Re: "testability" and "threshold" issues for cognitive -- and how to solve them?

> On Dec 11, 2025, at 7:21 AM, Lisa Seeman <lisa1seeman@gmail.com> wrote:
> 
> Hi  Gregg
> 
> The history of why COGA proposals did not make it through the AG consensus for 2.x, can be looked at in many different ways. I suggest we do not try to agree on this beyond saying they were not included (for the most part).

Lets agree to disagree on this.   COGA was never excluded - and has been a top priority since 2.0.      

> We can probably agree that  structural changes are needed if we want to be more equitable.

If you mean structural in the WCAG guidelines — then 100% we agree.    But I would say we need to do it to be able to get non-testable and non-requireable methods in front of authors eyes so they see them and more follow them.

> 
> In terms of AI being the right direction for better inclusion for Cognitive and learning disabilities (COGA).,. 
> 1. Most of WCAG could be replaced with good AI when it arrives. For example, the purpose and function of images could be achieved by AI. This is something we should build into WCAG 3. That will be good. 

Agree

> 2. However AI makes mistakes. Unfortunately people in COGA use groups are not consistently able to notice the mistake and compensate for them. The result is as a user group they are more likely  to be confused or misinformed by AI. In fact, there are also cases where they can be harmed by AI.

Agree

> So lets word WCAG 3  so that if AI on the user end can achieve an outcome, then the authother does not need to do it. But let us not ignore Cognitive and learning disabilities until this technology is really reliable. 

Agree.   This is the most important point we need to emphasize (and I include it in every presentation I  make on AI).

AI can do some things today
and it can do some things today better than 80-90% of humans
And it will be able to do many more things in the future 

HOWEVER - we should take very great care that we do not stop doing what works now — before we have proven (and consumers accept) that the AI can do it as well as or better than humans.  

And the best way to allow this to happen (since it will happen unevenly across different aspects) is to do  exactly what you mention above.   That is — to write the requirements such that the say what must be true — and that you can use or depend on AI to do it when - and only when - it is demonstrated to be sufficient for causing that thing (that requirement) to be true.    (not a dream -  but true in the real lives of people with disabilities as they use the web)

> 
> In terms of your classifications. Having different versions, such as a summary or easy language version is a great potential method for inclusion.

Not sure what this refers to — but it is impractical to have a different version of the content for every different level and combination of cognitive, language, and learning disabilities.      So the best approach will be when users can tune the level of both the linguistic and the interface  of the content to match their ability — and to be able to change it as their ability changes over time or even over a day or context.

> 
> In general, I would define accessibility as:
> Where the user could theoretically use the content, but is unable to because of the combination of:
>  their disability and  
> the design choices of the author ,
> then  that content is not accessible. 

That is very good.

I  would change the word  “theoretically”  to  “otherwise”

and I  would change “their disability”  to  their abilities  

and add   “mismatch between’


I generally define “disability”  as something that occurs  (not is but occurs) when there is a mismatch between a persons abilities and the design of the product.

I  also prefer to say that people “experience” disabilities rather than “have” disabilities. — but that will not work if we want to have funding provided for people who are experiencing disabilities.    People are more likely to provide funding, AT, training, or accommodations to “disabled people” or “people with disabilities”  than they are to people who are just  “experiencing” them.  (sounds like situational disabilities).

> 
> Hope it helps

Yes - very much.  And very nicely written response as well

thank you

All the best

Gregg


> All the best
> Lisa
> 
> On Thu, Dec 11, 2025 at 6:18 AM Gregg Vanderheiden RTF <gregg@raisingthefloor.org <mailto:gregg@raisingthefloor.org>> wrote:
>> Since my last email ws TLDR  -  let me repost a part of it separately 
>> 
>> 
>> I DO agree - that we there are critical things that are needed by people with cognitive disabilities that we have not been able to figure out how to make requirements for.  
>> Testability and threshold are both problems that have prevented us from having more requirements.
>> (The whole reason I  am working on WCAG 3 in retirements is that I am trying to figure out how to get more good guidance for people with cognitive disabilities into WCAG 3.    
>> And I  am also in my retirement working on an alternate approach that would be really great for people with cognitive disabilities.   One that would remove/solve both the testability and threshold issues. 
>> 
>> By “the threshold issue" I  mean the following.   
>> Just for discussion sake, let's say all people with intellectual disabilities, fell into five levels. 
>> If we could find a measure of simplicity (of text for example)...
>> which level of simplicity should we require  (what level of simplicity would be what was required?)
>> If you said 3, then people who were 4 and 5 would have to use text that was simpler than what they need (and they would have to be losing the information between level 3 and the level they were able to handle. 
>> And the content would still not be simple enough for people at levels 1 and 2. 
>> So do we discriminate against levels 1 and 2 and leave them behind?  (and simultaneously force levels 4 and 5 to use oversimplified text?)
>> 
>> Also, (as we have all experienced) once we understand something at a simple of a we can often understand it at a more complex level. How many times have you not understood something until someone said "let me give you a simple example”.   Then, once you've seen a simple version of the concept, the broad or more complex idea they were trying to express is suddenly clear. 
>> Thus, even for a single person there is no correct level of simplicity. 
>> There is the level that is needed for them to understand something new, 
>> and then there is a different level that they can handle once they understood it at the simpler level or if they are already familiar with the concepts.
>> 
>> I suggest that a much better approach 
>> is to work to have it so that each user can adjust both the language level — and in level of inteface complexity - to whatever level they need.
>> and this would be done at the user agent level (browser or AT) so that it could be set by the user and used across websites
>> and would allow users to select the type of interactors that they prefer and have consistency not just in a website but across websites
>> 
>> I think we should push for smart assistive technologies or smart browsers 
>> that can take whatever the content is on the page and present it to the individual at whatever level they need to be in order to be understood. 
>> This would not only eliminate the need for us to have a "testable" level of readability or consistency (which I think we will never have) 
>> but also eliminate the threshold problem described above. 
>> It would allow individuals to be able to adjust the text to whatever level works for them. 
>> Moreover they could do it dynamically 
>> so that they would only have to ask for simpler versions of pages that they had trouble with, 
>> and would only have to simplify sites as much as they needed to to understand it. 
>> Also, as noted above, once they understand the site basically,  they may be able to turn the complexity up to a higher level and still understand the site more completely.
>> 
>> This is the equivalent to having AT for someone who is blind or has low vision 
>> and has the ability to adjust the magnification to what they need but not more 
>> or to adjust the reading speed or other factors to be matching what they need but no more.
>> 
>> 
>>