Input Devices Matter

Transcription

1 Input Devices 1

2 Input Devices Matter Etch-A-Sketch Skedoodle 2 Images from

3 Tektronix From Buxton, et al, Human Input to Computer Systems: Theories, Techniques, and Technologies (not yet published) 3

4 Matching Tool to Task 4 Images from

5 Specific vs. General Devices Specific devices can be optimized for specific tasks Problems? General devices that can be adapted to many tasks 5

6 Types of Input Text Position 6

7 Text Input keyboards Two basic layouts for two-handed Qwerty Dvorak Better? 7

8 Qwerty vs. Dvorak Perceived Qwerty problems Many common letter combinations require awkward finger motions. (eg: tr) Many common letter combinations require a finger to jump over the home row. (eg: br) Many common letter combinations are typed with one hand. (e.g. was, were) Most typing is done with the left hand, which for most people is the weaker hand. About 16% of typing is done on the lower row, 52% on the top row and only 32% on the home row. 8

9 Qwerty vs. Dvorak Dvorak corrections Letters should be typed by alternating between hands For maximum speed and efficiency, the most common letters and digraphs should be the easiest to type. Thus, about 70% of keyboard strokes are on home row. The least common letters should be on the bottom row, which is the hardest row to reach. The right hand should do more of the typing, because most people are right-handed. Controversy around the speed differences 9

10 Qwerty vs. Dvorak Don t put too much stock in Wikipedia Problems are frequently perceived versus actual, and are based on a naïve model of typing Example: When you leave the home row, it can be good to stay off the home row Also controversy about speed differences? Actually there is none: There is absolutely no discernible speed difference between QWERTY and Dvorak People try to prove. Sometimes one faster, sometimes the other faster, majority of the time no difference If you know anything about science, this exactly implies that there is no discernible difference, and it is very highly probable that there is no difference at all

11 Ergonomics of keyboards To increase portability of devices, keyboards are frequently downsized: : Low-profile keys, projected keys, smaller keys All interfere with typing Much more significant a problem than Dvorak vs Qwerty keyboards 11

12 Keyboard Variants Soft/virtual keyboards Many ergonomic problems Feedback, resting of hands iphone significantly compromised However, increases the aesthetics of device by reducing thickness, overall size, and weight Good option when input can be significantly limited (e.g. mobile device, ipadas media consumption device) Bad if device requires text input (e.g. Surface pro: buy the type cover) ipad z10 Surface

13 Keyboard Variants Thumb keyboards physical virtual (ultra-mobile PC c. 2006) Surface Pro soft keyboard One-handed keyboards Frogpad Hold down space to shift hands 13

14 Keyboard Variants Chording Keyboards Doug Englebart Multiple keys together produce letter Very fast -- no targeting Twiddler is one example $199 Used for input by Thad Starner Currently working with Google on Google Glasses project Google glasses = another bad idea 14

15 Recognition-based text input Graffiti/unistroke/character Rubine s recognizer (featurebased stroke recognition) or some form of elastic template matching Handwriting recognition naïve Bayes classification algorithms Dictionary-based algorithms 15

16 Predictive Text Use characteristics of language to speed task Given characters typed so far, what letters are most likely to be next? Given characters typed so far, what could the word be? Examples Auto-correct T9 input Related techniques on smartphones 16

17 Text Input Expert-User Input Rates Desktop QWERTY 80+ WPM typical Record: 150 wpm for 50 minutes Mini-QWERTY (thumb keyboards) 60 WPM typical with training (Clarkson et al, CHI 2005) Soft keyboards 45 wpm T9 45 WPM possible for experts (Silfverberg et al, CHI 2000) Handwriting 33 WPM (Wilkund et al, Human Factors Society, 1987) Graffiti 2 9 WPM (Koltringer, Grechenig, CHI 2004) 17

18 Text-Input Recap A significant fraction of information conveyed to a computer is textual in form On desktop computers, keyboard is primary text input device Laptops may alter form, profile, or size in various ways to conserve space, which has drawbacks Off the desktop, many alternatives Frequently a compromise between portability and speed 18

19 Position 19

20 Absolute vs. Relative Position Absolute position implies direct mapping Examples? Relative position implies that change in position = change in screen position Examples? 20

21 Dimensions of Input More generally, there are a number of parameters to describe input devices: Absolute vs relative position Force stylus, joystick Linear vs rotary motion Mouse vs. trackball; knob One, two, or three spacial dimensions 21

22 Classifying Input Devices Need a taxonomy of input devices Example taxonomies from CS literature Logical or virtual input devices (Caruthers, Van den Bos, & Van Dam, 1977) Human perspective: (Foley, Wallace, & Chan, 1984) Physical properties: Card, Mackinlay and Robertson (1991) More recent and encompassing classifications by Buxton et al. ( ) Broader list of possible input devices arranged according to taxonomy 22

23 23 Classify by Function

24 Classify by Property Sensed William Buxton,

25 Classify by Property Sensed Card, Mackinlay, Robertson,

26 Taxonomies Recap Evaluate new kinds of devices and their appropriateness to a task Imagine new kinds of devices Example: 6 dof mouse 26

27 Expanding the Input Space Using multiple input devices Example: Benefits: Problems: Particularly useful for manipulating linked parameters Symtone Physical gestures Example: Benefits: Problems: 27

28 Recap Input is primarily text and positional in modern GUIs Text input performed by keyboards or variants of keyboards Positional input by pointing devices Full taxonomy of pointing devices can include many dimensions Can also look at increasing the space of input More pointing devices (e.g. one per hand) Other input techniques (e.g. physical gestures) Next Day: Input Performance Models 28