Lecture 3: Pointing Devices and Fitts’ Law Brad Myers 05-440/05-640: Interaction Techniques Spring, 2016 © 2016 - Brad Myers

Topics Note: “pointer” or “pointing device” (not “mouse”) Types of pointing devices Various properties: Direct vs. indirect Cursor? Relative vs. absolute Rate vs. position controlled (esp. for joysticks) How many states supported? (hover?, not-pressed?) Single or multiple touch Measuring effectiveness Important measures: Speed, accuracy, comfort, learnability Fitts’ law © 2016 - Brad Myers

Some Pointing Devices (List from HW#1): Others Mouse Laptop touchpad IBM Pointing Stick on Thinkpad laptops Touchscreen with fingers (phones, tablets) Touchscreen with Stylus Wii controller pointer in the air, pointing at a web page on a "smart TV" Microsoft Kinect using your hand to point at a web page on a "smart TV" Large "Smart Board" direct touch wall-size display Game controller connected to a PC to control the cursor Contour's "RollerMouse Red plus" Trackball Others Step keys Light pen Laser pointer at a screen Joysticks Game consoles © 2016 - Brad Myers

“Cursor” Same word used for both pointer position and text input position Are usually different Text cursor Will discuss with text entry lecture Pointer’s cursor often changes shape to show next action “Cursor” in this lecture Often no pointer cursor if direct touch © 2016 - Brad Myers

Light pen Starting ~ 1950s Sketchpad, 1963 Camera in pen looks for pixel on screen Calculates position based on timing Disadvantages: Need to hold hand in the air Low resolution Photo credit: https://design.osu.edu/carlson/history/lesson2.html © 2016 - Brad Myers

Step Keys Usually to move text cursor Sometimes can move pointer cursor as well Microsoft Windows, can move window with ALT-Space, “m”, arrow-keys Moves pointer cursor as well © 2016 - Brad Myers

Mouse Bill English and Doug Engelbart credited with the invention of the mouse (1967) Comfortable, fast and accurate Engineered to not move when push a button Cursor moves straight as hand moves in an arc Various gain functions Pixels per cm © 2016 - Brad Myers

Joysticks Often used in aviation Computer input for games Most are “rate controlled” self-centering springs Movement controls rate of cursor movement “isometric” – stick doesn’t seem to move Position controlled Absolute position of stick controls position No spring – stick stays where leave it © 2016 - Brad Myers

IBM Pointing Stick See lecture from 2014 from Ted Selker “Trackpoint” Rate-controlled isometric joystick Significant iteration and experimentation “10 years of human factors work” Material of the stick matters Placement of buttons Transfer function Training helps performance © 2016 - Brad Myers

Tablets Rand Tablet: 1964: http://www.rand.org/content/dam/rand/pubs/research_memoranda/2005/RM4122.pdf Handwriting and graphics input CAD/CAM Move stylus or puck or finger on a surface – not a display screen © 2016 - Brad Myers

Touchscreens Stylus versus finger How hold a tablet? Tradeoffs: Naturalness Accuracy Amount of content that can fit on the screen Blocking the view of the content How hold a tablet? Palm & hand removal? © 2016 - Brad Myers

Large Touchscreens Projector or TV Vertical or tabletop Smartboard Original Microsoft “surface” Single or multiple people Classrooms, museums, small meetings, etc. © 2016 - Brad Myers

Remote interaction with large screens Examples Laser pointing at screen Brad A. Myers, et. al. "Interacting At a Distance: Measuring the Performance of Laser Pointers and Other Devices." CHI'2002. pp. 33-40. pdf. Wii controller Microsoft Kinect – just fingers Very inaccurate Tiring © 2016 - Brad Myers

Issues Direct vs. Indirect Relative vs. absolute Direct – point at a screen Light-pen, stylus, finger Often no visible cursor More natural Indirect – movement moves a cursor Must be learned Issue – direction of movement (up or down is arbitrary) Transfer function – can be more accurate and faster Must have a cursor – Shape can be used as feedback Software can move position of the cursor Relative vs. absolute Relative – only movement used = mouse Absolute – required for touchscreen Touchpads can be either Image credit: http://www.drdigitizer.com/learn.html © 2016 - Brad Myers

Issues, 2 How many states supported? Single or multiple touch Can have “hover” state? Pointing but not “pressed” or “selected” Common with mouse, not with touchscreen Is possible with magnetic stylus on touchscreen Location but not pressed How many buttons? Stylus buttons on side of pen Other dimensions, like force of press – “3D touch” Single or multiple touch Resistive vs. capacitive vs. other sensing For big displays – which person’s hand? © 2016 - Brad Myers

Testing How decide whether an input device is “better” than another? Important measures: speed, accuracy, comfort, learnability As usual in HCI, need to decide tasks and relative importance of measures Often (but not always) will have tradeoffs Focus on speed and accuracy Since are numeric, can use standard statistical measures JMP, R, SPSS, Excel, and others to help with calculations Fitts law Slides by Jeffrey Rzeszotarski © 2016 - Brad Myers

Fitts’ Law 1954, Paul Fitts Figure out how quickly people could move a pointing device to a target and select it Predictive model e.g., keystroke-level analysis To compare pointing devices Throughput – combines both speed and accuracy

Cognitive Processes + Physical Processes = Pointing Performance

Laser Pointer Example Source: hcibook.com

Card, Moran, Newell studies Card, English, Burr 1978 paper – distance Time is linear for distance for step keys Time increases with the log of the distance for continuous devices like the mouse © 2016 - Brad Myers

Card, Moran, Newell studies, 2 Card, English, Burr 1978 paper – width Positioning time for both the mouse and the joystick decreases with the log of the target size © 2016 - Brad Myers

T = a + b log2(D/W + 1) T = time to complete a movement a = fixed cost to start/stop moving & click b = inherent speed of device log2(D/W + 1) = “index of difficulty” or ID bits Higher with distance (D) and lower with width (W) T = a + b log2(D/W + 1)

ID = log2(D/W + 1) “double the distance, double the width” equals

Card, English, Burr 1978

Homework 1 Analysis You can actually do this with your data! The Fitts’ Tester spreadsheet (on Blackboard) roughly follows Card et al. experiment Compute IDs from Distance and Width Chart them and see if they are linear Use Excel “Fit Trendline” to get coefficients for a and b and compare to other papers

Homework 1 Analysis Keep in mind that the papers you read collected MUCH more data E.g., 1200 to 1800 trials (four to six hours) before learning curve flattened out – [Card, 1978] It’s okay to not have any measurable difference as long as you explain why that’s reasonable Check out the error rates too The laser pointer paper is good example of how to structure a report

Newer Fitts’ Law tests Use circles instead of two rectangles ISO 9241-9 standard Doesn’t fit as well on non-square screens Horizontal and vertical movements may not be equal difficulty Muscles used Card, English, Burr 1978 paper showed differences for joystick, etc. but not mouse Laser pointer study: up to 10x more wiggle vertically Device properties Contour's "RollerMouse Red plus" © 2016 - Brad Myers

Homework 1 Questions? Tom Moran Bill Curtis, Stu Card, and Allen Newell Source: SIGCHI Archives