RE: Costs of testing with Silver

Hi John,

> merging of those ideas is counter-productive (or at least confusing), as Wilco started this thread specifically about the CO$T of testing, and not of feasibility or testability.

There is overlap, in the non-text contrast SC the issue (that didn’t win the argument) was that it would take too long to test all the images, it wasn’t automatable. “Testability” has included whether it is possible to test in a reasonable time.

An example for implementation was accessible authentication, there were several methods [1] to avoid having to rely on passwords. However, there wasn’t a browser-native method so any organisation implementing it would have to do so independently, and currently that is relatively expensive.
(Since then webauth became a recommendation, we should revisit that.)


Re plain language:
> my concern then was around testability and scalability, which we could not get agreement on.

How is that different from cost? With infinite time and the creation of common dictionaries per topical domain you could test it, but I agreed that it was not practical in a WCAG 2.x context.


> we are here to create the right technical requirements and guidance to benefit people with disabilities.

Agreed, and we aren’t disagreeing very much. My point is we won’t have technical requirements to meet some of the user-requirements, so whatever conformance model we come up with will need to account for that.

I’m suggesting the structure starts with user-requirements, with general & per-technology criteria underneath that. Each criteria could be assigned a level, and certain criteria may not be technical requirements (i.e. not like the binary content requirements we have now).

Cheers,

-Alastair

1] https://lists.w3.org/Archives/Public/w3c-wai-gl/2017OctDec/1037.html