COMS W4170 Interaction Devices 2 Steven Feiner Department of Computer Science Columbia University New York, NY 10027 October 31, 2017 1 Stimulus Response (S R) Compatibility Whether mapping of user stimulus (e.g., limb motion) to system response (e.g., cursor motion) is compatible with regard to Compare orienting your mouse Direction Orientation like this vs. like this 2
Control-to-Display (C/D) Ratio [more generally, Control-to-Response (C/R) Ratio] Ratio of movement of input device (user stimulus) to movement of controlled display object (system response) for multiplicative transfer function Alternatively, C D gain = (C/D ratio) 1 High C/D ratio accuracy Low C/D ratio speed, space savings Notation varies! You will see {CD, C D, C:D, C/D} {ratio, gain} 3 Control-to-Display (C/D) Ratio [more generally, Control-to-Response (C/R) Ratio] Variable C/D ratio E.g., typical mouse transfer function high-speed low C/D ratio low-speed high C/D ratio Mouse Properties: Enables variable C/D ratio in Windows (on by default) G. Casiez and N. Roussel, UIST 2011 http://libpointing.org Note: Graph shows C D gain, not C/D ratio 4
An Early GUI Framework: Interaction Tasks J. Foley et al., 80s Basic interaction task is the input by user of atomic unit of info Text Select Position Quantify Orient Path 5 An Early GUI Framework: Interaction Tasks J. Foley et al., 80s Basic interaction task is accomplished by basic interaction technique Approach for performing basic interaction task using logical device Abstraction of a device based only on its output String, choice, pick, locator, valuator, stroke implemented by physical device(s) Actual device with characteristic properties Keyboard, mouse, rotary control, linear slider, joystick, touch screen, touch pad, scroll wheel, buttons, 6
An Early GUI Framework: Interaction Tasks J. Foley et al., 80s Composite interaction task combines basic interaction tasks Dialogue boxes Specify multiple units of information Construction Create objects Manipulation Modify objects (e.g., by scaling) 7 An Early GUI Framework: Interaction Tasks J. Foley et al., 80s Problem: The notion of a basic interaction task accomplished by a single logical device is an oversimplification Best suited to classical devices processed by simple computation What about Multi-touch? Gesture? Full-body? Face? Eyes? Pulse? Fingerprint? Emotion? 8
Simple Gesture Recognition $1 Unistroke Recognizer depts.washington.edu/madlab/proj/dollar/ / For each T (template) and each C (candidate to match) Resample point path to N equidistant points Rotate indicative angle (vector from centroid to first point) to 0 Scale to reference square and translate centroid to origin For each C (additional steps) Refine rotation to J. Wobbrock, A. Wilson, Y. Li, $1 Unistroke determine (heuristically!) Recognizer, UIST 2007 minimum distance depts.washington.edu/madlab/proj/dollar/ L. Anthony & J. Wobbrock, $N Multistroke between corresponding Recognizer, GI 2010 points of C and each T depts.washington.edu/madlab/proj/dollar/ndollar.html (GSS = Golden Section Y. Li, Protractor, CHI 2010 Search) yangl.org/protractor (More efficient template matching) R-D. Vatavu, L. Anthony, J. Wobbrock, $P Point Cloud Select T with smallest Recognizer, ICMI 2012 distance (best match) depts.washington.edu/madlab/proj/dollar/pdollar.html (Independent of # of strokes) 9 Keyboard: QWERTY First used on typewriter by C. Sholes, C. Glidden, and S. Soulé, early 1870s Depicted in US Patent 207,559, filed 1875) Note: No shift key. This early machine typed capitals only. 10
Keyboard: QWERTY Designed to prevent jamming of keys struck in sequence Based on digraph frequency analysis 11 Keyboard: DSK (Dvorak Simplified Keyboard) Developed by A(ugust) Dvorak, based on time-motion studies, n- graph frequency analyses Increase time fingers spend on home keys Increase alternation between hands, fingers US Patent 2,040,248, filed 1932 Vehement opinions for/against (e.g., http://www.dvorakkeyboard.com/, http://www.utdallas.edu/~liebowit/keys1.html) But, is it faster? 13
Keyboard: DSK (Dvorak Simplified Keyboard) Checkered history of comparison studies with QWERTY 1943 US Navy studies show DSK faster 1956 US General Services Admin study (Strong) shows QWERTY brushup practice more effective Original data destroyed, possible experimenter bias Theoretical comparisons suggest 2.3 17% improvement Studies of simulated typing based on digraph speed Anecdotal practical comparisons claim improvement Try it yourself https://www.howtogeek.com/263849/how-to-switch-todvorak-and-other-keyboard-layouts-on-your-computer-orphone/ 14 Keyboard: Half-QWERTY E. Matias, I.S. MacKenzie, and W. Buxton, INTERCHI 1993 Builds on User s knowledge of QWERTY L-R symmetry Try the demo: http://www.matias.ca/halfkeyboard/demo 15
Keyboard: Half-QWERTY Keys of one hand remain the same Spacebar acts as shift to mirror image keys Press release of spacebar with no other char before timeout is a space after timeout is a no-op Modifier keys (e.g., shift, ctrl) work as sticky keys Press once to modify next key only Press twice to lock and press again to release 16 Keyboard: Half-QWERTY Naïve speed prediction for a keyboard layout Two hand: Initial cap in sequence of caps = 2 strokes One hand: Initial flip in sequence of flips = 2 strokes, Each cap = 2 strokes Determine strokes for each keyboard layout typing a sample corpus Speed of KeyboardX relative to two-hand QWERTY = (strokes for two-hand QWERTY) / (strokes for KeyboardX) Graph shows two-hand keyboard strokes normalized to 1 E Matias, IS Mackenzie, and W Buxton, Onehanded touch typing on a QWERTY keyboard, Human-Computer Interaction, vol. 11, 1996, 1 27. (This article also has a richer analysis of typing speed and the results of extended testing by a small number of users.) 17
Keyboard: Half-QWERTY Study: 10 QWERTY typists, within-subject; at most one 50-min session/day (2-hand pre/post tests, and 1-hand blocks); eyes blocked Exceeded hunt-and-peck performance after 3 4 hrs; after 10 hrs, all reached 41 73% of their two-hand speed E Matias, IS Mackenzie, W. Buxton, INTERCHI 93 18