Calum’s Wee GNOME Blog

Yes, I know it’s been nearly two months, but I’ve finally been prompted by a meeting next week to do a short write-up of the sessions I attended at CHI 2004 in April.

It all started with, amongst other things, the presentation of a Lifetime
Service award to our very own Robin
Jeffries. Followed by…

Opening Plenary

Jun Rekimoto
unfortunately isn’t the most fluent of English speakers
(and why would he be), so this session was harder work than I was
hoping for first thing in the morning Amongst other things, he
talked about his work on pick and
drop, augmented
surfaces, and other collaborative environments– including a
classroom with a shared display
at the front of the class on which students were allowed to rate the
quality of the lecture as it progressed. Unfortunately some of
the more
interesting questions from the panel and audience were somewhat lost in
the
translation… which is a good trick if you can get away with it

Games: What’s my Method?

Chose this one in the hope that it would ease me in gently.
Cornily-staged as a game show in its own right, this session looked at
whether conventional usability techniques (ehtnography, contextual
enquiry, biometric measurement etc.) could apply equally to
games. The conclusion, following some entertaining video clips
from gaming usability studies, was an unsurprising “mostly yes”.

Video Visions of the Future: a Critical Review

A panel chaired by our very own Eric Bergman, looking at ‘vision
videos’
from the past (such as Sun’s Starfire
and Apple’s Knowledge
Navigator), and asking why many of the features predicted for the
present day never quite happened the way we thought they would. I
don’t recall many concrete answers (caveat: this was two months ago and
I wasn’t taking notes), but it was certainly fun seeing all those
videos again…

Mark my Memories

A short paper session, the most memorable of which was
probably
Stu Card et.al.’s 3Book: A Scalable 3D Virtual Book. Essentially a 3D represenation of any
scanned-in book, one of its more interesting features is its ability to
create degree-of-interest indices based on its contents. I left
unconvinced that the 3D aspect (accurately rendered down to the
animation of turning pages) was little more than a gimmick, though.

Also presented was Photo
Annotation on a Camera Phone. Making use of a phone’s camera
and internet connection to do something useful seemed to be a common
theme this year, but speaking as somebody who wishes that the only
thing you
could do with cellphones is dial 911 or a breakdown service, I’m afraid
it didn’t really grab me.

Designing the Humane Interface

Amidst glossy tales of Disney.com
redesigns and Carlson
Marketing corporate reward scheme templates, the paper that stood
out for me here was from the Universities of Dundee and St. Andrews: a carer-driven
system for “reminiscence therapy” with dementia patients, with which
they can stimulate conversation by accessing multimedia clips,
typically of
sights or sounds from the patients’ childhood (rather than of friends
and family, which apparently tend to have a negative effect).
This seemingly simple approach– reported here by
the BBC— seems to be remarkably successful.

Can You See me Now?

Three papers presented here: one on mouse and touchscreen
selection in the upper and lower visual fields, and one on how varying
icon spacing changes users’ visual search strategy, about which I was
mostly none the wiser afterwards.

The paper I was most interested
in was a comparison of the effects of quantisation vs. frame rate for
streamed video (of soccer match highlights, in this case). Its unexpected conclusion was that contrary to most service
providers’ Quality of Service policies, users actually prefer high
quality to high frame rate (for fast moving sports, at least).
And also that, perhaps because of the comparative novelty,
people seem willing to pay up to $10 a month for surprisingly-poor
quality video.

Finding your Way

From these papers, I was most interested in
IBM’s reMail talk,
given our
involvement with Evolution
and Glow.
Was disappointed, though… pretty much everything of any use is
already available in Evolution. And slightly bafflingly, one of
the features most popular with its users was apparently its Thread Arcs
visualisation, which allow you to see a selected email in the context
of its response hierarchy. Er… tree view, anyone?

Another camera-phone paper here too: the idea of this one is that
you could
retrieve information about
a particular building or other tourist attraction by taking a picture
of it. The picture is sent back to a server that compares it with
other textually-annotated pictures that people have taken from the same
location. If the photo is determined to be of something that
somebody else has already photographed, a web search is carried out
on the annotation, thus returning to your phone within a few seconds
all sorts of information about whatever you’ve just photographed.

Games

My personal highlight of the conference: a presentation by the guys
who
invented the ESP Game. Inspired
by the huge
fascination with sites like Hot or
Not, the team from Carnegie Mellon hit upon the ingenious idea of
labelling every image on the web– a massive, non-automatable problem–
by having you and I do it for them, in the form of a game.

It’s
easier just to visit the site
and play it (it doesn’t seem to work behind Sun’s firewall,
though), rather than read an
explanation. But basically it’s a kind of web-based Pictionary-style
affair in which two randomly-selected human opponents, unknown to each
other and unable to
communicate, have to agree on a single word that describes a
randomly-selected image. Words that have been agreed on for that
image by previous players are also “taboo”. Consequently, the
agreed-upon words are almost guaranteed to be descriptive of
the image (and there are some built-in safeguards to filter out those
that aren’t), as there’s no way to influence what your partner will
type.

At the time of the presentation, about
6-10 descriptive words had been recorded for around 4 million images
from google.com (although the labels are not yet, as far as I know,
used by google or any other search engine). Right now there’s
only an English version of the ESP Game, but the concept is of course
pretty much applicable to any language. (And apparently, even the
English version is very popular with Japanese students trying to learn
English…)

HCI Overviews

A turnout of about 80 people for our talk on open source
usability, which wasn’t bad for first thing in the morning after the
CHI Reception. And no tricky questions either

Sightseeing

A few photos
from Vienna
here…

Closing Plenary

By Tim Brown, CEO of IDEO, on
designing “technology experiences” and “experiences enabled by
technology”. Examples included a design study for
Prada, a huge interactive
display for Vodafone headquarters that can be controlled via your
mobile phone, and an art exhibit
featuring chairs that projected an image of your clothing onto the back
of the chair when you sat in them– the electronic equivalent of
hanging your jacket on the back. Tim also discussed the
differences between “top down” design, which typically results in a
scripted, controlled experience (e.g. Disneyworld), and the rarer
“bottom up” design, which gives a more organic, continually-evolving
experience (his examples included eBay
and NTT DoCoMo).

Everything wrapped up with a cringeworthy musical number to publicise
next year’s conference, and a video about Portland, Oregon that
I’m guessing still hasn’t quite finished yet.

Gah… somebody at Sun sent me a link to an article that he thought might be useful background reading for my possible OSS Usability article for Interfaces, and now I can’t find the email. If you’re reading this, can you send it again please?!

Alan Horkan (link still b0rked at time of posting) pointed us at a sneak preview of the results of a GIMP usability study by Relevantive today. How far open source usability has come that such a thing was even contemplated… and how far we’ve still got to go