Notes from IA Summit 2015 Day 2

These are my notes from day 2 of the Information Architecture Summit, held from April 22-26 2015 in Minneapolis.

Table of Contents:

Applied behavioral economics for IA and UX

Robert Neal (@robertjneal)

  • dual system theory - brain operates in fast intuitive mode (system 1) and slow, methodical, cognitive and energy heavy mode (system 2)
    • example: lilly pads cover pond by doubling every day; if 48 days to cover, how long to half? System 1 says 24 days, system 2 says 47
    • if system 1 is involved, the conclusion comes first and the arguments follow
    • applies equally to experts in their field
    • design for what you want people to think, rather than just present logical outcome – don’t leave inferences up to the reader, make them explicit
  • indifference map: two decisions, indifferent between them, e.g. more vacation vs higher salary; traditional economics says this remains constant over time, but actually if you choose one and then you change it you feel like you’re losing something, no longer indifferent
    • uber surge pricing: instead of presenting as a loss, just show current pricing and remove reference point to “standard” pricing; could also show projected cost over the next hour to give people an option, feel like they’re making a good choice to go now or wait
  • attribute substitution: people offer a reasonable answer to a question that has not been asked
    • e.g. set thermostat very low so house cools down faster, but that doesn’t actually work that way, cooling is at a constant rate - thermostat is asking what temperature you want the house to be, people are telling it how fast they want the house to get cold
    • maybe show how long it will take to get to the set temperature - predict what question users will try to answer and provide feedback to that question
  • anchoring - provide a reference point that changes people’s perspective (could even be unrelated)
    • judges read cases, roll die, then decide a sentence - judges who roll lower numbers give shorter sentences
    • presenting the lowest price first has a strong anchoring effect, other prices seem high; if start highest, other prices seem cheap
  • framing effect -
    • 10% chance that the patient will die vs 90% chance that patient will survive, more likely to proscribe in second scenario
    • show that you’re saving when you pay more for bulk price
  • base rate neglect - when presented with statistics, it’s difficult to pay attention to base rate in population vs story
    • people will say more likely to be bank teller AND feminist than just bank teller
    • Nike compares you to other Nike+, who are other active people, more likely to be more active than you than the base population; but you’d compare much more favorably to the base population but then you’d have less motivation to improve
  • books
    • Daniel Kahneman, Thinking Fast and Slow
    • Richard Daylor, Nudge
    • Gerd Gigerenzer, Gut Feelings

Back to Top

What we mean by meaning: new structural properties of IA

Marsha Haverty (@mjane_h) - slides

  • structural integrity of meaning across contexts - nature of information and meaning through lens of embodied cognition
  • relationships between surfaces, edges and textures is invariant structure, information–don’t have to have seen every possible angle of chair to recognize it; recognize from invariant relationships
  • actor observer brings goals and actions, meaning emerges where goal directed actor observer engages with information in the environment
  • visual, mechanical, chemical, linguistic information, gesture, introspection part of information environment
  • perceptural information’s invariant structure = affordances, perception-action coupling
    • how does an outfielder know where to catch the ball? perception-action coupling to angle relationship of the ball, outfielder moves to eliminate angle on the ball, series of course corrections to maintain angle position, drift to the correct position
  • invariant structure of language? don’t have a word for that
    • information-behavior coupling is how and where meaning emerges, but not static, goals, actions and environment all changing
  • meaning is flow; flows have
    • viscosity (ease of flow)
    • is information like water? pressure, temperature affect phase state of water
    • perception flows easily, reflexive; language is more viscous - awareness, associativity, requires attention
      • (graph of linguistic information on y-axis and perceptive information on x-axis; top left is entirely linguistic interface, very viscous, conceptual; bottom right is entirely perceptual interface, very reflexive, visceral; line from 0,0 to 1,1 is even mix of both, but more of each triggers emotion, then overload)
    • example: twitter before they added images, dominated by language with images (avatars) in the periphery
      • active scanning, concept hopping - high-attention scanning
      • when they introduced inline images, moved toward reflexive from conceptual
    • design projects probably fall into linguistic half, with some perceptual components
      • but designs get phase shifted by surrounding ecosystem to more extreme modes - mitigate by using perceptual cues for changes, but to interact you need more attention
      • car touch-screen radio is gesture-based, muscle-memory, using perceptual information to suit the nature of the environment
    • need to think about design itself and greater information ecosystem - need to know when to use higher-viscosity language and when to offload some information to perception
    • texture
    • controlled vocab - language dominated, intense concentration
    • taxonomy more perceptual, including hierarchy
    • content strategy maps both perceptual and linguistic, balanced
    • concept spectrum: concrete to abstract
      • more concrete concepts balanced between linguistic and visual, something abstract like “calculus” requires more attention to engage with - but Vannevar Bush’s differential analyzer makes calculus concrete, makes abstract concept more perceptual, visceral
      • can do the same with metaphor, context priming - providing concepts phase shift concepts into a more perceptual space
    • use ways we detect environment differences to communicate conceptual differences?
    • events - locomotion, transformation, occlusion - Kavid Kirsh, external representations; Karl Fast, epistemic interactions
    • permeability - engaging and controlling the flow of meaning
    • when we interact with the invariant structure of information, some things can filter that - understanding vs not
    • need to be able to detect structure of information and then coordinate our behavior to it; failure to do either is loss of permeability
    • continued engagement requires structural persistance - outfielder needs to continue to see where the ball is to adjust his position
    • tolerance - how precise must the behavior be to mainting coupling
      • dashboard panel, put fingers anywhere on the screen, drag vaguely up or down - high tolerance

Back to Top

Shaping organizations to deliver great user experiences

Peter Merholz (@peterme) - slides

  • internal services model - centralized, your in-house agency
    • pros: strong design comunity, wide range of projects, design costs understood
    • cons: design not strategic (solving one small problem in product team’s solution), little influence on important decisions, designers easily dismissed, us vs them
  • decentralized and embedded - each product team has its own designers
    • cross-cut with director of design, product, engineering across teams to figure out strategy
    • pros: design included throughout lifecycle, part of the team
    • cons: one problem for a long time, lonely, fractured ux
  • centralized partnership - create a set of design teams that service different products, plus a couple of core design teams; design leadership within the team coordinate with a couple of different product teams
    • each product team has it’s known design point of contact, but that point of contact has a shared design team behind them
    • centralized design community still delivers as part of product team
    • reorgs don’t disrupt the customer experience because organizing principle for design is a different model than the engineering organizing principle, more stable
    • structure of each design team:
    • 4-7 people with a spread of design skills
    • singlular strong middle-management managing down to team, across teams, and up to execs and stakeholders - hire carefully, grease in the gears if they’re good, sand in the gears if they’re not
      • allows better project planning - low-level designers aren’t credible to upper management, just have to do what they’re told
    • organize around business problems, not functions
    • double-diamond model - diverge in ideation, converge on definitions and requirements, create plan, diverge on iterative design, converge on implementation
    • usually one plan, then multiple executions - but better to iterate on ideation and prototyping than in execution, less of an investment
    • lean/agile? lean and agile focused on execution once you already have a plan, only second diamond
    • challenge: every product team needs design, how to distribute the design time between them? no control of their own destiny
    • challenge: partner teams don’t understand the cost of design
    • challenge: maintaining singular leadership as group scales: operational matters trump creative matters, one person can’t really do both - other disciplines split creative and organizational management, like manageing editor vs editor in chief
    • leverage: CEO can go to design team and just look around to see everything that’s going on in the company; design can be a driving force instead of recipient of the plan, make org more ux-friendly - but needs confidence, evidence, and commitment
    • tech -> features -> experience
    • e.g. first phone that gets on the interet -> blackberry -> iphone
    • this model still stuck at features, driven by optimal number of engineers working together
      • designers can create a rich experience, but limited by model that organizes by features and engineers; what does a model organized around what designers can deliver on, rather than what engineers deliver on look like?
  • complimentary set of designers in team is better than multiple unicorns working alone - no one can be the best at everything, but you can have everyone be the best at something in a team

Back to Top

Let’s hear it for the seams

Discussion facilitated by Dan Willis (@uxcrank) and Livia Labate (@livlab)

  • people make myths when don’t understand, like myth of either mobile or desktop
  • channels != devices, people are channels too
  • Michal Levin - multi-device experiences: consistent, continuous, complementary
    • consistent - bank same info on desktop and mobile
    • continuous - buy ticket on computer, use with phone app
    • complementary - phone as remote for tv
  • elegant seams can trigger emotions based on expectations - e.g. can be creepy
  • best practice? provide a ragged version when elegant version fails
    • netflix starts at the beginning if they haven’t synced to the right location yet
    • not easy for uber driver and passenger to connect if the gps fails
  • best practice? if you can define elegant but can’t quite get it today, do what you can but keep trying
  • digital enhancement doesn’t necessarily help ragged seam between people
    • how does access to data change people’s behavior? e.g. bar arguments
  • properties of elegant seams
    • affordances, semantic information - how do you know what to do next?
    • intentional emotional impact
    • direct manipulation of information
    • flows as expected
  • properties of ragged seams
    • easy to identify when something goes wrong
    • impose certain reactions
    • data constrained by touch points, data doesn’t flow as expected
    • cross a boundary that user doesn’t want you cross
    • a bump in the road - could be to your advantage if you need focus or attention
  • aspects of seams to think about
    • feedback level
    • the space it happens in
    • locale, like uber cars and passengers
    • time/duration
    • storage, data
    • fidelity of channel

Back to Top

Designing webs: IA as a creative practice

Paul Rissen (@r4isstatic) - slides

  • More than 2.5million program pages on the BBC website (1 page for every episode, every series)
    • includes e.g. Doctor Who all the way back to 1963 with info for every episode
    • ensure that every tv and radio programme … has a permanent, findable web presence
  • use the web as a medium for creative expression
  • content model of episodes of dr who
    • plot, characters, scenes
    • trace chacaters through action, through time
    • first thing that happens to one character, to another character - same event through different lenses
    • fabula - raw material of a story, events of the story world
    • syuzhet - the way the story is organized - may not be chronological
    • knowledge that the audience has
    • story world events are not straight lines, they are webs - just because the narrative is linear doesn’t mean the world below is linear
    • don’t have to recreate the linear narrative online, can create the world as a web of information
    • story telling is IA in its purest form
  • BBC - Mythology engine - navigate archive of Dr Who by plot
    • link from one clip to related clip that it refers to
    • same event may be multiple clips across cliffhanger episode
  • creating structure of information unlocks it from the silos of traditional media
  • same narrative index can work for news, documentaries, etc.
    • trace significant person through history and see clips of BBC coverage
    • trace narrative of past elections
  • Amazon now having X-Ray feature on video platform
  • Can IA be a creative discipline - use IA to be imaginative, silly
  • Scott McCloud - understanding comics
    • creation of any work goes from content, form, idiom, structure, craft, surface
    • UX talks about the last 3, first 3 should be investigated more - we know the web is changing but we aren’t investigating the fundamental properties of the medium of the web - material exploration
    • we start by looking at existing user behaviors, then port them into new medium with slight modifications, building to predefined plan, but are we taking advantage of the possibilities of the medium? use playful projects to see what we can do with it
  • even earliest forms of information hyperlinked, we just didn’t have the medium to express it fully
    • sculpt data into the most useful, delightful shape
  • the material is the web (urls, hyperlinks), the form is the network
  • IA of the internet of things? concentrate on raw information, regardless of form - not mobile first, take a no screen approach
  • the world we live in is not a physical one, we mediate the physical world with thought, “Ideaspace”
  • Internet of things and internet of fictional things
  • if we only concentrate on serious things, then when we make mistakes we’re going to run into big problems. playful things are a good way to explore the medium
  • laws of alchemy
    • alchemist must have a medium for any form of alchemy to succeed - we have the semantic web
    • alchemist must fully understand the structure of matter - that’s what we do as IA, every story is a web
    • symbols and objects are treated interchangeably so that action on one affects the other - internet of things
  • designing webs
    • not just websites, think webs
    • web is a medium to express creativity, not just a platform for existing media
    • engage with the networked world, internet of things

Back to Top

Evening keynote: Living in a time of (un)ethical algorithms

Elizabeth Buchanan (@stout_ethics)

  • Ethics is about what’s possible, good, just, appropriate
  • We hear more about people who are unethical, and we think about compliance more than ethics but they’re not the same
  • personal not the same as professional ethics
  • Ethical possibilities, not just challenges
  • What happens when ethical questions are not asked?
    • FB emotion experiment, AOL data de-anonymized
    • in a few years this has become our new normal. how much of our lives are being manipulated?
    • Eric Meyer FB year in review
    • Drop-down male/female, marriage license with place for man and place for woman
  • what values do we embed or sustain that we don’t realize?
    • implicit bias can be unlearned, and we can encourage and benefit from differences: values-sensitive design, community-based participatory design, anticipatory ethics, etchical algorithms, action research - situate stakeholders as empowered and informed, experience ethical opportunities in our work
  • ethically speaking, what has changed? What ethical issues are we now experiencing in light of pervasive computing?
  • algorithms smarter, faster, intimate (they know us pretty well, if they’re working well)
    • in theory, they make our computing experience better, solve epistomological, ontological problems, help us make decisions about what we do/experience/share/know/exist
    • we see what we want, ignore probabilities, minimize risks that uproot our hopes; but algorithms offer what makes statistical sense, maybe they can help undo our biases?
    • With big data, we can find correlations across anything (see spurious correllation website)
    • computer algorithms can be so complicated that it can be subtly influenced while appearing unbiased
    • simple algorithms create tyranny of the majority
  • ethical algorithms Jeremy Pitt
    • resource allocation, find way to allocate scarce resources fairly based on past, present, future
    • alternate dispute resolution - automate mediation process - e.g. loser of judicial case should still appreciate that the process was fair
  • IDEO codifying company research ethics - Little Book of Ethics
    • respect - honor participants’ limits and values
    • responsibility - act to protect people’s current and future interests - what are potential downstream harms?
    • honesty - truthful in communications and work
  • book: the winter of our disconnect

Back to Top