Silver Task Force & Community Group

06 May 2020


Lauriat, ChrisLoiselle, CharlesHall, PeterKorn, JakeAbma, Fazio, bruce_bailey, kirkwood, Makoto, sajkaj, Joshue108, JF, KimD, OmarBonilla, Rachael, MichaelC, KimD_, jeanne, AngelaAccessForAll
Shawn, jeanne
ChrisLoiselle, JakeAbma, sajkaj


<ChrisLoiselle> scribe:ChrisLoiselle

<CharlesHall> away for a few minutes

<bruce_bailey> Regrets for most of todays' meeting; i can only attend 1st hour and last half hour :-(

I'll scribe first hour

<kirkwood> Will need to drop after 1st hour

Scope of a task based assessment

ShawnL: Task based assessment and scoping

A task can span multiple pages. Could be based on parts of pages but not all page. What is the scope of each of those parts?

Testing in pages that aren't directly related to tasks. I.e. a link in a footnote that isn't related to a task. Or a heading structure above a certain task prohibits the actual task, how does that impact user in terms of criticality and severity of task?

PeterK: Can a task have sub tasks?

<jeanne> https://docs.google.com/document/d/18saMLdtaVlwuVodpZX2-AcIQ8fCvDRxYdZw4VpB2qKo/edit#heading=h.crww2inpwbvp

ShawnL: Yes, we mentioned an ACT structure. Compound and atomic were mentioned. Tasks that make up a goal. Subtasks that make up interaction.

<Lauriat> From yesterday: "https://www.w3.org/TR/act-rules-format/#expectations-atomic and https://www.w3.org/TR/act-rules-format/#expectations-composite were mentioned by Jake"

Jake: I re-read the ACT rules format for composite and atomic test rules.

The rules the ACT rules were set up , if you have atomic and composite rule, the task or subtask could be the next step.

We should look at the ACT rules on how tasks fit into that model

<kirkwood> I want to make sure we don’t forget links/references to legal “mumbo-jumbo” within a task sequence.

The only thing that ACT doesn't have currently is low, medium , high ranking structure.

CharlesH: Atomic vs. composite aligns how tasks are typically thought of in User Experience.

Challenges remain on ancillary items on how task is approached.

We need to describe a preferred path of high level / composite task

<Lauriat> Related, from Rachael Friday: "From https://uxdesign.cc/start-designing-with-goals-in-mind-4723df5f8e88 Task: A task is a step taken by the user during the process of achieving a goal. In order to achieve a goal, users have to navigate through multiple steps completing small tasks. The information architecture of a digital product is formed by tasks."

<Fazio> Wouldn't that a be apart of the owners specified conformance claim?

Jake: Should we provide a test framework for tasks rather than each individual task? CharlesH: We shouldn't be prescriptive on the composite level.

<Zakim> sajkaj, you wanted to ask whether there's an "incorporate by reference" concept in ACT

Janina: Any sub task may fit in various paths or flows. Re-documenting is not efficient. We should be able to incorporate as reference somehow.

Jake: If tasks is a third level that follows composite or atomic, where do we go from there? We could provide some examples, but the main goal is not to get trapped in complicated measurements of outcomes.

<Lauriat> +1 to Jake

JF: If we start using task based test for conformance piece , how are we going to talk to all tasks conforming? I.e. what is essential vs. not

<Fazio> omitting tasks would put them in legal jeopardy

<Fazio> a user can claim x y or z was never tested

ShawnL: These x the tasks included in conformance benefits the user. JF: These x tasks are are excluded. ShawnL:The conversation is around benefiting users based on tasks and what users are trying to do

<jeanne> +1 to Shawn

JF: Right now, this is at a page level including third party content. If we move to task based conformance, that we don't let people avoid non critical tasks.

DavidF: CharlesH's example on Norman's subtasks is a great resource. Leaving it up to the owner is included in the conformance claim. If they leave things out, they are liable from the legal standpoint.
... If I couldn't complete this task, here's the demand letter to fix the issue...

CharlesH: Scope should include composite and atomic level items. I.e. learn about our consulting services. The user may review comments on the blog posts. If the conformance statement includes blog posts, we wouldn't be able to fully state that the scope was conforming.

<Fazio> What if that wasn't intended by the designer?

ShawnL: If we provide from a technical standpoint, how someone can build toward conformance, it provides structure to how someone would achieve it. I.e. the owner and user have the ability to express on what they tried to do.

<Fazio> Again to Wireframes. They provide a stable baseline for task testing and claims

CharlesH: Scope includes what the owner believes what the sub tasks are.

PeterK: The legality discussion is worrisome. Going by task clearly specifies what they are and are not looking for. There may be multiple ways to complete a task.

I believe that reviewing by task, it benefits the user.

ShawnL: We need to look at how we are building up conformance around a task.

Jake: Parts of a page vs. entire page and conformance. The ACT rules talk to test cases. They look like tasks. Test cases are snippets of content. Each rule must have one or more test case. A test rule looks like part of a task.

The ACT Rules format does not prescribe what format ACT Rules can be written in (e.g. HTML, DOCX, PDF, etc.). However, ACT Rules must be written in a document that conforms to the Web Content Accessibility Guidelines [WCAG] or a comparable accessibility standard. This ensures that ACT Rules are accessible to people with disabilities. ACT Rule test cases are allowed to contain inaccessible content. If any test case contains accessibility issues[CUT]

Test cases are (snippets of) content that can be used to validate the implementation of ACT Rules. Each rule must have one or more test cases for passed, failed, and inapplicable outcomes. A test case consists of two pieces of data, a snippet of each input aspect for a rule, and the expected outcome for that rule. Test cases serve two functions, firstly as example scenarios for readers to understand when the outcome of a rule is passed, failed[CUT]


<CharlesHall> what Jake read aligns to the essential / useful / different tasks for criticality

JF: The reporting level. What is the unit of measure?

<PeterKorn> +1

Right now, it is the page. Partial conformance statement is there. If we are moving away from "the page" and into the "site" , how are we reporting that?

If tasks are being scoped, what are they, how are we grading / scoring?

<CharlesHall> perhaps the unit of measure simply replaces ‘page’ with ‘scope’?

<Fazio> I thought we were scoping Primary Purpose of site (all tasks - sub tasks)

Comparing VPATs , VPATS offer feature comparison. How do we include unit of measure and how do we review this?

<sajkaj> How do we declare a claim?

PeterK: Can we explore how this is being applied? I.e. we are looking at a website, here are the defined tasks. This is how it may be presented. Looking at it in totality would be beneficial.

<Lauriat> +1 to Peter

<Zakim> jeanne, you wanted to answer JF

Jeanne: Building on Peter's comments. On the proposal we were working on, how do we make reporting transparent? That reporting is what we are working on.

JF: A report card or dashboard is one way we help our clients. Scoping tasks , all tasks should be scope. What is critical / essential vs. what is not?

Jeanne: Please provide a proposal to help move this forward.

<CharlesHall> +1 to Jake. the scope could theoretically be a page if the task is contained

<Lauriat> +1

Jake: What if someone only provides building blocks for a page or a template? I.e. dashboard would be page based. However the dashboard should also talk to component level or task based dashboard as well.

<JF> How do you do comparative evaluations when everybody writes their own rules?

How to derive a representative sample?

ShawnL: For example, Google docs. Bolding text has many ways.., keyboard commands, voice commands, search, etc. For a task and individual steps to achieve the goal , there are many paths . The building blocks for each path are many.

Representative sampling / building blocks need to be defined. We need to make sure we cover the representative sample.

PeterK: Some of the ways of completing the task may be more comfortable for certain disabilities. I.e. cognitive disability based users may benefit from voice command rather than other ways of completing the task.

<kirkwood> +1

<Lauriat> +1

I think we should explore this.

<CharlesHall> Shawn’s description where some paths for the same task are out of scope is what UX calls the “happy path”. but we know that users will find other paths.

<kirkwood> I will help with that

Jeanne: I think that is a great idea. Perhaps a sub group to work more on this.

<Fazio> Good point Peter I encountered this with a major fast food client doing UX R&D our blind users every time wanted to use "hey Siri" which was unexpected

CharlesH: How granular would the example Shawn provided have to be to include the tasks needed for completion of the goal? What sub tasks ?

Jeanne: We started looking at a solution towards this yesterday.

ShawnL: We do a combination of unit testing and path testing and manual testing etc.

<PeterKorn> 1-

integration testing (automated)

Jeanne: If you already did this test, would you have to repeat it?

Shawn: Goal would be to do it once for a given path. I.e. voice commands, entry point = audio from microphone. Then action is triggered...contents would be different based on action someone wants to take.
... How can you present this in representative sampling? How do we include building blocks. ?

<CharlesHall> i don’t think sub- or atomic tasks are fair to consider representative sample. to me instances or occurrences better align to the idea of representative sample.

Jake: To Peter's comments on voice commands. I was looking at severity, from a technical perspective. I.e. label / name would help in voice commands. From a heading level perspective, the criticality is a human judgement call , someone may benefit more than others, may be using different assistive technology. Criticality may be guidance documents.

How does this make it worse for each type of user?

<JakeAbma> scribe: JakeAbma

<ChrisLoiselle> Janina: Transparency is important. For reporting, executive summary then this is all the tasks and browser combos we are reporting on. The reporting structure might be very long. Is that based on entity testing ?

<ChrisLoiselle> Is it company or the testing company? Or a central registry of claims?

<ChrisLoiselle> Sure, thanks Jake!

<scribe> scribe: JakeAbma

<JF> +1 and thank you for looking at my concern from a different vantage point

SL: things will look more complicated before we can make it simple

Janina: we might need a template

JF: we need a rosetta stone, each site might have another task flow
... so it would be hard to compare

<Zakim> sajkaj, you wanted to talk about documenting incorporation

<Zakim> CharlesHall, you wanted to respond to Shawn

CH: in a blog we only need to check a representative example
... for tools, all paths need to be in scope, some are more critical than other

<Zakim> Lauriat, you wanted to compare instances of blog posts to instances of formatting you can apply

<JF> +1 to ">>all<< paths need to be in scope"

SL: for different ways of adjusting styling, there's more ways to achieve them
... all the different ways are different building blocks

<JF> @shawn, but some blog posts will be more "accessible" than others?

SL: agree with all paths will require testing

<JF> HUge +1 to that Shawn

SL: but we need a mechanism to ensure people don't game the system, by saying this path is accessible, so the other doesn't need to be accessible
... different people might need one of the paths, so excluding should not be allowed

<Zakim> jeanne, you wanted to propose a solution

<CharlesHall> i still see atomic tasks that comprise a composite task ≠ representative sample

<Lauriat> Charles: so a representative sample would cover a selection of composite tasks that cover an overall surface area of composite tasks?

<CharlesHall> yes

Jeanne: we have to have a way to test CR, propose to define a task, define a representative sample, but not every option / iteration possible. so we can get it out there .

<Zakim> JF, you wanted to return to the blog post example

<CharlesHall> +1 to JF (content impacting sample)

<jeanne> Propose: Define the paths and the units or widgets of code or interactions. Then define an intersection so that each path is tested, and each unit is tested, but not every path with every widget.

JF: need to take into account content from CMS systems used by large companies, universities etc.
... how will this play out when using WordPress is what I use as an example

<jeanne> https://docs.google.com/document/d/18saMLdtaVlwuVodpZX2-AcIQ8fCvDRxYdZw4VpB2qKo/edit#heading=h.crww2inpwbvp

PC: how will we get all examples and discussion with solutions documented (as in a parking lot) so we don't loose it

<jeanne> https://www.w3.org/WAI/GL/task-forces/silver/wiki/Conformance_Issues_and_Decisions#Unorganized_List

JS: I have a place and added the link in IRC to store all ideas and useful commnets

Talk through some challenging examples

SL: would like to talk about different and emerging technologies

JS: would like to start with a simple one, than go to the complex ones...
... example suggestion is a checkout / ecommerce example, puchasing comething

SL: "I want to purchase haircut accessories"
... "scissors"
... "can go through home page, search etc."

<Fazio> device also?

SL: lets start from the home page

<Fazio> mobile tablet pc?

<Fazio> yes

SL: mobile tablet

<PeterKorn> Tablet via web browser or app?

<Fazio> also yeah

SL: use navigation via hierarchy in web browser

<PeterKorn> Going via hierarchy is interesting - user studies may indicate that only a very tiny fraction of users do that instead of search

<JF> +1 to Peter

SL: via via categories I might end up at the scissors...
... now I'm presented a page with 10 different options with scissors

<Fazio> As cognitive I need use descriptions with so many options what do I need

<CharlesHall> a compare function

<Fazio> good idea

SL: order by price, with pictures, the material...

<Fazio> +1 Janina

Janina: comment by users

SL: comparison function
... next, I found the one and would like to purchase, find the mechanism to add to cart

<PeterKorn> You are making an assumption that they path to purchasing requires the intermediate step of being "in my cart"

SL: can be overview page OR details page

<Fazio> hence possible sub tasks

SL: I choose to buy via cart

<PeterKorn> This bring to mind (for a later discussion perhaps this "offsite") the importance of the concept of "happy path" in our path analysis

<CharlesHall> and payment method if not stored

SL: go to the cart and see overview with checkout button

<PeterKorn> Also is the shipping destination a11y as well...

<Fazio> Conformance Claim could list possible paths, designer's primary path identified by wirefraaming, then alternate paths with a timetable for testing primary oil down

<Fazio> on down not oil

SL: lets assume I've added all info and and process is done

<Zakim> JF, you wanted to ask if using shawn's model, does that make all "home pages" in scope?

SL: per guideline we need specifics for what needs to be in scope

<JF> You made a very dangerous presumption there shawn

<Fazio> I like this testing structure because it identifies a map to remove those readily achievable barriers, which is legalize language in accessibility law. It also provides justification for the map by identifying primary path down

SL; for the path I described the homepage is in scope, the ability to go through this page is part of what we need to work out

JF: we can't say users do or don't need or want parts of the content, we don't know that. Some users may still want / need it.

<CharlesHall> content on the splash page is orientation and wayfinding to verify i am in the right place

<Lauriat> -1 to the debate, I agree with JF on that.

<sajkaj> But it's the commercial interest to get you to also buy X

SL: other content, maybe related, is part of another path and should be judged at than path

<CharlesHall> there should be something similar to non-interference though

<PeterKorn> I think we also need to put the question of whether the path approach can be applied in a way that addresses John's concerns to a later agenda topic, so we can get through the more granular work of all of the details of the path approach

<Zakim> jeanne, you wanted to talk about how it is currently done today for for task base assessment

<sajkaj> There is an accessibility interference case, for blind users but also COGA, where relqated products interfere with obtaining the thing people came looking for

Jeanne: as long as I've been doing this work, there has always been a way of using both web page total and task base testing when doing audits

<JF> not *Objecting* - questioning

JF: Johns concern is that parts might be left out
... the combination is where the power of the Silver approach must be

<Lauriat> -1 to the sub-page task concept, tasks need declaring from what users would do, from how developers would make it.

<PeterKorn> +1 to Jeanne. Defining the task based approach for the places where tasks are best is different from saying that paths are sufficient to do everything.

<Fazio> Would that make it retroactive though?

SL: we need to focus on the user paths, not the developer building paths.

<Fazio> Like Socrates

<Fazio> LOL

<JF> +1 to also looking at buying a pizza from a site that features a game

<PeterKorn> I need to drop a few minutes early, to prep for my next mtg.

<PeterKorn> +1 to Jake's point. I think we can provide guidance on how this should be done, but then have to have some trust for the site creator.

<jeanne> +1 to Jake

<jeanne> -- but we need to figure out so we know how it can work

<KimD> I suggest using a model similar to this as a framework for how we approach tasks: https://www.userfocus.co.uk/articles/prioritise.html

<OmarBonilla> +1 to Kim

<ChrisLoiselle> scribe:ChrisLoiselle

I can scribe for first hour.

What did we learn from the first session and how does that impact the scoring proposals?

<jeanne> https://docs.google.com/document/d/18saMLdtaVlwuVodpZX2-AcIQ8fCvDRxYdZw4VpB2qKo/

Jeanne: Methods are informative, and can be updated easily. Methods have one more tests, (i.e. ACT tests).

<sajkaj> So scoring is informative, and different companies can compete on doing a better job of scoring?

Non interference with task completion possibly fits in methods.

How is overall score is presented?

The "how" of scoring is informative.

<JF> +1 to Janina

Janina: What is normative about scoring, other than you need to meet these outcomes, show us how you do it?

<Fazio> Isn't their complaint inconsistency?

Jeanne: Let us work out the details and then we can talk to philosophical differences later

<Lauriat> +1

<JF> scoping makes a big difference in scoring - ya

We talked about different proposals this morning on how to break down a task. Tasks have sub tasks. Tasks must be specifically laid out . It can be a single page or span multiple pages.

We are looking to ACT on composite vs. atomic tasks.

How to we include items that have been already tested?

Shawn: I don't think we may need to do that. I.e. filling out conformance claim on two reviews vs. one.

Janina: How and whether we say something depends on how we require them to specify the claim.

<JF> Q

How transparent is the reporting?

<Fazio> +1 JF technologies change updates effect older devices etc

JF: Relying on older validation reports , perhaps the older they are , the less important they are to the conformance claim. I.e. 2 1/2 years later . Jeanne: Timestamp puts date on the testing.
... I'm talking to multiple time stamps on a given report.
... Not just the reporting (timestamp) but also how does it impact the actual score.

<Fazio> good point JF

<Lauriat> +1 to working through those cases, definitely.

Jeanne: Tasks and sub tasks, what is the criticality of those tasks against the main objective or outcome?

<Fazio> Quantative analysis drives that decision in UX

<jeanne> https://www.userfocus.co.uk/articles/prioritise.html

Shawn: We define a task, these are the steps that make up the task. I.e. this is how someone would use the product. Testing with users validates that assumption.

<Fazio> +1 SL

Defining the tasks per a cognitive walkthrough alone is not ideal, but worthwhile. Testing with users is definitely very important.

<OmarBonilla> +1 to the importance of user testing

DavidF: Pulling from the conversation from last call on Peter vs. Shawn on how someone may want to do a certain task. In my own experience, we develop scripts and how we design the product, however users may totally go away from how someone would validate this.

JF: End users vs. large orgs have valid points of view on the topic.

<Zakim> Lauriat, you wanted to speak to the middle ground between automated and testing with users.

Shawn: Usability testing and everything in the middle of end result needs support. We need to support companies as well as end users.

<kirkwood> +1

conformance model needs to help those that may not be able to do that extra testing and review.

Jake: Taking into account assistive technology and all users is very hard to do. Staying close to a composed rule will be better to guide . Going one more small step on severity may be beneficial. Criticality is difficult is hard to define. Critical for who?

Jeanne: Defining a task based scope was typed on the task completion scoring document.

The unit of measurement is more flexible. We don't want it to be a page. Moving away from pages.

Rules need to be set up for task completion side of things vs. representative sample method (i.e. wcag em scoring method)

It can vary by guideline. If a test doesn't exist for that technology, then they test for functional outcome for each guideline.

If task completion flow is used, it is used to determine the criticality of test results

<jeanne> ChrisL: Are we also looking at frequency of a failure? If something has already failed in one area, are they being failed in multiple task flow.

JF: How do we compare apples to apples ? I.e. newspaper vs. car sales vs. pizza order? Tasks will be different.

Jeanne: Looking at details and working through examples helps a great deal on how we should be evaluating. I.e. data driven and real examples.
... Talks to https://docs.google.com/document/d/18saMLdtaVlwuVodpZX2-AcIQ8fCvDRxYdZw4VpB2qKo/edit#heading=h.a3c70c4fqun and the testing proposal example 1, newspaper

<Fazio> Not to put the cart before the horse but the text layout is killing me to search through and read through

<JakeAbma> https://docs.google.com/spreadsheets/d/1jNdmR2rBtRBOcexEHXKLwXiz62MhgAI95ktw8sFUcYM/edit#gid=320325119

Jeanne: Continues talking to Test Results. Points to spreadsheet.

<jeanne> https://docs.google.com/spreadsheets/d/1jNdmR2rBtRBOcexEHXKLwXiz62MhgAI95ktw8sFUcYM/edit#gid=320325119

Combined Proposal Scoring

We are looking at tab in spreadsheet that is labeled as Newspaper.

<Fazio> I like the spreadsheet layout so far

Jeanne: I went through on 1 - 3 adjective score ...on a value judgement I assigned ...

JF: Weighting of score and calculation is not present , or how those 7 functional user requirements factor in to scoring piece.

<Lauriat> +1 to the question, since those go into severity. Looking forward to better understanding how things come together.

Jeanne: I didn't include these items in this review. For my example, I wanted to review Functional Outcome of programmatic, visual , expect.

JF: Total instances reduction increase score?

Jeanne: Actually it is the opposite due to weighting of impact (scrolling to the right) table labeled constants

Weight for Essential 10 %

Wight for Different task -30 %

adj 1 is poor 50 %

adj 2 is medium 75 %

adj 3 is good 90 %

<Fazio> I like the idea of separating reports so there isn't an information overload

Jeanne: I wanted to bring pass / fail on par with overall adjective based score.

Looking at the OLD - experiments tab in the spreadsheet the raw socre is 2.65 and weighting was applied simply to come up with a raw score to what adjectival scoring represented.

<Lauriat> Scribe change?

Jeanne: Overall this was a good site for headings, serious mistakes but not a large number of mistakes.

Thanks to all, I need to drop off the call. All valid points and great input!

<sajkaj> scribe: sajkaj

js: Invites people to play with the spread sheet template tab, expirimenting with different values

sl: Suggests do so in a copy! :)

js: Useful conclusion -- "no way to get all"

<Fazio> is the actual weight still up for debate, and you just assumed a number for the example?

js: Conclusion: impportant to have a number items that are important;

jf: How defined essential?

js: On the task that I defined

jf: other headings are valuable, though.

js: Yes, but not essential to the task

sl: So each task would have it's own instance of this kind of template?
... Someone would need to define all the tasks for the site?

js: Would probably need to test fixed categories
... Noting this is to look at the headings guideline--not more

sl: Is this task "scan the headlines?" Or, "locate specific headline?"

js: Latter

sl: Clarifying what goes under essential ...
... For me newspaper assessment would be finding a specific headline
... Believe most people on a newspaper site likely browsing, though perhaps some specific coverage

js: Wonders what principle?

jf: Too granular

sl: How does a user find X
... Scan, visually or otherwise ... is this it? this? etc
... Starts with a decision tree of is this related
... The headlines enroute are therefore still essential

ja: Notes this close to how they currently audit
... We may create some basic ways of interacting, top tasks -- login, nav
... Also content focussed, so structure, headings, color, etc
... Can expose ml that isn't essential, but similarly marked to the essential ones

js: Also looked at top level headlines -- which were improperly ml
... Visually they were fine, but programatically incorrect -- therefore socred 9 and 2 respectively
... Didn't affect the final score enough
... So, this is an issue. A flaw in this approach
... This is why it's important to test the details
... We want the resultant score to correctly reflect how accessibility the site is
... In our case it's a design flaw, not gaming.

<Zakim> Lauriat, you wanted to say that sounds like boiling the ocean, especially once we expand technologies involved, but if we can define these rules at the guideline level, it can

sj: Also doesn't help people do the right thing

sl: the same tech may have different impact based on content type, e.g. newspaper vs XR town

js: Notes weighting is adjustable in this ss; let's expiriment
... Adjust at far right Z, aa
... Being unhappy with the first model, I tried working via impact--but it became too late for good work

jf: Recalling the BB "currencies" model

js: I was trying to go for something integrated
... pass/fail but also adjectival

jf: Don't believe BB ever meant fir pass/fail

js: Was mirroring current testing, then doing more advanced testing a la Rachael
... Wanted to try a merge hoping for more simplicity
... Reiterates scoping has major impact on scoring
... Wanted to show how that can happen
... We need to work on that in task definitions
... Including more waters down the impact

jf: Giving away scoping responsibility facilitates gaming

js: Narrow scope cannot define broad conformance; we can make that a normative no brainer

sl: We have two definitions of scoping
... One scoping a particular task--we're still working on that, and no conclusion yet
... We did agree you can't claim more than you've tested for a collection of tasks
... How do people with different interactions use headings to achieve their task? What might derail someone?
... Notes bad ml in this example could keep some from finding the article
... alt-text in the footer wouldn't interfere, nor a heading in a footer
... Anything autoplaying audio/video would be massively impactful
... Believe we can make this work, but need to explore further and need to define boundaries of critical vs noninterference
... Also testing with another guideloine

js: Agrees

<jeanne> +1 shawn

ja: Notes I tried using two currencies and ended up essentially as Jeanne

<Fazio> high for COGA

<jeanne> +1 DavidF

ja: Also worked on page which was almost perfect and found not always critical with unstyled headings

<Lauriat> +1 to Fazio

<Lauriat> Also potentially high for low-vision, which could result in all of them wrapping and looking like paragraphs.

ja: Think we're coming to improved testing from what we've had before

a: If not 100%, but still much better and more informative, is that OK?

+1 to Jake

<Lauriat> But +1 to the point, even though the evidence didn't quite follow.

jf: Notes severity depends on disabiolity type

sl: Depends on the disability

ja: Agrees severity is granular
... We need at least to document all this
... Believe we're in violent agreement!

<Lauriat> Agreed. :-)

ja: We need the bar somewhere, because we need to account that something may be more severe for one pwd than another
... This would be so much better than what we have now

jf; Offers to write more example pages

<Lauriat> Thanks, JF! Agreed, we all should.

js: Thursday AM conversation is fitting functional needs into our calculus

jf: Notes he may not agree with low-medium-high -- perhaps still needs numeric
... likes 5 better

ja: We can create those docs

jf; agrees


js: Wants to shift topic back to this morning's scissors example ...
... Not for everyone, but for our expirimental examples

<jeanne> https://docs.google.com/document/d/199O_OtKPWqjNX9lfUwztn7MohHcgMDtNJiCsOfe-iaM/edit#

js: Hopes this might pull the entire discussion together
... What goes into socring?
... Had been thinking a matrix might have worked, but not sure and would like to discuss some ...

sl: Goal is the scissors purchase

<CharlesHall> goal = composite of atomic

sl: To reach our goal we have several levels of tasks

js: See so many unnecessary burdens at the beginning -- finding the path through the menus

sl: Without going to the level of press TAB, etc., ...
... From the view of working inward ...
... If a user can't get to their goal, should result in a very poor score
... Should distinguish failure but was close
... But also some kind of language to express how deeply it fails, i.e. for how many users -- functional needs conversation in the AM
... But the failures need to come through clearly
... Opposite also true ...
... If we can demonstrate how each kind of essential need can go through the user journey, score should reflect that
... But middling should show, as well as more easily should be reflected, and these were real good
... our process needs to show why things work, and why they fail

js: Chris' work testing nytimes.com is a good example
... NY Times example would not reveal certain categories because of nesting errors
... We're not requiring correct nesting, but in this instance it became an obstacle

sl: Don't know we can do this guideline by guideline, but overall result is key
... Users must be able to get to the next step, then the next
... So for scissors, if some ml impedes some users achieving the next step, we need to expose that and express it as a failure

js: How do you see this being tested? As ux research with pwds?

sl: Primarily as a series of steps to perform
... Should be possible for pwd with varying AT
... May not be great, but is possible
... Thinking in terms of conformance to this task ...
... Where scores are all over the map, whether by functional need or based on functional step ... the overall failure needs to overpower all the other goodness that may be present
... Let's try functional needs at different granularities tomorrow.
... We have the recurring COGA that when every step is marginal there's little chance of overall success

<Fazio> +1 shawn

<Fazio> we mentioned before tabulating instances for cognitive fatigue

Summary of Action Items

Summary of Resolutions

[End of minutes]

Minutes manually created (not a transcript), formatted by David Booth's scribe.perl version 1.154 (CVS log)
$Date: 2020/05/06 20:01:30 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.154  of Date: 2018/09/25 16:35:56  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: Irssi_ISO8601_Log_Text_Format (score 1.00)

Succeeded: s/e/e?/
Present: Lauriat ChrisLoiselle CharlesHall PeterKorn JakeAbma Fazio bruce_bailey kirkwood Makoto sajkaj Joshue108 JF KimD OmarBonilla Rachael MichaelC KimD_ jeanne AngelaAccessForAll
Found Scribe: ChrisLoiselle
Inferring ScribeNick: ChrisLoiselle
Found Scribe: JakeAbma
Inferring ScribeNick: JakeAbma
Found Scribe: JakeAbma
Inferring ScribeNick: JakeAbma
Found Scribe: ChrisLoiselle
Inferring ScribeNick: ChrisLoiselle
Found Scribe: sajkaj
Inferring ScribeNick: sajkaj
Scribes: ChrisLoiselle, JakeAbma, sajkaj
ScribeNicks: ChrisLoiselle, JakeAbma, sajkaj

WARNING: No date found!  Assuming today.  (Hint: Specify
the W3C IRC log URL, and the date will be determined from that.)
Or specify the date like this:
<dbooth> Date: 12 Sep 2002

People with action items: 

WARNING: Input appears to use implicit continuation lines.
You may need the "-implicitContinuations" option.

WARNING: IRC log location not specified!  (You can ignore this 
warning if you do not want the generated minutes to contain 
a link to the original IRC log.)

[End of scribe.perl diagnostic output]