An interesting debate in the accessibility community was whether it made sense to accessify the artistic tag cloud service Wordle.net. This is a service which converts text from either a Web page or a pasted text into an artistically rendered tag cloud like the one below based on the Penn State Education Technology Services Homepage
As you can see the output is an abstract art piece composed of all the keywords from the text laid out in a jigsaw pattern. Some are horizontal, some vertical and all are different sizes. You can choose fonts, color schemes, and layout segments. Is this worth accessifying? Note that gallery images don’t even include a basic ALT tag.
When asked about making an accessible version of Wordle, the developer Jonathan Feinberg and others stated that since the output was abstract art, creating a non-visual representation was irrelevant. This is an interesting counter argument because even accessibility experts recommend blank descriptions for decorative images on a Web page.
But is Wordle output completly visual? Not really because the output is extracting the list of words from the Web page, and further makes the more frequently words larger. If you click on the Edit, you will find an option to open a list of words found, copy the list and paste into another file. This is fairly close to being accessible…assuming the screen reader can process the Java menus.
Still when these facts were pointed out to Feinberg, his argument was that the visual aesthetic impact was most important, and thus makes accessibility irrelevant (at least for screen readers). More importantly, it would require him to recode the entire site to make the lists for older items available (now I think we discover the truth).
But are beauty and accessibility incompatible? First, the accessibility community isn’t seeking an equivalent aesthetic experience, but an equivalent informational experience. For instance, a text list would tell all users that the second most important word on an educational technology site is “students”.
Second – if you were interested in aesthetics, why not match the colors and fonts with tones and voice styles? What would be the abstract tonal possibilities? I remember a seismic monitor of Mount Vesuvius which mapped a tone to seismic vibrations. Normal vibrations made soothing background music, but the stronger the vibrations, the louder the music. It was a perfect match of beauty and information.
What’s the upshot? It should be noted that Feinberg developed this as a side project and it’s not for profit (right now). He is not really obligated to add any enhancement unless he feels the desire to.
On the other hand…I think Feinberg’s comments really show a lack of understanding about accessibility. The accessibility community suggested some alternative representations (the list of words and word counts), but Feinberg commented that either users know about better text analyzers or that a third party take “30 minutes” to implement the code.
Ironically though, Feinberg is willing to make changes to accommodate the quirks of Devanagari and Arabic scripts (not easy). It’s a shame that he can’t accomodate a simple ALT tag or link to a text file