Re: Functional Outcomes MUST cover all benefits OR must be duplicated

Hi Jake,

Thanks for raising thoughtful questions.

We have been looking at these issues in the last couple weeks as we 
build out the content and test the new conformance model. Sarah Horton 
did a lot of work last week with a spreadsheet model to test the 
consistency of the content and look for patterns.

I recommend reading the minutes from Friday and look at the consistency 
spreadsheet.  The recommendation we looked at Friday was to roughly 
define Outcomes as having an AND relationship with the guidelines (you 
have to have [this] Outcome AND [this] Outcome AND [this] Outcome.).  
There is an OR relationship with Methods: (you can do it [this] way OR 
[that way].  Most of the time, the OR will be technology related and 
probably wouldn't need a decision tree.  In the case of Alt Text, 
decision trees have been valuable tools in the past and could be useful 
going forward.

I want to emphasize that this model needs more detailed testing, but as 
a model for moving forward to FPWD, this looks viable and I think 
addresses both of your questions.  Please bring any additional concerns 
to the group.  It is helpful to keep working out the details of the model.

1) Email summary of minutes of 4 September 2020 
<https://lists.w3.org/Archives/Public/public-silver/2020Sep/0010.html>

2) Consistency spreadsheet 
<https://docs.google.com/spreadsheets/d/1_Vu0ix-d-Qrv1wDZYQhfUX6jICE_bRalypp1rtcie8w/#gid=1109648765>

On 9/7/2020 3:46 AM, jake abma wrote:
>
> Hi all,
>
> Just another issue we must have correct or discuss at least before 
> publication I think.
>
> --------------------
>
> As Guidelines are not normative but (Functional) Outcomes are, they 
> must cover all benefits for all Functional Groups and Functional Needs 
> we try to tackle.
>
> This means the "so... bla bla" statement should be broad enough to 
> cover all benefits OR a bulleted list might be needed with the 
> benefits (and are the benefits normative then?).
>
> --------------------
>
> On the other hand, if we use bulleted lists for Benefits, then all 
> methods and the scoring / tests MUST cover all benefits also otherwise 
> they are not compatible (Charles Hall commented on this in the 
> functional needs subgroup).
>
> --------------------
>
> If this is not a "Catch All" for (Functional) Outcomes, we might need 
> to split / duplicate Outcomes covering different Benefits (?!)
>
> --------------------
> EXAMPLE 1
> --------------------
>
> "Provides semantic structure So can convey a sense of hierarchy"
>
> In this case the benefits of navigating or locating are not mentioned, 
> also the Functional Needs are not covered as it's not in the normative 
> text.
>
> Three options for this example:
>
> 1. (long sentence, covering all benefits)
>
> "Provides semantic structure So can convey a sense of hierarchy AND/OR 
> users can navigate AND/OR users can locate"
>
> 2. (use of bulleted list)
>
> "Provides semantic structure
>
>   * So can convey a sense of hierarchy
>   * So users can navigate
>   * So users can locate"
>
> 3. (split in 3 Functional Outcomes)
>
>
> "Provides semantic structure so can convey a sense of hierarchy"
> "Provides semantic structure so users can navigate"
> "Provides semantic structure so users can locate"
>
> --------------------
>
> This is just an example of the challenge with the Functional Outcome 
> texts being normative, more examples are not difficult to think of.
>
> Another option would be to separate the Benefits from the functional 
> outcome and mention them as something like: " Benefits might be but 
> not limited to: bla, bla and bla"
>
> --------------------
>
> At the moment I think the Functional Outcomes as we have now are to 
> open to interpretation and probably will not make it as normative text 
> to be tested and scored.
>
> Of course happy to illustrate of dsicus.
>
> Cheers,
> Jake
>
>
>
>