2017-2 Human Computer Interaction.fex

Introduction
========

user research method:
	formative: 
		helps to understand the problem and the users to inform design
	build: 
		create new version of the product using the formative methods
	evaluative: 
		help to understand how well design works
		may inspire new formative methods
human computer interaction:
	human: 
		single user, or friends, colleagues
	computer: 
		machine running program, distributed, heterogeneous
	interaction: 
		user provides input
		machine communicates result
		user interface main point of interaction between user / machine
	
human vs computers:
	strength of human:
		signal detection under noise
		recognising complex signals (images, voice)
		recognising complex configurations (scenes)
		concentration to the essential
		adaptation to unexpected situations
		learning aptitude
		memorizing cohesive information
		intuition
	strengths of computers:
		superior if problem can be algorithmically formulated
		detecting / recognising known signals
		fast and reliable reaction to known signals
		measuring and counting
		storing large amounts of incoherent data
		syntactic symbol manipulation
		reliable, fatigue-free repetition of operations
	
components of cognition:
	perception (using sensory systems):
		Sight (visual)
		Hearing (auditory)
		Touch (haptic)
		Smell (olfactory)
		Taste (gustatory)
		Balance and acceleration (vestibular)
		Body awareness (proprioception, kinesthetic sense)
		Temperature (thermo reception)
	memory:
		sensory
		short-term STM (controlled cognitive processes)
		long-term LTM (declarative, procedural knowledge)
	action (using motor systems):
		Arms, hands, fingers
		Head, face, eyes
		Vocal system
		Legs, feet, toes
		Jaw, tongue
	attention flow:
		stimulus
		sensory organs (eye, ear, ...)
		sensory register (visual, auditory, ..., may activates the motor system)
		symbol recognition (uses LTM and STM)
		LTM and STM may activate the motor system
	anatomy eye:
		light travels through cornea (skin around eye)
		crystalline lens make image sharp
		iris controls amount of light which enters the eye
		retina contains rods and cones
		rods see light (scotopic vision) but slow reaction
		cones see colour (photopic vision) & fast reaction, S M L types
		fovea contains cones and where humans see sharp
		optic nerve transports information to visual cortex
	acuity of visual field:
		foveal vision (fine details), about 2degrees of sharp vision
		peripheral vision (motion), visual acuity decreases with distances from fovea
		horizontal field 60deg nasal, 90deg temporally
		vertical field 60deg up, 70deg down
	usefulness of visual field:
		low character density 15deg
		high character density 1deg-4deg
	eye movements (gaze movements):
		saccades (repositioning of fovea, 30ms (10ms-100ms), bad perception)
		fixations (dwelling on a point, 230ms (150ms-600ms), 90% of the time)
	implications for perception:
		reading speed fixed (10words/sec)
		context dependance of gaze movements
		
model human processor (MHP):
	abstract understanding of perception, memory, motor system
	can be used to estimate execution time, error rates, training effects
	practice leads to skills which reduces cognitive effort
	three processors with associated memory and runtime:
		perceptual system for sensors & buffers
		cognitive system for working memory content
		motor system for movements
	total processing time:	
		overall system runtime is sum of all runtimes
		240 ms = t_perceptual + t_cognitive + t_motor = 100ms + 70ms + 70ms	
	perceptual processor:
		how it works:
			receives sensor signals and stores them in buffer (one buffer per channel)
			perception time about 100ms (50-200ms) t_p
		blochs law (for t < 100ms):
			Response = Intensity * TimeOfExposure
			can substitute intensity, duration and maintain constant effect
			enforces limits on framerate for animations and videos
	cognitive processor:
		how it works:
			operates on chunks of information
			processing time about 70ms t_c
			divided into STM and LTM
		STM:
			as working memory
			control of cognitive processes (concious such as reading & multiplying, automatic such as visual searching)
			repetition and building of connections stores information in LTM
			can process limited by number of chunks, but not complexity
			capacity 5-9 chunks for 2 sec, 2-4 for 7 sec
		LTM:
			as remembering memory, build on experience
			divided in declarative (facts, past events) and procedural (skills)
			capacity practically unlimited, but hard to organize for retrieval
	motor processor:
		how it works:
			controls and runs motor system such as hands, arms, ...
			processing time about 70ms
		open loop:
			70ms
			no perceptual control
			just motor processor commands
			in pen experiment the number of changes in direction
		closed loop:
			250ms
			with perceptual control
			motor processor corrected after input from perceptual processor
		
Fitts Law:
	predict movement time of rapid aimed movements
	measure throughput of movements (reaching for stuff, clicking on icon)
	very powerful, widely used, comparable over different experiments
	fitts thesis:
		fixed information-transmission capacity of motor system
		most limiting ability is the processing of sensory input
		tradeoff between speed & accuracy
		ID = # of bits required to specify movement
		MT = movement time
		IP = ID/MT index of performance
	formula:
		difficulty I_d = log2(2D/W) for W target width and D target distance
		movement I_m = 100ms/bit
		T_pos = I_m * I_d
	implications:
		doubling distance adds constant to execution time
		doubling target width is similar to halving distance
	bandwidth factors (performance of use):
		bandwidth of muscle group which uses device (human)
		precision requirement of task (application)
		bandwidth of input device (device)
	target acquisition:
		in intervals
		come close fast, then correct
		each proceeding is slower & more precise
	visual feedback loop (using MHP):
		t_p (observe hand position), t_c (plan movements), t_m (move hand)
		one feedback cycle is (t_p + t_c + t_m) = t = 300ms
		first cycle e*D moved, second e*(e*D), .. for e error, D distance
		movement stops when (e^n)*D <= 0.5W for W width of target
		resolve to n, use T = I_m * I_d to calculate I_m
	in practice:
		icons should be activatable at their edges (cause easier to hit)
		larger icons which are far away
		toolbars which appear directly between cursor
		upper-left corner buttons
		round menus rather than select
		
shannon:
	process:
		information source -> transmitter -> produce signal
		signal is modified by noise
		received signal -> receiver -> destination
	formula:
		C = H(signal) - H(noise) = W'log((S - N) / N)
	shannon formulation:
		A is amplitude (of information source)
		W is width (of target)
		a and b are empiric values, use linear regression
		movement time MT = a + b log (A/W + 1)
		information source is the amplitude, width is the noise
		determine a, b for device by varying A, W in experiments
		adjust the target width such that 96% hits
		
extensions of fitts law:
	more dimensions:
		UI's are mostly 2D, buts fitts only 1D
		can generalize the formula for 2D, 3D, ...
	angular motions:
		consider rotations instead of linear motion
		just replace A, W with angulars
	steering law:
		predicts time for trajectory movements
		fits nested menus, handwriting, ...
	continuous steering law:
		steering along curve
	
applications of fitts law:
	evaluate input devices:
		empirically find a, b for different devices
		compare devices
		leads to optimized mouses
	optimize user interfaces:
		empirically find a, b for UI
		optimize UI elements for predicted movement times
		leads to optimized keyboards
	does not include:
		body asymmetries (left vs right hand, flexion vs extension)
		parallelization strategies (two hands, multiple fingers)
		cognitive factors (reaction time, searching time, ..)

Interaction Paradigms & Computational UI Design
=========
	
UX milestones
	mouse 1964, usable at 1983
	touch sensing table 1985
	smartskin 2002
	idea growth: 
		invention
		refinement & augmentation
		traction (when its finally used)
	
interaction paradigms:
	commands: directly typing in commands with the command line
	dialogue systems: onscreen / speech-based dialogue systems
	searching and browsing: list / grid with all items
	direct manipulation rules: 
		visibility of objects and actions (self-explanatory)
		rapid, reversible, incremental actions (allows to recover)
		direct, visual manipulation of object of interest (eases memorability)
		
emerging interaction paradigmas:
	context sensitive UI: 
		knows where you are
		predicts your next action
	natural language interface: 
		speech commands
		AI & search still bad
	AI UI issues:
		not self-revealing
		difficult to understand what its capabilities are
		severe novelty effect (only used at the beginning)
		social acceptance issues
		high cognitive load
	augmented reality:
		quite old, but old phase was not successful
		next phase probably now

UX livecycle:
	costly, unrealiable, slow, subjective, non-accumulative
	multiple iterations often better than single idea
	getting the design right (refine existing design to perfection)
	and the right design (change approaches to overcome local minima)
	analyze: understand user works and needs
	design: create interaction design concepts
	prototype: realize design alternatives
	evaluate: verify and redefine interaction design
	
model-based UI optimization:
	designer proposes stuff (driven by design studies)
	optimizer generates model (driven by computer science)
	model is evaluated (by behavioural sciences)
	then use model to create optimal user interface
	in general:
		minimize sum:
			frequency of i used after k
			effort from j to l
			random variables i to j
			random variable k to l
			(random variables determine tested layout)
		solve with: 
			greedy (locally optimal choice)
			dynamic programming
			constraint programming
			branch and bound
		multi-objective optimization:
			includes additional possible factors
			performance (speed, accuracy)
			experience (satisfaction, aesthetics)
			effort (energy consumption, fatigue)		
	linear assignment:
		frequency factor:
			entry value
		effort factor:
			fix value 1
	keyboard:
		frequency factor:
			use dictionary
			use world frequency
		effort factor:
			with fitts law
			a + b log(D_ij/W_j + 1)
	menu-optimizer:
		adapt keyboard optimization
		include SDP:
			search (time to find item, increases linearly with items in menu)
			decision (time to decide upon item, determined by experiment data)
			pointing (time to point to right item, determined by fitts law)
		consistency:
			pull same items together
			penalize wrong positioned items
			remove not matching items
		frequency factor:
			number of items in menu
			position of target menu
			number of trials
			decision entropy
		effort factor:
			fitts law
		
experimental design
======

evaluation types:
	formative:
		early in the design process
		understand the user:
			sanity check the right thing is built
			observe and understand current processes
			generalize observations into functionality
			create scenarios for actual usage
		prototype interfaces:
			low fidelity techniques (paper, video)
			interactive (limited functionality in high level language)
	summative:
		check if solution works, has been improved, compare with other solutions
		measure quality:
			analytics (experts)
			empirical (measure how well task is completed)
		measure statements:
			less errors, faster completion time
			check if target metric is reached on average
		metrics:
			effectiveness, efficiency, satisfaction

study:
	cause and effect:
		cause has to precede effect
		cause and effect need to correlate
		all other possible explanations must be ruled out
	infer causality:
		experiment where cause is present (independent variable 1)
		control group where cause is not present (independent variable 2)
		choose dependent variable (like grade)
		show correlation
	characteristics for empirical:
		objectivity, reproducible, relevance
		external validity (good sampling, big sample sizes)
		
analyze data:
	how to analyze data?
	view data:
		display max, min, n
		mean (average over values)
		median (number at middle of value)
		mode (most frequent value)
		range (value range, max - min)
		check histogram (uniform, symmetric, skewed, bimodal)
	representing error:
		sum of squared error (sumof diff value to median)
		variance (divide by n-1)
		standard derivation SD (root of variance)
		standard error SE (SD divided by sqrt(n))
	t-test:
		difference mean_test - mean_control, divided by SE_test
	confidence interval:
		replicate experiment extensively
		persist confidence interval
		then do statistical tests with the confidence intervals
	writing up results:
		define statistical procedure
		describe samples with median, mean, confidence interval
		describe difference with statistical values
		visual representations
	
HCI studies
=====

survey of statistical problems in HCI
	60% fail to report data appropiately
	30% use wrong statistical assumptions (use non-applicable tests)
	30% do over testing (using same samples for different purposes)
	20% do inappropiate testing 
	for published papers, in medicine report failure at 90%!
	see wikipedia.org/wiki/Misuse_of_statistics
	why:
		p 0.05 value is arbitrary
		only results published (if something does not work, its not published)
		lack of awareness
	problems with bigger samples:
		more participants stabilize slowly, therefore costly
		better measurements can also help to increase power
	robustness:
		statistical analysis robust when similar datasets create similar results
		plot robust when difference clearly visible
		interpretation robust when similar datasets imply similar interpretations 
	possible solutions:
		prohibit usage of p values to draw conclusions
		researchers should refrain from final conclusions
		
study fundamentals:
	within subject design:
		everybody does everything
		each subject sees all independent variables
		need to permute presentation order to avoid learning effects
		full randomization implies n! groups, use latin square counterbalacing
		shift condition by one place per row, reorder rows, reorder columns
	between subject:
		only one condition per group
		each subject sees only one independent variable
		need to balance groups to account for different skill levels
		personal interviews or random assignment
	independent variables:
		what is compared
		categorical (unordered discreet, countries)
		ordinal (ordered discreet, month)
		cardinal / interval (continuous values, height)
	dependent variables:
		what is being measured
		precise, unambiguous measurements
		collect quantitative feedback (how many, how much)		
		difference between behavioural (what is done) and attitudial (what is said)
	extraneous variables:
		what else is varied, but can't be accounted for
		gender, age, nationality
		
how to do a study:
	goal of study (what to investigate):
		set a research question
	study design:
		define a goal, set a research question
		define how to evaluate your research question
		what is being compared (independent variables)
		what are they compared with (dependent variables, metrics)
		what is being varied (extraneous variables like age)
	subjects design:
		within design (but need to counter balance to avoid learning effects)
		between design (but need to account for different skill levels)
		define sampling (for example convenience sampling)
	measurement design:
		define metrics to measure your research question
		think about multiple data sources 
		find effects in different sources (triangulation)
	possible metrics:
		number of keystrokes, mouse clicks, page visits
		self-reported values, observations
		task completion time
		error rates
	standardized questionnaires:
		NASA TLS (perceived task load)
		SUS (system usability scale)
		UEQ (user experience questionnaire)
	study procedure:
		what participants need to do
		create a handout with detailed instruction
		define what has to do before the task
		how the task has to be done
		what has to be done after the task
		include open questions
		include section for experiment conductor (starttime, endtime, notes)
		include section at the end to say thanks
	study evaluation:
		define how the measures will be tracked
	study hypothesis:
		define whether qualitative (explain) or quantitative (measure)
		define hypothesis and their reasoning
	organisation:
		ask people personally if they want to sign up
		use a nice tool for them to register (for example typeform)
		shortlink to pass around
		confirm event personalized
		bring something to eat
		lottery if unable to pay participants
		"thank you" email with results of study
	write report:
		use standard format like apa6
		abstract for a short summary
		introduction section contains motivation, explains product / functionality
		methods section explains how study was performed
		results section visualize quantitative results (p values, graphs)
		discussion explains qualitatively the seen results (why, how to fix)

interaction design:
	applies	to desktop, mobile, web pages, ...
	aiming:
		support users rather than replacing them
		enrich the user experience
		10 minute rule (learn & use time for new users)
	identify users:
		all stakeholders
		study setting, not individuals
	general process:
		requirement analysis
		prototype continuously adapted by evaluation results
		evaluation with studies
	design process:
		establish requirements
		develop conceptual model (define interaction)
		produce rough models
		experiment with alternatives
	scenarios:
		informal narrative descriptions
		"real world" events and the resulting interaction
		develop into use cases
		maybe create storyboards
		establish requirements with customer
	sketches:
		quick, plentyful, disposable
		clear vocabulary (not an implementation)
		constrained resolution (capture only the concept)
		refinement keeps concepts
	prototypes:
		turn sketches into interactions
		can use paper (cheap, encourages creativity)
	workflow to working system:
		brainstorm different ideas (lot of sketches)
		choose idea, and create rough interface (sketch variations)
		task centered walkthrough (low fidelity prototypes)
		fine tune interface with heuristic evaluation (medium fidelity prototypes)
		usability testing (high fidelity prototypes)
		limited field testing, alpha/beta testers (working system)
	some targets:
		rewarding, aesthetically pleasing, motivating, helpful, enjoyable, satisfiable
		efficient & safe to use, easy to learn & remember, good utility
	design principles:
		visibility (show action user wants to perform)
		feedback (visual, audio, tactile reaction of what is happening)
		constrains (restrict actions to reduce errors)
		mapping (natural control-action connection)
		consistency (same workflows, grouping)
		affordances (give clue of what is going to happen)
		simplicity
	text:
		novice readers assemble words letter per letter
		advanced readers skip assembling
		avoid repetition and long paragraphs
		avoid centred text
	considerations:
		allow for colour blindness
		only small part is actually seen sharp
		attract attention with movement or colour to important information
		continuity (eye completes image)
		figure / ground (eye assumes something to be in the front)
		remind users of their actions (persist search query, use adaptive UI)
		don't expect users to remember (show current step / progress, cleanup)
		recognition easier than recall (shortcuts harder as finding menu element)
		provide visual cues (show preview, show past actions)
	visual hierarchy:
		organize content using size, prominence, content relationships
		vary text size, bold, order, grouping
		use spacing and hierarchy for structure
	gestalt laws:
		emergence (form complex pattern from simple rules)
		multistability (figure/background selection)
		reification (brain assumes more content as explicitly shown)
		invarriance (3D recognised in 2D)
		closure (complete shapes)
		similarity (similar objects form group)
		proximity (close objects form group)
		continuity (assume overlapping objects are simple shapes)
		
paper prototyping:
	steps:
		create user profile
		decide on task
		create prototype
		perform walkthrough
		plan study
	good task:
		critical for product
		goal that matters for user
		finite set of solutions
		clear ending point
	run study:
		facilitor (guides the process)
		computer (simulates the system)
		observers (document the study)
		helpers (provide help if asked to do so)
	construct paper prototype:
		list screens for each task and create them
		break down screen into elements (can be switched)
		hand-drawn vs screenshot mix
	good at:
		find issues with requirements
		detecting unclear concepts
		problems with navigation, workflow
		define documentation and help requirements
		find issues with layout
	bad at:
		find interaction issues
		input methods (like scrolling)
		response time
		animations
		
video analysis:
	why:
		observing users can be more revealing than asking them
		revisit scenes many times, focusing on different aspects	
		collaborative analysis
	captures:
		basic actions
		verbal communication
		facial expressions
		gestures, body movement
		gaze
	gain access:
		management (in their interests, no interference with work)
		workers (get insight into skills / issues, say not about optimising)
	ethical considerations:
		informed (everything which may occurs is disclosed)
		consent (rational, mature, voluntary agreement, free from pressure)
		formal informed consent required
		give time to ask questions
		inform detailed in public places
		agreement of parents in schools
		preserve privacy (blur, remove sensitive remarks)
		determine how data is stored (time, place, usage, access)
	recording:
		sample recordings
		wide angle captures scenes but loses details
		short-angle captures interaction but loses expressions
		external micro often required
	get users to speak:
		ask users to think aloud
		work in pairs to hear discussion of problems
	structuring data:
		preliminary catalogue, describing all interviews which took place
		transcribing segments for future reference
		transcribing visible conduct for future reference
		
touch:
	FBTouch:
		basic touch optimization, multi-touch extensions
		touch-enhanced interface (more spacing)
		touch event handling
		touch event tracking with easy matching
		touch event capture and delegation
		touch device detection
	W3Touch:
		logging, inspection, segmentation, adaptation
		rather poorly optimizes non-mobile sites for mobile

cross-device applications:		
	eye-free pen interaction:
		what:
			handheld mobile device which can change colours, erase, stroke, undo/redo
			use pen to "write" on whiteboard
			recognised stroke then projected on whiteboard
		study:
			compare three UI's
			eyes free uses touch gestures to command
			classic uses big buttons
			popup was displayed on the whiteboard
		study process:
			design (conditions, tasks, questionnaires, measurements)
			implement (study setup, adapt software, logging)
			pilot (test issues, time, instructions with pilot subjects)
			recruit (find participants, schedule appointments)
			run (keep conditions stable, collect data)
			report (statistical analysis, interpret results)
		study results:
			eyes-free & classic faster than popup
			no other significant results
		study problems:
			participants used classic UI also eyes free
			too few participants
			not so great logging
	cross device testing:
		what:
			can debug on different device connected by wireless
			shared css / javascript editor
		study:
			can't do proper study cause no time
			instead 2-hours lab study with inexperienced developers
			browser debugging tool as baseline
			prepare remote debugging
			two tasks to solve, one rather easy
			recorded video, questionnaire, measured time, code written by participants
		study results:
			participants liked tool
		study problems:
			tasks not well balanced
			not all features under test were used
			participants were bad at judging their skills
		
digital pen & paper:
	to bridge paper-digital divide
	automated form processing:
		pen strokes captured and stored on pen
		data uploaded to pc	
	paper point:
		control powerpoint with printout
		free pages for free drawing
	multimedia pen:
		capture audio, real time processing
	iPaper:
		create active areas inside pdfs and link to multimedia
	EdFest:
		tourist guide with paper documents & pen
		interactive event brochure
		can rate attractions, navigate on map
		noone cared
		lack of clear goals & hypothesis other than impress
		lack of collaboration with designers, systems people
		lack of frameworks made the application way to complex
		needed nearby laptop to function properly (lol)
	Print-n-link:
		use digital services to search/retrieve cited publications
	PaperProof:
		edit on paper, changes are reflected on computer
	iGesture:
		draw gestures on paper, which will then be executed on pc
	iTable:
		map projected on table, can be written on

adobe lightroom:
	easy to use photoshop
	task-based, modular
	different applications:
		wedding (post-processing, publish to webpage)
		landscape (well prepared, filters & post-processing)
		studio (well prepared, direct review, post-processing)
		wildlife (catching the moment, lots of pictures)
		sports (review in short time)
	first phase:
		user interviews (phone & personal visits)
		walking through the activities of recent photo shoot
		starting with high-level overview of last job, then communicating timeframe to discuss certain activity
	impact of first phase:
		general framework for workflows in the real world
		users were patching together wide range of tools
		found out which feature of PS are crucial
	define task based environment:
		used card sort to ask photographers how workflow should be structured
		0. cards with tasks written on are given
		1. exclude all cards which were not applicable
		2. add new cards with missing functionality
		3. grouping, duplicating & arranging cards
		4. creating names for the groups
		5. repeat 1&2 for feature cards
		6. add the feature cards to the task groups
		highly consistent workflows & features used could be shown
		summarized results on poster
	key features lightroom:
		workflow top right (library, develop, slideshow, export)
		user driven attributed (mark images)
		faceted search & metadata (search by tags, labels, attributes)
		in place image manipulation (marking, moving)
		different visualization (fast change UI layout)
		organize images into folders
		select images based on ratings
		compare images
		simple post-processing
		export as slideshow, print, web
	online interview:
		general background questions
		scalar agreement questions (likert scale)
		multiple choice questions
		open questions
	online interview considerations:
		personal information can be sensitive
		ask at the end of questionnaire as user more likely to answer
		provide slider/ranges where applicable
		recruited using social media
		offer reward if it does not distort subject group
		as short as possible
		only few open questions
		no jargon
		what to do if running issues contains mistakes
	study results:
		browsing folders more common than search
		keywords, smart collections not often used
		
communications:
	everybody communicates; denying communication also communication!
	not only worlds but also meta-info exchanged depending on speaker and target
	rules of conversation:
		next speaker starts talking, current speaker continues or chooses next
		next speaker chosen by asking question, making request, inviting opinion
	irritation:
		if rules disobeyed, interrupt constantly, ignore cues
	specifying length of talk:
		state length explicitly, or finish with explicit sentence
		shift body, change focus, use gestures
	adjacency pairs:
		multiple concurrent conversations in one using phrase pairs
		first phrase determines which conversation continues
	new forms of communication:
		e-mail (clear turn taking, one to many, recipient controls pace)
		emoticons (clearly communicate intention of sender)
		social media (supports, relationships, lightweight conversations)
		rich media (face-to-face communication like video calls)
		lean media (text-only chat)
	collaboration:
		task-related goals with known team of small size
		collaboratory are big collaboration groups
		emails, calls, conferencing, shared files
		goal-directed, time-limited, identified parters, assigned tasks, cross-review
	effective collaboration:
		abolishing emails may help
		social media for workplace
		simple tools as doodle can be very effective
	crossover:
		wikis, blogs, chats
	social media:
		tagging, rating, review
		large scale
		playful, unknown parters, act independently
	crowdsourcing:
		large-scale collaborations
		unclear motivations, quality, ethical, social implications
	effective crowdsourcing:
		use algorithms which allow concurrent editing, undo/redo
		use privacy sensitive environment (explicit push of changes, not WIP)
		awareness of others work helps to avoid conflict
	awareness mechanism:
		awareness visualization (displays who did what)
		awareness computation (finds out who did what)
		consistency algorithm (sync stuff)
		structured document (allows concurrent editing)
		operations (persist actions performed)
	organisation forms:
		face to face interactions:
			same place / same time
			use presentations, whiteboards
		continuous task:
			same place / different time
			project management, team rooms
		remote interactions:
			same time / different place
			video, chat, screen share
		communication / coordination:
			different time / different place
			emails, blogs, shared files, forums
		
ambient information:
	inform without distracting (like steps in house warn of arrivals)
	easy switch between periphery and centre of attention
	nice to know, non-distracting, push rather pull
	communication zones:
		device with interaction, notification, ambient zone
		active zone depending on distance of observer
	examples:
		informative art (image different depending on ambient)
		info canvas (user chooses visualization if condition met)
		composition (coloured, sized squares visualize weather)
		live wire (moves faster if more bits in channel)
		AuraOrbs (bubbles visualize presence of coworkers)	
		medication boxes help to remember
		umbrella lights up when ugly weather outside
		light buildings according to events happening inside
		
other input methods
========		
		
beyond mouse and keyboard:
	mobile spread:
		mobile are more common than desktops
		responsive design focuses on adapting content rather than different modes of interaction
	developers study tilt-and-tab:
		js library for recognising tilt and tab gestures
		goal:
			easy to use according to developers
		task:
			create slideshow
			small tasks including all major features
			group work like real company
			larger number of participants
			but maybe only part involved, no lab environment
		participants:
			web students
			perfect set of skills, but forced to do the exercise
		evaluation:
			grade of exercise, questionnaire
			lot of feedback, but not allowed to link grade/feedback
	
gestures:
	form of non-verbal or non-vocal communication in which visible bodily actions communicate particular messages 
	motion of the body which contains information
	types:
		symbolic (single meaning within culture)
		iconic (about size, shape, orientation)
		pantomimic (use of invisible object)
		deictic (specify object, directing attention)
	define a good gesture:
		design (by developer or user)
		memorability (how well remembered form / action)
		consistency (in form and action)
		customization (how users customize it)
		registration (part-gestures recognition)
		segmentation (when it starts/end)
		conflict (distinguish from others)
		completeness (all actions covered by set)
		
psychology:
	Inattentional Blindness:
		attention is directed to different things; big changes can go unnoticed
		unusual changes hard to detect (like fading building facades)
	Perceived Dominance:
		perception of factors such as power, intelligence based on appearance
		height, age, facial features