The winning design from a recent competition by the Guild of Accessible Web Designers (GAWDS) has several accessibility issues which have been discussed on the GAWDS mailing list. One of the issues mentioned was relying on colour alone for links in the main content.
Author: Gez Lemon
GAWDS Redesign Competition
The Guild of Accessible Web Designers (GAWDS) recently held a competition to redesign their site. From what I understood, the purpose of the competition was to lead by example, and illustrate that accessible designs don't have to be boring. Unfortunately, the winning design has a few accessibility issues, as do many of the other entries.
Relying on Colour Alone
One of the comments I made about the winning design was that links were identified using colour alone, and assumed this contravened guideline 2 of WCAG 1.0; in particular checkpoint 2.1: Ensure that all information conveyed with colour is also available without colour, for example from context or markup (a priority 1 issue). According to Mel Pedley and Jim Thatcher, I'm wrong in my assumption. I'm prepared to accept I may have missed the point of this checkpoint, but their reasoning does not convince me. In response to my comment, Mel Pedley wrote:
Having looked at the design sans style in a graphical browsers and using Lynx (black on a grey background, I don't see how this impacts. It's perfectly accessible without colour and, as such, remains in line with WCAG 1.
The links were not distinguishable on the monitor I was using, and certainly weren't when I printed the page. I know, I know; I need a new monitor, replace the toner in my printer, and get new glasses. Having made such an assertive decision on behalf of people with visual impairments, Mel went on to explain how it's difficult to select a colour that would work for everyone:
I'd actually be very careful about trying to even attempt some sort of 'ideal' colour contrast within *any* design. No such beast exists. The needs of the various groups often run completely counter to one another. What benefits the visually impaired (high contrast) actively discrimates [sic] against the dyslexic (low contrast with strong preferences for the red/yellowe [sic] spectrum). Neither extreme is of much use to those with cognitive problems (high colour) or those with forms of autism (shades within a colour). Then there are people suffering from the various forms of colour blindness (require coloured elements to lie at opposing points in the colour spectrum).
Which left me wondering why she was so sure the colour issue was of no importance, and remained inline with WCAG 1.0.
Jim Thatcher shed some light on the issue, offering a variation of the example the W3C offer on Techniques for checkpoint 2.1.
Checkpoint 2.1 has nothing to do with contrast or CSS, but the problem of "fields marked in red are required" or push the green button. Don't convey information bv [sic] color alone. Use other means in addition to color. "Fields in red with * are required" and "push the green OK button."
Interestingly, the W3C's example comes under CSS Techniques, which is a strange place to put the example if this checkpoint has nothing to do with CSS. The only visual clue for a link in the GAWDS winning design is provided by colour alone. When someone else highlighted this to Jim, and suggested this could be a problem, he provided the following response:
Nope. Links and their attributes (like visited) are conveyed in the object model and reflected by mouse pointer, by focus rectangle, by "link" in a screen reader or high pitch voice in home page reader.
As Derek Featherstone of WATS.ca correctly pointed out, the object model isn't exposed to visitors, except in the final rendering of the page. Derek went on to point out that visitors unable to distinguish the links would either have to hunt them down with the mouse or a keyboard, unless they were fortunate enough to be using a screen reader.
The biggest clue as to whether checkpoint 2.1 applies to links can be found near the bottom of the W3C's CSS techniques example:
This will show you where you need to add redundant cues (example: hyperlinks are usually underlined on Web pages), or whether the cues are too small or indistinct to hold up well.
Not only are they talking about CSS, but explicitly mention links in the example. My interpretation of checkpoint 2.1 is that it is applicable to links. If links are part of a navigation system, providing a clue other than colour isn't applicable, as it would make sense by the context. When links are intermingled in the content, my advice would be to ensure you relay the fact it's a link by some other means; such as underlining the link.