Web Accessibility for Individuals with Cognitive Defects

As I am responsible for the user-interface of
GroupServer, I am also
responsible for its accessibility.
I try and follow the accessibility guidelines, but supporting those
with cognitive disabilities is something that I avoided looking at in
detail: it seems too hard.
However, I decided that the time had come to have a look at some of the
issues with cognitive accessibility when the
ACM Technical Interest Service pointed out
the following paper.

Javier Sevilla, Gerardo Herrera, Bibiana Martínez and Francisco
Alcantud, “Web Accessibility for Individuals with Cognitive
Defects”
, from ACM Transactions on Computer-Human Interaction,
14(3), September 2007

Valiantly, Sevilla et al. have created a system that can present
web pages in a form more suitable for those with cognitive
disabilities.
In their (sadly unnamed) system, the page-information is described by an
ontology.
Another ontology defines a visualisation for the content.
By using them together, different pages can be presented to those
without, and those with, cognitive disabilities (specifically mental
retardation).

The pages designed by Sevilla et al. for those with cognitive
disabilities take cues from the TEACCH methodology, which is normally
used in the creation of teaching aids for people with developmental
disorders, autism, or both.
The system breaks browsing tasks down into small steps, similar to the
series of steps required to withdraw money from an ATM, or complete
a Wizard dialog.
The steps required to complete the current task are shown at the top of
the page, with the current step highlighted.
In each page, up to seven options are presented:

  • The Back button,
  • The Home button, and
  • Up to five links from the page.

No other components of the browser interface — such as the title bar,
status bar or menu-bar — are shown.

The system also removes the “distracting” elements from the page;
for the homepage in the study by Sevilla et al. there were 118
distracting elements removed.
This raises a serious concern with the evaluation: the advantages that
Sevilla et al. measured — and they were considerable — could be
due to the page having 118 fewer elements, not the automatic
deconstruction of the browsing task into small steps.
I am also troubled by the correlations drawn between IQ and
user-understanding, as the number of participants in each condition was
very small.

However, ensuring that the pages are clear from distracting elements
should benefit all users: support user’s task!
Doing this will be cheaper than having to define an ontology for
each site, and an ontology for each visualisation of each site!
The research by Sevilla et al. is fascinating, and I will watch with
interest for further developments in this area.


Mobile Data Speed

The following strikes me as a good exam question for a honours-level networking course.
Contrast and compare the new mobile networks being introduced in the Western Pacific:


Flexible Work is Your New Day Job

Alice and I have just agreed new hours of work. The standard hours that she works now don’t fit so well with her other priorities. So instead, Alice will be working 1100-1930, Tuesday to Saturday. On Saturdays, Alice will work remotely.

Does this make me a good employer? Actually, I think it just makes me not a stupid one. Alice has been here for a little over a year, and in that time has grown from an ace linguist with a geek streak, to a productive software engineer. In the same time, she has extensively automated and systemised service delivery to our main client, completed a stack of arbitrary hacking and documentation tasks, and been a lot of fun to have around.

Our business is about collaboration towards a shared vision. Smart, hardworking people like Alice, who share our vision, are a fundamental part of our business. Lots of the work is head-down coding which is well-suited to after-hours. And we are experts in online collaboration! We already collaborate closely with Richard, who is in Wellington. In fact, we collaborate online, even when we are in the same room. While some of our customers are in the same timezone and physical location, many are all round the world. And our customers’ users can want support from us at any time of the day or night. Alice is amazingly diligent and cares deeply about doing a good job. This might sound like a reference for her, and if that’s what it turns out to be, then call me for more. But my hope is that Alice will be working with us for a long time yet, wherever in the world she travels.


Mobile Computing

I needed some tech support with my home computer, so I took it into the lab (our office) where there are plenty of friendly geeks. If I’d taken it by car, I would have had to park outside the office, take the machine upstairs, then either drive back home and bike in, or go up the parking building and walk back. Either of these would have taken longer than walking (one virtue of living in the provinces), so I walked in, carrying the computer. That worked fine but my hands felt like death when I arrived.

Rob of egressive, kindly fixed the boot priority so the machine was good to go. Taking inspiration from Michael’s new tramping pack (will we see that here, Michael?), I rigged this front pack for the trip home.

Portable Computer

Webstock: Participants are the Producers. Deal with it. But how?

I went to Webstock 2008 for inspiration. I came away overflowing with the buzz, and the insights into how stuff works and what’s happening. There is so much to take in at the event, that making sense of it barely begins before walking out the door. But the most inspiring impact of Webstock for me, is the questions that remain unanswered. One week on, here are my reflections on these questions, and how they are propelling me forward.

Developers, designers and web entrepreneurs think that we are changing the web. We go to events like Webstock to get better at it. In our zeal, we can miss what’s really happening. The web is changing us. In fact, it’s changing everything. It is turning the process of production and consumption that we know on its head. It’s providing consumers globally the means to define the way the world is, to decide what happens, what is made, and what truth is.

At Webstock, people like Amy Hoy, Kathy Sierra and Jason Santa Maria showed how to manipulate people to do what we want on the web. But Liz Danzico cautioned that whatever we do, people will mess with it, and Tom Coates showed that our product is not our website but our data. We have no choice but to encourage users to share their data with us, and allow others to use and manipulate that data in any way they want. But how do we do that without being evil?

We have to follow research, gather data and make rational choices. But we also have to look outside the limits of our thinking, to embrace change and uncertainty. We don’t know whether we will fall by the wayside, or fall to the dark side. We don’t know whether putting more control in the hands of users will lead to creative diversity and democracy, or to tribalism and polarisation.

We are at the very beginning of this. All we know is that this web, this world, is about people. And that being in it requires us to be ourselves; to say and do less, and to observe and listen more. To open ourselves to relationships with others, perhaps to everyone, and to be spontaneous and creative as we find a way forward into the unknown.

Participants Used to be the Producers

We are entering the third era in the history of human participation in society. The process of production and consumption is being inverted. Instead of centralised production and mass consumption, we are entering an era of participant production. This is not not the first era of participant production that we’ve had. In fact humans have more history of participant production than we do of mass consumption. It’s just that this time, it’s scalable, potentially to all humans.

It is More Scalable to Centralise

The first participant production era ended at about the time we invented cave painting. Prior to that, communication was extremely expensive and difficult to scale. As the opening scene from 2001 A Space Odyssey shows, without words or other symbols, you have to show people stuff for them to find out about it. With words, you can have an oral tradition. This works really well in small groups, where everyone contributes as much as they consume. It works in parallel with subsistence material production, where there is very little specialisation. It relies on frameworks that enable messages to be communicated through generations through repeated hearing and telling of stories using quite regular forms. Because this system is distributed, it scales well to large numbers of people over space and time, but it does not lend itself to tactical organisation of large numbers of people.

Specialisation of production requires scale in human organisations. To scale human organisations, you need people who know what to do. When communication is expensive, the cheapest way to do that is to make all the decisions centrally, and just tell everyone else what to do. This requires some kind of one to many communication medium that minimises loss and distortion. You need some kind of persistent symbol, like a cave painting. Cave paintings can last for millennia and reach lots of people, but they’re not cheap. The problem is not knowing how to draw, or getting the materials, it’s getting access to a cave. There simply aren’t that many of them. To get your ‘tag’ on a cave wall, you have to fight off everyone else. Cave painting is not about prowess in killing animals, it’s about prowess in killing humans.

Technology Lowers the Cost of Communication

With a mass communication medium like cave paintings, you can start to scale human organisations to larger sizes, despite the high cost of communication. You still use the legacy media, of course. The favoured communication medium of most large empires before the current millennium was the sword. As technology lowered the cost of communication, the first to acquire access to it were the elite. You had to be elite to get access to it, and we used it to retain our elite position. Literacy, the printing press, broadcast media. This paralleled the centralisation of production. The few control the means of production, and communication, and the many passively consume. The problem arises when you need tiers of people to be well-informed. When the printing press and mass literacy start to lower the cost of education and information for people outside the centre, you start to be able to have larger organisations, and even to have democratic political systems.

Cheap Communication Enables Participants to Produce Again

The more technology you have, the more technology you get. And all technology is information technology. But the lower the cost of communication, the more autonomous people become. This is good news for scaling large organisations, especially in dynamic environments. The problem is just too hard for people at the centre to solve, so it gets pushed out to the periphery, where people are well informed enough to make the best decisions in particular domains. Then production crosses the boundary of the organisation and you have participant production.

Of course, none of this is news. ClueTrain, Thomas Malone and innumerable others have been talking about this for years. What was new at Webstock was glimpses into the pace and processes of it playing out. If you are thinking about “user generated content”, you are missing the point. The traditional production model is being turned on its head. The web is putting the power of production into the hands of every participant. This era of we, the few, creating sites that are consumed by the many, will be fleeting at best.

Mistaking the Web for a Mass Medium

In Web 1.0, the web was built out in the image of broadcast media. We all had the opportunity to extend the reach of our brand. Content was king and with it, we lured the unsuspecting. Then we got too tired to create all the content that we needed to keep up with the content that people generated themselves. We assumed the guise of a service, enabling people to connect with each other by sharing their data.

Mass exploitation hasn’t stopped yet, of course. At Webstock we learned about some insidious techniques that are used to lure users into our sites. (Insultingly, many of them are derived from the design of malls. Or are we the insulters?) Using combinations of visual design, information architecture, graphic resonance, multi-path search and navigation, by exploiting unconscious functioning of the human brain in the design of visual and textual cue, and the conscious adoption of a personality, a voice, we lure users to visit, to stay, to buy, to share personal data and content, even login information for other sites.

Game Over for Privacy Online

Read almost any terms of use, and you’ll see that you sign over control of your data. With a service like Gmail, it is not only the users whose data becomes available to the service provider, but anyone who they email or who emails them. Flickr now (optionally) supports geotagging. Legions of young and old adults use social networking sites, mostly without asking why it is free. Photo-sharing and social networking sites allow people to associate photos with you, without your consent. Nigel Parker gave an excellent 5 minute presentation about this at Webstock.

Your Product is Your Data!

You take two dull lumps of data and put them together and you make something interesting. Both Nat Torkington and Tom Coates made this point. Chicago Crime is the poster child for this, but any mashup is an example. Tom Coates says that if you run a website (and there’s actually no such thing as a website, by the way: they are all applications), the web interface is not your product. The data is. 90% of Twitter’s traffic is via their API. You will live or die by accumulating users’ data, letting them use it in useful ways, and getting others using the data, in diverse ways. The more this happens, the more secure your future will be on the web.

OpenID: There is Hope for Public Space Online

So this means that we, at Webstock, are both those who are losing control of our identity online, and those who are seizing control of identity information from others? Actually, I think people’s lax attitude to the privacy of their data is a temporary phenomenon. Simon Willison gave an urgent and compelling presentation about how OpenID provides a distributed system where each of us can choose just who to trust with our personal information, and how much to share with anyone we don’t trust. It is a significant enabler of something like public space online.

Frameworks Not Artefacts

Liz Danzico pointed out that whatever you create, whether physical or virtual, however participative your design process, you only ever find out how people use something when you launch. That process is iterative, as people respond to how others are already using it, and become involved in dynamic, self-conscious creation process. She likens traditional design processes to classical music, and proposes designing just frameworks, that allow a co-adaptive design process with users. Could we, like modal jazz artists Coltrane and Davis, be good facilitators of improvisation, by making great frameworks, rather than glittering artefacts?

Citizen Justice

Russell Brown notes how, for example in the Augustine Borrell case, the media are racing the cops to the story. Could things could go horribly wrong? But does he consider the possibility that the people ourselves are gaining the means to beat both the police and the media to the story? Of course we know how horribly wrong that could go, too. In the first instance, if you give people the tools to mess with stuff, they make mess. Witness MySpace and Piczo. Over time, however, some lovely stuff floats to the top. And we are just at the very beginning of learning how to create things that self -organise and scale. No example is as good as the Wikipedia.

Design for Participation

Another example of a framework-based process is WordPress. WordPress supports four basic tasks: write a post, change a post, change the design and manage comments. Within those limits, an ecosystem of themes, plugins, and a variety of users has spawned. The existence of mashups themselves is an example. If you don’t provide the functionality for people to remix your data, then people will figure it out for themselves. Even outside the web, in social and organisational change, often the means is a simple infectious idea (like the hipster PDA, or no pants day) that people implement and adapt themselves. Imagine a web where it is all like that. Where everyone is creating as fast as consuming? Where websites, art, music, movies, news, criminal investigations, surveillance, science, environmental management, and things we can’t imagine yet, are produced by decentralised clusters of people and systems that continuously form, decay and reform? Where trust, truth and authenticity are negotiated. Where everything changes in a moment. This is the Web we are participating in now.

Mike Brown closed Webstock with a heart-felt reiteration of the Maori epithet: He Tangata, He Tangata, He Tangata ([asked what matters, I answer] it is people, it is people, it is people). He also shared a quote from Sir Edmund Hillary (can anyone tell me what it actually was?) about the importance of relationships, and relational abilities in the 21st century.

Knowing these things, I am glad that I am not a geek, not a designer, and not even a natural entrepreneur but that my strengths are in dealing with people, groups, communities and organisations. I am glad that, with GroupServer and OnlineGroups.Net, we have chosen to focus on collaboration in groups and communities. I am glad that our privacy policy is to let users control their own data. I am glad that we give away the code to our software. The questions remain however, about how we are going to attract millions of people, give them something structured enough to be useful, and loose enough to mutate into thousands of different forms. How are we going to stay true to our values of humanity and collaboration, or could these be our greatest asset? These are the questions and challenges that motivate me to be doing this. At Webstock I got to be in a large room full of people, all facing, and feeling the same challenges. We were enthralled and entertained by the speakers, we ate, we drank, talked and laughed, we grooved to Twinset, the Ukulele Orchestra and Craftstock, we flaunted our bags and teeshirts, and felt proud to be part of a profoundly inspiring web event, that is right here in Aotearoa.


Dumpster Diving

Dan found a Lexmark X215 multifunction printer-copier-scanner-fax in a dumpster.
After recovering it, dusting it off, and giving it a try, we found that it was always jamming.
Despite my protests that I am not a hardware geek, I ended up taking the machine apart; I found that its Mylar comb, used to guide the paper, had a bent tine.
Two minutes of work with a pair of scissors, and OnlineGroups.Net has a new printer!


Mode Switch

At Kiwi Foo Camp, I had an excellent opportunity to discuss the
proposed
mode
switch

in Microsoft Internet Explorer 8.
Over the weekend my position changed, from being mildly in favour of
the mode switch, to being mildly against it.
I am still in favour of mode switching as a concept, but the proposal
does not solve an existing problem, and introduces too much uncertainty.

Mode switching is not a bad thing in of itself.
In a Web browser, a mode switch occurs
when the browser switches from one way of processing a document to
another.
Sometimes a switch is the result of an explicit user request: when the
user prints a document, it is is converted to black and white, and extra
margins are added.
More often, a switch will occur when the browser detects different
types of document.
For example, the
the
specification for the
css
overflow property

requires the browser to behave differently for
xml
and
html
documents.

The particular type of mode switch that has caused the fuss on the
Web recently is an extension of the current practice of detecting a
malformed document.
This causes the browser to switch from
standards mode to
quirks mode.
Broadly speaking, in standards mode the browser strictly follows the
formatting instructions contained in the document; in quirks mode the
browser tries to figure out what the author of the document intended.
All browsers do this.
The
proposal

is to extend this scheme so there is more than two modes, and allow the
author to specify the mode that should be used to render the document.
This will allow a new version of a browser to behave like an old browser
when rendering old documents, making browser upgrades easier to
distribute.

For browsers other than Microsoft Internet Explorer, upgrades are not
normally hard to distribute.
Few changes need to be made to documents when a new browser comes out,
as the browser implements changes smoothly; a process known as
progressive enhancement.
Progressive enhancement was used to transition users through all versions
of Netscape Navigator, Mozilla, and Firefox.
The upgrades to the Opera and Safari browsers are also handled through
progressive enhancement.

However, progressive enhancement cannot be used to transition from
Microsoft Internet Explorer 6 to 7 because the former is so broken that
there is no smooth path between it and the latter (saner) browser.
Any organisation that has pages that are designed
specifically for Internet Explorer 6 cannot deploy
Microsoft Internet Explorer 7 without breaking their Intranet
applications.
Because of this, many large corporates have stuck with Internet Explorer
6, as Internet Explorer 7 does not implement the new mode switching.
Unfortunately,
Microsoft Internet Explorer 8 will not have an Internet
Explorer 6 mode
, so its only switch will be from the
not-so-broken IE 7 mode, and potentially-great IE 8 mode.
While nice, this does not solve any problem that we have here and now.
Most sites should be able rely on progressive enhancement to move
from IE 7 to IE 8 — just like the sites will rely on progressive
enhancement to switch from Firefox 2 to Firefox 3.

The root of the problem is that corporates are stuck with poorly written
applications that are designed specifically for Internet Explorer 6.
To my mind, deciding to support one particular browser, especially in a
corporate environment is foolish: the companies would be better off with
a Visual Basic application, which would allow better work-flow, provide
better widgets, and be more stable across system updates.


Usability Mistakes

Just before I went to Kiwi Foo Camp, I created a major usability bug in
GroupServer, which was mostly due to forgetting to add feedback to a
form.
This caused a large problem for Alice, who had to manage the support load
while I was away.
Thankfully, I managed to correct my error and the page is now better than
ever!

Like many systems, GroupServer requires the user to enter his or her
password twice when it is set.
This is to ensure that the password is set to the value that the user
intended, because GroupServer (like many systems) does not show
the user the password that he or she entered, for security reasons.
I thought it would be nice to add a JavaScript check to the page, to
ensure that the two passwords matched.
In my prototype the Set button was enabled when
the passwords matched, or disabled (greyed out) if the
passwords did not match.
It worked fine for me; I knew why I was not able to click on the
button because I wrote the code.
Unfortunately, as the page moved from prototype to production I forgot
to add feedback to the page.

The disaster unfolded the day before Dan and I went to Kiwi Foo Camp,
when we signed up 800 people to an OnlineGroups.Net site.
When each user browsed to the site for the first time he or she was asked
to set a password… using my broken page.
Poor Alice had to field all the queries from users who could not figure
out why they could not click on the Set
button.

My solution — which I implemented in a very bleary state the
Monday after Kiwi Foo Camp — was to add feedback to the Set
Password page.
My initial implementation was to modify the password-checking code
so the passwords were dynamically-checked as the user typed.
This was better, but not good, as the feedback appeared as soon as the
user started typing!

After
some
prompting by Tim Erickson
,
and a good nights rest, I implemented the version of the
password-checking code that is now deployed on OnlineGroups.Net.
In the current version, the Set button
is always available, and the passwords are checked when the user clicks
on the button.
If there is an error, a message appears in the same position as
the button, moving the Set button down.
This helps the user notice the error message, as it appears where he
or she is probably looking.
After deployment, we signed up another 500 people to an OnlineGroups.Net
site, and everyone has managed to log in without difficulty.
Much to my relief.


Kiwi Foo Camp 2008

Last weekend, Michael and I, and some Christchurch web folk that we knew (and some that we didn’t) ventured to Warkworth for Kiwi Foo Camp 2008. Kiwi Foo Camp is an invitation-only gathering of web, art and science people modelled on O’Reilly Media’s Foo Camp. This means 150-odd people speaking fluent geek all weekend, in groups of all sizes, at all times of the day and night, and in states ranging from intrigue, through bafflement, hilarity, inebriation and exhaustion. The format is unstructured. Someone said “the best thing about conferences is the conversations in the corridors, and foo camp is all conversations in corridors”.

On the way up in the car, Julian, Seth, Michael and I were limbering up with geek talk of our own. I was wishfully speculating about mass-customisation. You know, people are sick of consuming the same goods as their peers, so they are start to make stuff themselves. Kids that have barely stopped scoffing at their parents are making and crafting and hardware-hacking. But that’s only because the Web taught them that they can, and it’s cooler to customise a beanie than a MySpace profile. But making things is time-consuming, especially say a fine-knit merino beanie. Before long we’ll be picking a design from a user-contributed online gallery, and ordering a custom-made, individually configured and fitted item, that is produced in a plant that spits out a thousand unique items in a minute. Hmm, we all nodded, how cool would that be.

In two days at Foo Camp, I’d met the people who are making this happen. First, you select your design, from the prototypes at felt.co.nz. The Bioengineering Institute at the University of Auckland have already modelled your body shape, size and posture. You combine your fitting information from there, with a 3D model of the object you’re purchasing, and view rendered simulations of you using or wearing it, while you vary colour, material and adornment parameters in your price range. When you’re happy, you place your order with Ponoko, who fabricate and ship it.

Of course, this too is wild speculation. Felt is actually an online marketplace for hand-made products. The Bioengineering Institute will be glad when they’ve modelled the human eye. And Ponoko cuts things out of plastic sheets. But they’re all doing this stuff with speed and sophistication. And they all had someone at Foo Camp. And they are only a small sample of the amazing and diverse people that I met there.


There is no folder

I predict that the Journal on the XO laptop will be a roaring success,

as it is designed around one person

— it is One Laptop Per Child, after all —

and it avoids the problem of Save overwriting documents.

I came up with this prediction after reading

a review of the One Laptop

Per Child (OLPC) XO Laptop

from which I saw a link to the

OLCP Human Interface Guidelines,

specifically the section on how file management works.

We believe that the traditional “open” and “save” model commonly used for

files today will fade away, and with it the familiar floppy disk icon…

Instead, a more general notion of

what it means to “keep” things will prevail. Generally speaking, we keep

things which offer value, allowing the rest to disappear over time. The

Journal’s primary function as a time-based view of a child’s activities

reinforces this concept.

(From

The Journal,

OLPC Human Interface Guidelines.)

I have had similar thoughts myself: my thesis detailed a system that just

used time to organise documents.

Considering that it was academic software, which had its flaws, it was

surprisingly good.

My work on document organisation was inspired by three observations:

  1. Documents are rarely retrieved once finished [1],
  2. Most email clients organise documents (messages) by time, and
  3. Save deletes the prior version.

Inspired my my results, I have been keeping most of my documents in

one folder (/home/mpj17).

For retrieval I use temporal cues and search — and it works.

Without folders, I no longer have to think about where to store a document:

it goes in my home folder.

For retrieval, I used the Recently Used section of the

Open dialog, or search.

The reason this works is that generally old documents are not

useful documents.

The fact that the site-style for

my work

is the oldest document in my home-folder is not really that much of an

issue, because I am unlikely to need to retrieve the two-year old file,

and it stays out of the way at the bottom of the list.

The down side of my system is that Save will still delete the

prior version of my document, unlike the XO Journal system, or indeed

the file system in VMS, or the Elephant File System.

The Recently Used section of the Open dialog is

also useful for the documents that are not stored in my home folder: code.

I use the Recently Used list a lot because I am generally

working on a few related modules, such as the

html

for a page, and the supporting

Python code.

However, the

GroupServer

code is stored in a strict folder hierarchy.

This is because it is maintained by three people besides myself,

and everyone (and

Zope)

has to be able to figure out where each code module is, what each code

module does, and how each module relates to the wider system.

In his article in The Register, Brian Hurley expresses a concern

about the children adapting from the Journal model of document

organisation to working with the XO code, which is organised in

folders.

I suspect that this will not be a problem.

People commonly use different means of organising different

documents: music is organised in a music player (such as iTunes),

email is organised by a email client (such as Outlook), pictures

are organised by a photo-organisers (such as iPhoto), and Web pages

are “organised” by the Web of hypertext links.

The addition of the Journal to organise personal documents, and the

folder hierarchy to organise code is not a big imposition.

It is definitely better than using a system designed to organise

code (in the

Multics system)

to organise personal documents.

[1] Barreau, D. and Nardi, B. A. (1995).

Finding and reminding: File organization from the

desktop.

Acm Sigchi

Bulletin, 27(3):39–43.