Re: After today's call (Comments to John Foliot's alternative scoring proposal)

Hi Patrick,

>  *SC like 1.1.1 in WCAG 2.X requires a huge amount of subjective and
contextual analysis from an auditor (and two auditors may still disagree in
the end) once you move beyond a binary "has an alternative / doesn't have
an alternative" and into "but is the alternative actually
appropriate/sufficient/comprehensive" ... I have my own doubts that this
could ever be properly achieved going forward.*

And yet, this is the exact problem we have today: evaluators attempting to
subjectively decide (for other users) whether a textual alternative is
"good enough". Says who?

I for one remain troubled with that ambiguity as an overall resting place
today, so how do we improve on that? And more importantly, how do we
account for the problem of subjectivity in a regulatory fashion? (Or, to
view this through a negative lens, how do I *prove *to the judge that my
text alternative is sufficient, when the litigant says it isn't?)

Remember, my proposal is for *measuring conformance*, NOT for measuring
usability, which I argue cannot be measured at scale - ever. (And it is at
this point that I usually reference Jamie Knight's oft-quoted statement
about after his presentation knowing the needs, issues, and perspective of
*one* person with cognitive disabilities...)

HOWEVER, to your specific reference to SC 1.1.1: we have today multiple
references on how to create appropriate text alternatives
<https://www.google.com/search?q=how%20to%20create%20appropriate%20text%20alternatives>,
including one directly from the W3C
<https://www.w3.org/WAI/tutorials/images/decision-tree/> - a 'decision
tree', which broadly speaking is a type of Protocol (ref:
https://www.merriam-webster.com/dictionary/protocol - defintion 3B),
whereby it looks at possible scenarios, and sets out 'objective goals',
while at the same time not attempting to define what is or isn't "good".
And as I noted in my deck, we already have similar kinds of 'Protocols'
today, including the US plainlanguage.gov, and the EU
https://www.plainlanguageeurope.com - clear and useful guidance on how to
meet the objective of writing in plain language, without specific
mechanisms for 'measuring' success or failure.

So, in my proposal, if an entity "publicly promised" to use a Protocol as
part of their internal authoring and UX style guide(s), then at least I can
feel somewhat confident that they are using vetted best guidance on how to
achieve success *in the context of their application/content. *And since
Patrick we both know that fundamentally a lot of "how to succeed in
accessibility" is an educational problem, adopting protocols (which are
instructional in nature) is my suggestion on how to move that needle
forward. (And while I use enterprise-levels terms like authoring and UX
style guide here, even the smallest of entities can "inform themselves" by
those protocols, without formally adopting them as an internal document -
remember, in my proposal, all they are doing is making a public promise to
adopt and use the protocols - there is no mandate on how they do that, only
that they do.)

Does this then guarantee that all alt texts will be "better"? Not at all.
But it does hopefully codify the 'learning' piece a bit more, because in my
proposal, entities will be making "public promises" to do certain things,
which I can both go read and then return to your site/content and evaluate.
But because that good guidance lacks a measurement piece in the first
place, it specifically allows for subjectivity. However, for my
hypothetical judge, he can now go read that subjective (and
hopefully neutral) guidance himself, and then apply his 'judgement' (in the
role of judge) to what you have authored. So adopting protocols helps
contain, constrain, and guide subjectivity.

> *Unless you remove all instances where a human has to make a value
judgement, and reduce all guidelines to purely technical/mechanistic ones
(which yes, will then be nice and unequivocal, but also won't have much
relevance to a user's actual experience when using the content).*

Effectively, from a pure-play *conformance* perspective, that is exactly
what I am proposing Patrick; that attempting to measure and evaluate
subjective determinations is a very real problem today, and adding more
subjective evaluations in WCAG 3 increases that problem, it doesn't reduce
it. So yes, in my proposal, they are no longer "guidelines" but rather unit
tests which can always be unambiguously tested for: pass or fail, and
subjective determinations are handed off to Protocols.

I do understand that this is somewhat shocking to many, but I personally
cannot see another path forward, and it is clear from the first round of
feedback that industry are very unhappy with the amount of additional
subjectivity in our current draft, so I assert we do need to come up with
something different.

This then was my suggestion, but of course others are encouraged to offer
alternatives as well.

JF



On Thu, Aug 12, 2021 at 5:38 PM Patrick H. Lauke <redux@splintered.co.uk>
wrote:

> On 12/08/2021 22:01, John Foliot wrote:
> > So yes, I am proposing that we only "measure and score" that which can
> > be accurately and uncontroversially measured and scored; that as a goal
> > we seek to squeeze outthe known subjectivity we currently have in WCAG
> > 2.x even further, and that we avoid at all costs adding more
> > subjectivity to the scoring and conformance
> Just tangentially on this particular point then, considering that even a
> seemingly simple SC like 1.1.1 in WCAG 2.X requires a huge amount of
> subjective and contextual analysis from an auditor (and two auditors may
> still disagree in the end) once you move beyond a binary "has an
> alternative / doesn't have an alternative" and into "but is the
> alternative actually appropriate/sufficient/comprehensive" ... I have my
> own doubts that this could ever be properly achieved going forward.
> Unless you remove all instances where a human has to make a value
> judgement, and reduce all guidelines to purely technical/mechanistic
> ones (which yes, will then be nice and unequivocal, but also won't have
> much relevance to a user's actual experience when using the content).
>
> But yes, I'm sure that industry/big players would love this sort of
> mechanistic approach that can be applied at scale by just running
> something through Axe or Lighthouse or whatever. Or something that can
> just be remediated by putting a mechanistic plaster/overlay on top of an
> existing site.
>
> P
> --
> Patrick H. Lauke
>
> https://www.splintered.co.uk/ | https://github.com/patrickhlauke
> https://flickr.com/photos/redux/ | https://www.deviantart.com/redux
> twitter: @patrick_h_lauke | skype: patrick_h_lauke
>
>

-- 
*John Foliot* |
Senior Industry Specialist, Digital Accessibility |
W3C Accessibility Standards Contributor |

"I made this so long because I did not have time to make it shorter." -
Pascal "links go places, buttons do things"