Talks, Fall 2001
1) BJ Fogg: How computer technologies can motivate and influence people
2) Jen Mankoff: Alternatives to desktop computing (GUIR)
3) Victoria ----: Ethnography

BJ Fogg - Persuasion

understanding how computer technologies can motivate and influence people
9/18 Tuesday;

hcc schedule
room 202 south hall, T 4-5

computer persuasion doesn't only hold in consumerism ...
"baby think it over" (teen roleplaying: baby)
5-a-day cd-rom for kids
"vr" bike

... and websites, of course
example of early site: an auction (early 'net was populated by gamers)
not shopping - playing; "win" if you win auction
entertainment vs. persuasion
computing timeline: functionality -> entertainment -> ease of use -> persuasion
"captology" - computers as persuasive tools

"functional triad":
tools (increase capability),
media (provides experience),
social actors (creates relationship) <- think tamagochi!
most technologies are mixes of these things
3 functions = 3 ways to persuade

computer as a tool: can change thinking patterns (like a scheduler: structures time)
as a medium: virtual reality, simulations, cause/effect, rehearsal .... (perhaps most
powerful way to persuade: vicarious living) example: "alcohol 101"
as a social actor: chat? reciprocity (favor -> indebtedness -> favor back) between humans and *computers* ?? (do people have this perception?)
reciprocity study:
first part of experiment: rate things for desert survival; *consult computer for help* (in ONE case, the computer gave great information; in the OTHER, the computer gave really bad information)
second part of experiment: people either went back to the same computer or to a different computer, then did a rather boring, repetitive, and endless task (rating colors from darkest to lightest for "computer's benefit") until they got sick of it and stopped findings: if computer was HELPFUL, people generally did MORE comparisons in the second task and "felt better" about the interaction; opposite for bad advice

how do tools persuade?
tunneling: go through a set sequence of steps (like putting sw registration in installation)
tailoring: personally relevant info. to change opinion ("chemical scorecard" **check out**)
conditioning: operant conditioning (reinforcement) example: slot machine
reduction: complex -> simple (one click away to buy things, e.g.)
suggestion: well-timed decision point (like the "speed limit __, your speed __" things; software update reminders) <- of course, the speed limit thing could be seen as persuasive if people thought their licenses might be recorded ... it's about what people PERCEIVE a tool does, not what it ACTUALLY does
self-monitoring: monitor themselves so they can modify their behavior to match a predetermined goal (like the heart rate monitor: beeps if out of target range) surveillance: probably the most common persuasive technology, unfortunately ... (cell phones, 'net monitors ...)

... so which influence ME? Which are most effective in general?

now hopefully able to recognize these persuasion techniques: raise awareness
(**it would be great to have a kid's class or a general-public class about persuasive technology**)
use persuasion RESPONSIBLY
where does persuasion end and coercion begin?

computers don't get tired (neither do telemarketers :~))
many ways to persuade
credibility built-in: "advanced technology," "smart"

believe a computer more about some things than others? What are MOST effective? Comfort level of "talking" to a computer about personal things?


Jen Mankoff - alternative to desktop computing (guir)

10/3 Wednesday 4-5 306 soda
alternatives for output: audio, pda's, moveable painting
alternatives for input: camera, microphone, motion detector, communication
"ubiquitous computing"
treat different needs, different environments
ambiguity: what does user want?

recognition support (speech, handwriting)
currently, recognition is becoming ubiquitous, but recognition is difficult to use: prone to errors (interface problems)
part of thesis: OOPS (Mankoff 01)

separation of interaction: mediation, recognition, application

ambiguous hierarchical events: add a pop-up mediator
(**how to make these mediators NOT distracting?**)
**how to make these mediators support different kinds of input (e.g., writing pad ambiguity -> keyboard OR mouse OR speech ....) for different needs of users?**

mediation subsystem: automatically identifies ambiguity by recognizing multiple "nodes" (if each event is seen as an extension of a tree)
another option: correction library - keep track of corrections in past to predict

event dispatch: sensed event arrives, goes to an input handler, is run through receivers, then through mediators
(**how LONG can all this take? in average case?**)
change: sensed event arrives, dispatched to ambiguity-unaware components, dispatched to all recognizers, then mediators ... then any ACCEPTED leaf nodes are dispatched to ambiguity-unaware components

**her OOPS toolkit - more modular?** (from comparison to 1990's GUI?)
can be used with any application, even if it doesn't know about potentially ambiguous input?

speech used with browsing? mouse used for dictation? how to rectify the mismatch? how to redesign

for model: cognitive and motor capabilities of users (what if a user can't easily switch to a less-error-prone input device?)

window is a good non-technical example of an ambient "display"
- history (for example, the fountain that's tied to stock price - it only tells CURRENT state, not anything in the past)
- transitions - summary to detail (say the fountain was tied to all nasdaq - how can someone extract the price of a SPECIFIC stock?), interactivity
- art

interaction design becomes experience design - computers are everywhere
lots of exploratory work is needed; many related areas (ubiquitous computing, etc.) or

**differences: word processing - "backspace" - NO change in mediator (less to distract? work on not breaking flow of thought - no focus on the TOOL, but on the task you're trying to do with the tool)**

**with stock example: have little graphs for each stock (history of nasdaq PLUS history of individual stocks) - then it reminds me of one of those pictures composed of lots of little pictures :~))**


Victoria _____ - Field Work


Why do we do fieldwork?
exploring settings for design opportunities
exploring activity for specific requirements
critiquing specific existing design in situ
getting feedback on preliminary design
evaluating new model/prototype

as much about stopping BAD design as pointing out good

methods: interview (structured questions), observation (expensive), participatory intervention (advice, brainstorming), experimental evaluation (too much "firefighting" due to design lock-in), questionnaire (risky, sketchy)
*ideally, use these in concert!

quite a headache to get responses!
-gaining entry to sites to study: time consuming, back door vs. front door (w/ gatekeeper) how much data to collect? what's too little, what's too much?
-video, audio, stills, trash :~)
analysis - quantitative, qualitative
communication: MOST IMPORTANT!
-designers hate to read long things (esp. bad news)
-stories, case studies, etc. best


representations of social space give: inceidental awareness ane learning, opportunistic social interaction around the work, kudos for effort through visibility of project

computers make work invisible! none of the above are presently supported
**ambient display for what people are doing?**

computers are overloaded

why NOT to do fieldwork?
time-consuming and expensive (even though it saves engineering effort)
"obvious" what users want or need (ignorance is bliss ....)
market research is adequate (but only part of what you want, and it depends on how you want to use it)
design-driven vs. requirements-driven (but revolutionary tech. should still meet real needs; users may not be completely right but shouldn't be underestimated)

case study - xerox "new enterprise" (xne) venture
portable document reader (tech-led, too-different design imperatives)
user studies: try to find specific requirements, get feedback on preliminary design, evaluate improvements

meant to be like a tab of paper - <3 lbs, no applications, had a scanner, very high-resolution ... basically to go through documents quickly and easily

made into a "prestige" device
market research: some thought it'd be for high-power execs, others for the "consumer" market

"not a computer"; too-ambitious deadlines

design in "people's heads, in people's hearts" - can't take away designers and expect product to keep going!

carefully think about WHO it's designed for: easy to have two groups that think entirely different things!

**hindsight is 20/20 - any techniques for anticipating problems? how about for people without training in psychology/social science?**

avoid design lock-in: evaluate EARLY with rough prototypes (also, if users know it's an early prototype rather than a finished product they're more likely to be honest) - consult with target market

**book: "how to lie with statistics" :~)

consultants BETTER at telling the truth (though sometimes they schmooze so they could get hired AGAIN ...)