Re: Moving personlization forward - wording suggestion

HI Lisa,

It misses several of the points I raised last week [1, 2], I’ll try to make a simpler list here:

  *   The first bullet could be satisfied with a button on the page that adds icons to two links. It is a loophole for people who want to do the minimum. Also, for those with good intentions there is no apparent scale of what *should* be done. I.e. include everything from the coga-personalisation spec, or just a sub-set?

  *   Some of the coga-personalisation spec is defined in a similar way to ARIA where there are clear ways to apply it. However, some is very open (e.g. coga-context) and there is no concrete guidance for how or when to apply it.
You would have arguments like “I think the context should be gender-based” whilst someone else thinks it should (also) be location based, with no way to resolve those arguments.
Those parts of the spec are not a suitable basis for a WCAG SC, although they could be provided elsewhere.

  *   Achieving personalisation with the 1st bullet (on-page) appears to be more work than with the 2nd bullet (meta-data only), as you would have to define the context for both, AND then build the mechanism for the page.
Why have the 1st bullet?

  *   It is not clear if the author is responsible for what happens when a user applies their own icons (expanding the links). Just applying meta-data is one thing, but in practice that will fall apart if there is no specification for what should happen when it’s used.

  *   There does not appear to be any user-side technology to use this yet. I followed the BBC link but hit a dead-end, it appears the company no longer produces an icon-browser.

Overall, rather than trying to get ‘something’ in and then throwing *everything* in, we need to take a similar approach to other areas of the W3C in moving the platform forward (as Chris, David and Jason have said in several ways). That means starting with a focused set of things to achieve.

I recommended removing the first bullet and focus on a 1.3.1/4.1.2 approach for the remaining requirements, so essentially relying on the personalisation spec for the details.  That brings me back to a previous suggestion [3]:

“1.3.x Contextual information: Where a defined vocabulary is available, contextual information can be programmatically determined for sections and controls.”

However, for that to work the spec will need to remove the ‘open’ classifications as they are too subjective to form a basis for testing.

Kind regards,





Received on Tuesday, 4 July 2017 14:56:05 UTC