D. Grant Campbell: You know, babyboomers are getting old. Many of us are getting old. We are getting old. Many of us. Now I don't care about the next generation, but somebody's gotta take care of me when I grow old.
Problems: the number, the duration, the cost. The skills. What I'm not
seeing and needs to be addressed is what happens when my generation
gets to that age. The problems of a technologically literate
generation. My mother never learned to use the net and will not; I,
however, know how to use it and will not want to give it up. The
dependence. The need. There is an undeniable need for maximum
independence and for assistance (medical, financial, shopping,
networking), and for contact.
Alzheimer's disease (AD) patients can be very enthusiastic bloggers and
email users, and their journals can be extremely important to them and
to those who are coming to terms with their disease.
Three questions. Neuropsy, caregiving, and medical research
connections. How do Alzheimer's patients differ from others in their
ability to recognize and use categories and labels? Can informational
websites that cater to elderly information users be designed to
facilitate these characteristics? Are the categorization needs of AD
patients consistent with those of non-AD patients? When my mother
became ill, she bought herself a kettle that automatically turns itself
off. I was an instant convert. I would love to say that if we can find
ways to design sites that are good for Alzheimer patients it would
result in sites that are better for the general population. However the
evidence I have so far is inconclusive.
Declarative memory is accessible to conscious awareness ("explicit
memory"). Episodic memory is connected to sensory experiences. Semantic
memory is the memory for specific facts. It tends to have limited
sensory reliance and is rather fragile. Non-declarative memory is
implicit memory.
What is retained in early/mild AD? Sustained attention - you can retain
a thread of thought. Procedural memory is very robust, in particular
musical ability, which seems to hang on for very long. There was a
person who could play the trombone to the day of his death, though you
had to hold it to his mouth. One lady lost motor ability, but her
musicality and ability to play the piano did not change at all. This
may be significant; we don't know as of yet how much of our ability on
computers is procedural. Reading and auditory comprehension is
retained. Capacity for conceptual knowledge, figurative language,
grammar (not spelling), social skills are retained.
What is compromised? Selective and divided attention is compromised.
Selecting something that is relevant among a range of options.
Extracting signal from a background is difficult. Episodic and working
memory is affected. Also, semantic memory - free recall of words,
categorization (naming entities, assigning them to categories, using
hierarchies).
Here's a snapshot of a page in the Google Directory. This would be challenging for an AD person. MedlinePlus is hard too.
In AD, familiar names are over-extended to similar entities. E.g. "dog"
refers to cats and rabbits. There is a bottom-up degeneration. "German
shepherd", the specialized labels is lost first. Then more general
labels like "dog". Very general terms like "animal" are actually lost
last, even if they were learned later in life.
The theory of parallel developed processing has suggested that the
reason for that is that the mind can be modeled as an information
space. When the person has a concept triggered to them, that sends a
signal to the brain. The effect to the brain of dementia is to
spatially dislocate associations. What researchers argue happens is
that this noise affects more general categories last because the space
they occupy is large enough that a good amount of damage is needed
before they are "knocked out of orbit" into another category's space.
That goes counter to our intuition - we would expect the categories
that were learned first, not those that are the most general, to be
lost last.
There is a shift from semantic categories to perceptual categories.
Controls favor semantic associations, while ADs will associate based on
shape, orientation, color - e.g. associating a carrot with similarly
shaped rocket rather than a bunny or potato.
I'm cautious about the last effect, I've only found it in one study.
There is a shift from prototype-based categorization to exemplar-based
categorization. If you have the concept of dog you tend to have in your
mind a prototypical dog, and you assess categories based on closeness
to the prototype. AD patients tend to group things according to
perceptual similarity. Paradoxically the ADs may deal much better with
very distortioned images. This could be significant for our use of
icons. There was an AD who kept going through a door with a big red
"stop". when asked why, she said "it means I have to stop, and then
go." She didn't see it as the prototype of a general prohibition.
Possible solutions. Facilitate movement through hierarchies. Explode
terms to give the items below. "The mutated synonym ring". If you have
an inability to make distinctions at a specific layer, what you may
need is maybe a synonym ring that takes "dog" in and suggests "cat",
"horse", ...
The "theory theory" approach to semantic cognition of Rogers and
McClelland. Cognition involves 1. use of naïve categories; 2.
context-sensitivity; 3. the cause-and-effect conditions of a particular
scenario or domain.
From the caregiving literature: "Caregivers could use this mantra: no
Alzheimer communication is meaningless." ADs retain a strong sense of
purpose.
Where is this leading us? To the use of tagging systems, possibly based
on RDF; the harvesting and organizing of resources based on
locally-defined tasks and purposes; locally-established domains of
services, caregiving and social structures. Seeding the information
with very local combinations and groupings.
And I really hope it works.
Q. (Leanne Bowler, McGill U.): Where does this fit within your research agenda?
A. My primary focus is the semantic web. This is a sidetrack arising of a family situation.
If RDF attains prominence, it will result in the ability to reuse
information. The voice in the back of my head tells me that we may end
up creating our own divisions and ideological rifts in RDF. [SP: No doubt about it, if you ask me.]
[This talk is about increasing
flexibility for users of content management systems (CMSes). Not sure
whether the oppression term is really the right one here.]
Rosalie Ehrlich: First two phases of content management: 1. Webmaster; 2. Templates.
AIfIA CMS survey: If you could make one significant improvement to CMSes, what would it be? Answer: FLEXIBILITY.
Because we can identify needs now, but we can't always identify future needs...
Stewart Brand on a design problem: Buildings are designed not to adapt. Ultimately they do, but not well.
Theory of meta-design. Systems must 1) Evolve, therefore can't be
completely designed prior to use; 2) Be designed for evolution; 3)
Evolve at the hands of the users.
Phase 3 of content management is components and layouts - maximize
flexibility through use of components that can be combined into various
page layouts. We can create generic components.
Kevin Roberts: You've got three seconds to impress me. People develop
rapid navigation patterns. One of the behavior efficiency strategies is
to examine the look and feel of a site to evaluate credibility. cf.
Stanford credibility.
Nielsen: Users are not designers. Designers are not users. Not! Users
may act as designers. We need to move towards this. We need to think
about the skills involved in those roles.
Empowerment: ability to keep site fresh, opportunity to innovate, learn
and use new skills, feeling of ownership. Oppression: could necessitate
more time to manage, uncertainty about whether the layout works,
requires time to learn.
Urban planners struggle with evolving cities: What creates the essence of a city? How best to serve its inhabitants?
Look at New York. First unplanned, then planned grid, build upward
(skyline symbolizes the city), accidental occurrences and serendipity.
Berlin. Unplanned, then controlled by the kings who needed to ensure
that no building was higher than their palaces, then contained, then
controlled to keep the European flavour.
To create a CMS that provides enough flexibility without overwhelming
users: 1) gather requirements and evaluate - how many people will use
the system, how committed are they to managing the content, is the
publishing process centralized or not, what is the approval process,
how far in advanced is content planned, does the organization encourage
innovation? 2) Perform a content audit 3) Determine cost 4) Design
Thank you.
Comment (Jorge). Wikipedia is to the city what corporate websites are to Disney world. Masterplanned cities.
Comment 2. The city metaphor breaks down for CMSes. Users contribute
content, not structure. Maybe it's more like a house where you get to
paint corners.
[By the way, this talk takes place in
the poster room. I notice IAs make killer posters as compared to what
you usually see in poster sessions]. What do you think? [] links to this post 5:22:28 PM
[SP: Very interesting session on knowledge sharing across organizational boundaries through a pattern repository.]
A Yahoo! Case Study
Erin Malone, Matt Leacock, Chanel Wheeler
We have a paper in the proceedings that provides more detail on what we'll talk about today.
Is there anyone who doesn't really know what a pattern is? It's a
"Reusable solution to a problem in a context", as Christopher Alexander
has described it in the field of architecture. You can use that for
other things like interaction design.
We are a siloed organization. Our people sit in different places. Not a
lot of knowledge sharing going on. What is the standard for X, is there
a convention? What are the other folks doing? Is there a way that they
can learn from others so they don't have to reinvent the wheel? It's
haphazard right now. There's about 120 people in the user experience
design teams asking questions. The company figured that it's pretty
inefficient. My predecessor and boss recognized the org had grown too
large and we got a central group to build a common repository. We found
that having the money to be dedicated to it rather than do it in our
spare time means it actually gets done.
There's value in the brand and the networkedness that comes from
working from a common knowledge base. Our pattern library supports this.
I used to work with an internal group to build an intranet. We didn't
want to go around begging for money, so it needed to be cheap. [php, movable type, drupal, other logos pop up] We wanted it scalable [some logos go away], customizable and extensible, easy to use (unlike coders, designers are a fickle bunch) [some other logos vanish], and conducive to collaboration. A bottom-up feel to it. Categorization. The answer was Drupal.
It has broad functionality, blogs, calendar, strong taxonomy system.
Active developer community - I want to do as little work as possible.
There's a new version about every four months. It has a very abstracted
engine. It's not the greatest at everything, but the taxonomy part is
very strong.
Pattern Library Workflow. One of the first things was to work on the
patterns first. I was very influenced by Stewart Brand's Shearing
Layers model. You can look at IA in a similar way. The yahoo.com URL
doesn't change often, other aspects change more quickly.
We bribed designers with lunch and cookies on the campus. We got
wireframes made. Findability: first, an index. Drupal has parent-child
relationships. This wasn't really scalable. We wanted a categorization
scheme. We did about 20 attempts at a priori categorization, but it was
hard without having any content for starters.
Each pattern has a pretty structured format. [...]
We came out with a ratings scheme, to assess not really the quality of
the pattern but whether you should be using it. Levels were: Avoid
(anti-pattern); Exercise Caution; ... Proven Performer; the Yahoo! Way.
What we learned. Don't let the technology define your process. Also,
treat this like you would any other product development. Paper
prototyping, get to know how people use it. Use a blog or a wiki,
something rudimentary to see what emerges, and then go towards
something more rigorous.
It's really important to make the content consumable. I tried to take
the mystique out of the pattern writing process. At one end you had
people who were afraid or doing it, on the other end some people would
put out 30-page novellas.
How a designer in a hurry reads a pattern: reads Solution, and looks at
the Rating. Those are above the fold. The rest is yadda yadda. The
Rationale helps back up designs for project teams.
Everybody gets a warm fuzzy feeling in contributing to the group. But
this motivation can vanish in the presence of deadlines. We did a
raffle for an iPod to incite contribution. Community recognition for
quality & quantity of work. We tied it in with people's quarterly
goals.
Another thing. You need your content to be credible. Don't invent. We
have people who are rolling out solutions and testing them out. We want
to capture proven solutions. It's a grassroots thing, it doesn't come
from an ivory tower. The review process matters. Who rates? If anybody
can rate it's problematic. We drew representatives from across the
business. We put in safeguards, results don't show up before we have
enough votes, we have secret ballots.
It is very important to define and operationalize roles. We have a
librarian, content authors, reviewers, an evangelist. It's also
important to communicate. We've been extremely transparent about what
we're building and what our motives are so people don't feel it's been
shoved down their throats. We did lots of roadshows to all kinds of
people.
You have to repeat the message that people should visit and use it,
that new patterns come up often. If you build it, will they come? Don't
count on it. This is not a static site. It's alive. You want to come
back. That's the message.
The champions will help you, write the patterns, propagate their use.
People who are in pain, dealing with four roles, need to learn about
how this can help. This matters to our international teams especially,
as they have limited resources.
We make a subset of the information available for our vendors.
What didn't work? I've been at the company for 11 months. There was an
initial push for using technology. Before I came in they had hired an
outside consultant to write a pattern library. This guy didn't actually
come in to interview the designers to see how they work. His approach
was very top-down, academic, and we had to throw it all away.
Things we haven't figured out. Trying to pull representatives from
different divisions to review patterns was hard. There was the "You
have been CHOSEN to be a reviewer!" idea that didn't work very well.
Successes. We have 55 patterns right now, written by interaction
designers and researchers. 12 have been reviewed and 12 are under
review. About 120 are in the review queue. This tool is being used by
design teams across the world. We implemented comments on the patterns.
Some patterns have become hubs to discuss design. That's good because
it means people who are in different businesses or locations talk to
one another, builds a sense of community. It has been more successful
than free forums, because it provides a focal point for discussion.
People talk about having gone to the pattern library. It gets used. The
last indication that this has been really successful: three locations
have hired people to translate the patterns.
We want to build a code library with CSS, a visual asset library
because nobody knows where they are. In the end we may have packaged
toolkits. Teams with small resources can pull stuff off the shelf, it
frees their time to spend on the harder problems.
Q. Timeline?
A. Months.
Q. Is it prescriptive?
A. No. People make the call.
Q. (me) Monitoring is RSS or email or both? How about Yahoo! leading an open effort in this area?
A. Email right now. I'd love to open it up. It has to go through legal though.
Q. Standards vs. patterns vs. guidelines vs. best practices - what's the difference?
A. Patterns can be cross-referenced. You can have one big pattern that references other dependent patterns.
Q. Editing workflow?
A. Submissions are published pre-review. The editing is ongoing.
[SP: This is important work. I hope they find a way to open this effort to the outside world.]
Peter: We thought about global websites and websites all over the
world. We've been talking with practitioners. It's a really deep
subject and there's lots going on.
We've been looking at the wide range of IA, things like "can you
translate a taxonomy"? Culture, George Lakoff, people have different
categories. Strategies. Methods like our IA practices, how do they work
in a global setting. I know a lot of you have really good experiences
and stories - let us know. Let's concentrate on IA - we don't really
want to focus on charsets or on interaction design.
Livia: We had an internationalization BOF yesterday. The localization industry is strong.
Jorge: Somebody told me "We're moving to 26 languages in one year"
Peter invites the floor to talk.
Comment 1. (Peterme) We've been thinking about centralization vs.
decentralization. The poor shmucks are going to have to deal with what
the core folks put out. the HQ people think they can do it centrally
and just translate outward. The global sites are essentially
marketing and sales sites. For me the concern is, I get very nervous
assuming that all cultures can follow one uniform architecture.
1b. You have evergreen and deciduous content. Evergreen can be
centrally designed and translated, while deciduous should be dealt with
locally. It's a spectrum. Maybe think about content groups within your
system to bucket things.
1c. If they have local offices you should get them involved in the process.
1d. (Livia) Centralized vs. Decentralized, and top-down vs. bottom-up.
(Peterme) it seems to come down to cost. (Livia) If the org you're
dealing with has made a decision, you'll have to compromise.
2. (James) In North America and Japan we have large markets. Moving to
Greece or Bulgaria, where the market is 1/10000 the size, we have to
scale down. The people can't necessarily do it all.
PeterVD: Okay, who has learned something that they can share?
3. Contrast Asia with Europe. You need to acknowledge that some
variation will occur in your IA. What you wind up with is a fairly
static tree where the changes occur at the leaves. I wouldn't count on
your navigation being perfect globally, but as a first step you can go
that way.
4. We are implementing a global intranet. The navigation is fairly
static, but the content is locally created. Maybe we should ask if the
navigation suits everyone. We haven't.
(PeterVD) You can enforce standardization, but people find workarounds.
5. I helped make a site about i8n for a very large multinational org.
The culture of the company overrode the local cultures. (It was a
mixture of tech and non-tech people.)
6. Several people report that when there's specialized activity you often don't find cross-cultural variation.
7. People will often say up front that they're different. But we've implementing something uniform, and it worked.
8. (Jorge) We have to distinguish between functional and optimal, though.
9. We talked to people in China, Japan, Korea and India. In our domain,
we didn't see a difference between the tasks they wanted to do. This
made it easier to do it in English, then translate.
10a. In my anecdotal experience, Western folks seemed to have a
top-down vision, while the Japan folks were very bottom-up. They didn't
want to start from the abstract but rather from the details.
10b. It's very interesting to compare the English and Arabic versions of arabia.com.
11. (Peter) I'd like to highlight that we have a lot of second- or
third-hand information. Maybe we need more first-hand insight.
12. Having lived in Northern Canada for many years and gone to Asia,
I've been wondering if our validation techniques really work. In Japan
if you ask a direct question there is an overriding politeness where
people will say "yes" when they actually mean "no". I wonder if people
have worked on adapting test methods.
13. I know of one instance where testing procedure failed to the point that one of the subjects broke down and cried.
14. It's very important to have a bilingual person to bridge, both for design and collaboration, expectation-setting reasons.
15. At a strategic level having someone who really understood the local
culture is superior to using a straighforward usability test - they
make the assumptions come out.
16. (Jorge) Everyone belongs to several cultures. Everyone we design
for belongs to the Internet culture. Working with someone who doesn't
belong to Web culture can be very frustrating.
17. (Rashmi) I agree, the early adopters have a common culture.
18. Brazilians preferred English to straight-Portuguese (non-Brazilian) translations.
19. A lot of our local websites we have a large credibility gap. The
translations are poor, or there's less content. We see it because
people route around them, they prefer going to the US website.
20. Price is a big issue.
21. (Jorge) horror story. I bought computers on a local site and got a
call a week later asking if I was aware that I'd have to pay for
shipping from Miami, which is where the computers were being built.
22. (Peterme) Cultures are becoming transnational. As a San Franciscan I may have more in common with someone in Montreal. What do you think? [] links to this post 9:16:56 AM
Panelists: Thomas J. Froehlich, David Robins, Don Turnbull, Andrew Dillon, Nancy Kaplan, Peter Morville
[SP: this is a partial transcript.]
Thomas Froehlich, Kent State U, LIS/IA/KM program: Curriculum Planning Process at Kent
Our program started about 6 years ago, when I raised my hand at a
meeting. It's been going downhill since. (Or uphill, depending on your
perspective.)
We offer a Masters in IAKM. Interdisciplinary degree, can enter from
any undergrad degree because the core covers the foundations pretty
well.
Core courses:
IAKM 60001: information literacy (e.g. search engines, commercial databases, searching skills, knowledge organization)
IAKM 60002: metadata, controlled vocabulary, semweb, ontologies etc.
IAKM 60003: Information Design in the Digital Age
CS 61001: Structure of Computer Science - as interesting as the title. We'd like to get rid of it.
IAKM 60005: Information Technologies, management viewpoint
IAKM 60006: Strategic Info Managment
ECON 62015: Economics of Information
+ one Research Methods Course
I was promised 5 faculty members and got 0.5. Without SLIS our idea
would have just died, but the SLIS Director was generous. We are housed
in SLIS.
We had freedom to be creative in setting content, unencumbered by
faculty turf wars or proprietary claims. But interdisciplinarity makes
it hard to control content.
Content sources to define the curriculum: Analysis of job ads, AIfIA
survey of competencies, advice and blogs of field pioneers and leaders
(Morville, Rosenfeld, Merholz, Wodtke, etc.)
Issues: avoiding excessive redundancy throughout the course set,
ensuring that a wide variety of skills and software packages are
included. Serve prospective consultants as well as future large team
leaders. Anticipate future trends. Find adjuncts. Course waivers.
Peter Morville
I occasionally teach at UMich, where I got my degree back when the
school still had "Library" in its name. We use the polar bear book.
It's fun for students to see their project work being put to use in the
real world.
That's all I have to say about IA in education. So I thought I'd play troublemaker here.
In the world of practice we've had endless, agonizing arguments in
defining IA. the polar bear book has four definitions. Some might say
this is "wishy-washy" or "flip-flopping". I'd call it an act of
diplomacy.
I wrote an article called "Big Architect, Little Architect" and it got
me a lot of flak. Here's JJG's (in)famous diagram of the elements of
user experience. It helps people understand all these different
disciplines and how they fit together. A few rebel IAs refuse to live
inside Jesse's boxes. Peter Boersma: "Big IA is now UX". In practice,
yes, but in academia, I'm not sure.
When we look at sister disciplines (Interaction Design, Knowledge
Management) it gets messier. I'd argue that understanding
organizational politics is absolutely essential to success in IA
practice. Same with understanding marketing. (Note that marketing
people will likely object to marketing being a component of IA.)
I'd like us to move beyond this and take "findability" as the encompassing umbrella - School of Findability sounds good.
The community of Information Architecture is much larger than the
community of Information Architects - many non-specialists are
interested in our field. The IA Institute is important, its job board
is active and matters.
I'm biased. I've consulted to many large corporations, but there are other places where IA is practiced.
Peter praises the "Non-Designers' Design Handbook".
Remember that we're defined also by what people ask us to do. I'm often
asked to develop interactive sites, online communities, things that are
not my core focus.
It's fine that the first generation of IA was self-educated, but I
think for the next gen it will have to be at least partially done the
traditional way. I think that in this applied field it's absolutely
critical to have a close connection between academia and
practice/business. Thank you.
