are an exciting topic for the industry and designers which disseminate work outside the academic channel. We find it necessary to reference at least some patents and design concepts in acknowledgement of this original body of work outside academia.

In summary, we are confident that we can improve and focus the paper significantly to make this interesting and timely topic available to the HCI community.