The Principles of Optimal Design for Digital Inclusion
Presentation given at The Rix Centre Conference ‘Me! Multimedia and e-learning for Inclusion’
Date: 18/01/2009
Venue: University of East London, London, UK
Article
In the first 10 years of the World Wide Web, the emphasis in accessibility was largely concerned with the particularly severe problems confronting blind and visually impaired people. With the growth of convergence there is a new opportunity to implement optimal digital information design to meet a variety of user requirements from automobile drivers to those with special needs where such attributes as defaulted simplicity and accessibility features, predictability and consistency are applied across platforms in such areas as user interface design, taxonomy and information display. There is also an urgent requirement to break textual hegemony.
All the major research studies into digital information accessibility, notably the Report written for the DRC [i] show that those who experience the greatest difficulty suffer from severe visual impairment or blindness. It is no surprise, then, that at the inception of the World Wide Web (w3.org) Accessibility Initiative (WAI) in 1997 that the key emphasis in establishing a self-regulatory industry standard was on access by these groups of people covering three major areas:
Blindness & Visual Impairment Web Access Requirements
- The capacity to customise text (size, font, colour, leading, kerning, justification, background)
- The ability to navigate
- The necessity for graphics to be described (alt tag and long.desc).
Meanwhile, many countries, notably the UK, pressed on with the regulation of television accessibility in the area of signing, sub-titling and audio description but, in spite of massive regulatory and licensing powers, did nothing about accessibility in the exploding mobile phone market.
When WAI published its Content Guidelines [ii] in 2001 with their 64 check points it was obvious that, in the absence of legislation and regulation, implementation would be patchy; it also soon became clear that the whole corpus lacked credibility because it had never been user tested. It was also becoming increasingly clear that it was no longer effective to establish digital accessibility standards in user interface silos for the PC, television and mobile phone. For example, web standards called for text customisation whereas TV subtitles were not customisable; and although web navigation had been a big issue from the start, Electronic Programme Guide (EPG) Accessibility proved to be a thorny issue [iii].
In order to increase WAI compliance acceptability I concluded in my Inaugural Lecture at City University [iv] that at least all the major Web guidelines could be boiled down to three golden rules.
Golden Rules of Web Accessibility
- Enable customisation
- Enable user interface/platform neutrality
- Create multi modally
The last of these deserves a note of explanation; to create multi modally is to make information as if it were television, i.e. combined text, audio and pictures. To implement this is relatively costly but many aspects of it have become a specific or implied legal requirement.
Perhaps strangely, given the blindness and visual impairment age related demographic, the emphasis on web access took priority for many years over broadcasting access. Now that this distinction is becoming ever more blurred, it is time to take a new look at digital information accessibility, formulating simple guidelines appropriate to the different user interfaces within a common framework. The starting point for this enterprise should be the concept of optimal design which I define as follows:
Definition
Optimal design is the sum total of all features each of which produces a positive cost/benefit for the producer.
There are three points to make here:
- First, commercial suppliers may wish to exceed this criterion from own resources, CSR budgets or in partnership with the voluntary and public sectors
- Secondly, the public sector may wish to regulate or legislate additional requirements
- The public and voluntary sectors may need to finance accessibility beyond that provided by optimal design.
It is important to distinguish commercial defined optimal design from 'universal design' or 'design for all'. What I want to show very briefly is that many aspects of good design for accessibility are either highly cost beneficial or neutral; here are three basic ideas:
The Basics of Digital Design Processes
- Defaulted simplicity and accessibility features
- Consistency; and
- Predictability
Defaulted simplicity and accessibility features
This first principle should not be contestable. If you have learning difficulties you should be shown the simplest information array as a default; if people want to turn on complexity features they are welcome. The best kind of simplicity is that which is based on elements ranked by individual user behaviour but if that is not possible then ranked aggregate user behaviour is better than nothing. This enables a rational prioritisation which allows 'cutting from the bottom' when selecting a small number of elements.
It is also improtant that ranking is not restricted to classes but can take place across classes. For example, on many sites such as the BBC, football is neatly sub classified under the main heading of sport but the number of hits on football far exceeds the number of hits on some other main headings; so football should be a main heading alongside sport. Pure, theoretical taxonomy is as much a matter of intellectual preference as it is a matter of tidiness. The idea of simplicity should also include taxonomical self-discipline, separating content from navigation prompts and eliminating authorial whimsy.
In respect of defaulting to accessibility features it is obvious that it is easier for a seeing person to switch off speech than it is for a totally blind person to switch it on. None of this costs anything.
Consistency
Consistency is a matter of corporate or network compliance. Again, citing the BBC, it currently has an archive access pilot on cable which limits the number of categories of programme to nine whereas the number of categories in its web access pilot is twelve. Thus, the same person, accessing the data of the same organisation, is presented with two different taxonomies according to the user interface. Fixing this would cost nothing.
Predictability
Of course you can be consistent but unpredictable; this happens when a corporation or network feels the constant need to change its taxonomy or navigation. This is a particular problem for people with learning difficulties and car drivers. For example, I am driving into Brighton and I log on, using audio, to the civic information site. If I know that its taxonomy is made up of nine elements the first of which is number 1 Arts and if I also know that under Arts the sub-class for number one is 'Today' then I can speak into my hands-free device and say "1, 1" and it should read the arts events going on today. It would be even more helpful if the same convention was adopted by Bedford and Barnsley so that anybody accessing a civic network would know the key numbers she usually wants. this may sound rather dirigiste but if we want our digital environment to be a network rather than a jungle we need to think about these ideas.
Adopting this kind of predictability might improve productivity but coordination would cost money.
This kind of consistency and predictability, however, needs to be underpinned by much more discipline in three areas:
Basic Design Discipline Environments
- User interface
- Taxonomy
- Displays
User interface design
15 years ago the 0-9 digital array was different for:
Varieties of numerical display c1990
- Telephones
1 2 3
4 5 6
7 8 9
- 0 - - Calculators
7 8 9
4 5 6
1 2 3
0 - - - ATMs
1 2 3 4 5
6 7 8 9 0
- Telvision/hi-fi remote controllers
1 2 3 4
5 6 7 8
9 0 - -
The last of these categories still presents problems. It may well be that the degree of standardisation will be rather lower than we would want and so an alternative would be to take advantage of user interface disaggregation and focus on giving each user an interface of choice that would work with all processors. Research into this idea is urgently needed.
Taxonomy
In many ways, as I hinted earlier, taxonomy is as much an intellectual choice as a pursuit of tidiness. Whatever your preferred taxonomy, however, it is becoming ever more clear that Miller's proposition on taxonomical memory [v] is being reinforced by investigations into taxonomical effectiveness. Miller's research showed that people could not readily hold more than seven classes of item in their head when putting items into classes but it is now emerging that the greater the number of classes of items the more difficult people find it to remember into which class they deposited an item [vi].
Miller's original seven was shown as (7 + or - 2). Building on this work, and taking into account the basic cultural user interface of the telephone, Roy Stringer and I [vii] proposed that the maximum number of classes and sub classes in any digital information array should be nine. There is an obvious balance between links and 'clicks' but any system based on individual or aggregate preference should be able to arrive at nine major classes plus 'other'. There is not currently enough research into the way people exercise preferences within relatively small taxonomies but my guess is that it will follow the 80/20 rule with 80% of choices accounted for by the top classifications and the rest being covered by 'other'.
Displays
Although, again, it is impossible to impose overall discipline in digital design, it would be extremely helpful if there could be some consistency in the way that digital information is displayed as a default within and across platforms. To give a simple example; you can think of television EPGs as a matrix of time and channel. It should not take much research to show whether the default should be time on the vertical axis and channel on the horizontal or vice versa. Where two taxonomies are shown in a matrix the general rule should be that the one of most relevance to the user should be vertical; this is because in systems where only cursor keys are used vertical navigation between classes is much more consistent and rapid than horizontal travel. So if the user is interested in what to watch now he will want to get the time right first than look for programme choices; but if she is a comedy fanatic she will want to get the genre first and then wander at her leisure through the available time slots. The key point, however, is that the general default should represent aggregate behaviour which can then be amended to suit an individual.
Widening the idea slightly, it is difficult to see why there should not be a convention about a standard differentiation and location for content selection and navigation prompts. Most web sites don't look like considered architectural drawings, they are more like Jackson Pollock pictures.
In the digital environment none of this is likely to cost money and it is likely that in standardising hardware features the cost of coordination would be out weighed by the saving on production costs.
Finally, I want to make a major point about our basic cultural assumptions. If you think again about the BBC it is obvious that it is a broadcaster which projects images and yet all its searching is text based. How strange! There is a current debate about whether an 'image wall' to supplement the text taxonomy should contain only personalities and shots from programmes or whether it should try to use abstract ideas. Should there, for instance, be Marlene's hair care or should there be a pictograph for hair care? Again, research is needed here but the general idea of an 'image wall' is utterly sound, as long, of course, as the images are the default with a prompt to change to text! This will require some initial investment but it will vastly increase uptake.
In summary, then, this new era of convergence cold present us with a massive problem but such attributes as:
Helpful Accessibility Attributes
- Default
- Aggregate and individual ranking
- Customisation
- Consistency; and
- Predictability
Should be encouraged not only for people with special needs but also for the majority of the population. When I was working on the Sky EPG we had a motto: "If you can't find it, you can't buy it!"
[i] Disability Rights Commission (DRC UK): The Web: Access and Inclusion for Disabled People: A Formal Investigation conducted by the Disability Rights Commission, TSO, London, 2004 http://www.drc-gb.org/PDF/2.pdf
[ii] World Wide Web Consortium: Quick Tips to make Accessible Web Sites from: Web Accessibility Initiative (WAI): Web Content Accessibility Guidelines WCAG (Version 1.0, 2001, http://www.w3.org/WAI
[iii] http://www.vista-epg.org.uk/
[iv] Carey, Kevin.: 'A Comparative Analysis of Guidelines for Access by Disabled People to Digital Information Systems : Some Proposals for Simplification': Inaugural Lecture, City University, London, 25th May 2005 (available on request humanity@atlas.co.uk)
[v] Miller, George A.: "The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information": The Psychological Review, 1956, vol. 63, pp. 81-97 http://www.well.com/user/smalin/miller.html
[vi] Ormerod, Professor Tom, University of Lancaster 'Lost then found: Applying psychology to problems of digital image handling. PACCIT, Royal Society, 27.6.06 http://www.paccit.gla.ac.uk/public/RoyalSociety/
[vii] Carey, Kevin. and Stringer, Roy.: "The Power of Nine": Library and Information Commission Research Report 74: London 2000: ISBN 1-902394-46-1 - ISSN 1466-2949 & 1470-9007