2.4.11: Adversarial Conformance

By admin

2.4.11
CARDINAL

: Adversarial Conformance

Updated

November 2, 2023
DATE

, originally posted

October 16, 2023
DATE

;

3
CARDINAL

Comments

This post is part of

RSS Club
ORG

, rewarding those who still use RSS to read and/or share content. These posts are embargoed from my regular post feed and the socials for an arbitrary number of

weeks
DATE

. You can see all the

RSS
ORG

-only posts at

AdrianRoselli.com/category/RSS
ORG

. Tell your friends (to get an

RSS
ORG

reader).

This post is part of

RSS Club
ORG

, rewarding those who still use RSS to read and/or share content. These posts are embargoed from my regular post feed and the socials for an arbitrary number of

weeks
DATE

. You can see all the

RSS
ORG

-only posts at

AdrianRoselli .com
PRODUCT

/category/

RSS
ORG

. Tell your friends (to get an

RSS
ORG

reader).

I made a demo for

WCAG 2.2 Success Criterion
ORG


2.4.11
CARDINAL

Focus Not Obscured (Minimum). I built it strictly to capture screenshots, so while it is not a good demo as a standalone (which is why it is not embedded here) it is handy to demonstrate something.

Conformance


Three
CARDINAL

of the following screenshot examples pass the

SC
ORG

but

only one
CARDINAL

is useful to users. It comes down to this

SC
GPE

language (emphasis mine):

When a user interface component receives keyboard focus, the component is not entirely hidden due to author-created content. WCAG

2.2
CARDINAL

Success Criterion

2.4.11
CARDINAL

Focus Not Obscured (Minimum)

User interface component is defined by

WCAG
ORG

as a part of the content that is perceived by users as a single control for a distinct function . I argue each button in my demo is a single control.

This fails the

SC
ORG

because the button is hidden, even though its focus style is visible just above the fixed footer.

This passes the

SC
GPE

because the topmost row of pixels from each button is visible. Note that there is not enough information for a sighted user to know what the buttons do.

This also passes the

SC
GPE

. Now we can see each button has text, but because we only see the very top of the letters we have no idea what they say.

This passes the

SC
GPE

and has the benefit of being legible to a sighted user.


Two
CARDINAL

of the

three
CARDINAL

passing examples are terrible. That last passing example is not much better, however. Throw your favorite 1.4.12 Text Spacing bookmarklet at my

2.4.11
CARDINAL

demo and watch all the buttons disappear.

Adversarial

Since

two
CARDINAL

of these buttons could result in doom for

Dave
PERSON

, it stands to reason that HAL designed this interface — creating a de facto anti-pattern to potentially drive a particular outcome. Or maybe just for the lulz. HAL had an interesting sense of humor.

This is something I call adversarial conformance. HAL the

UI
ORG

designer can claim that it followed WCAG even though it’s clear that HAL expended effort on minimally conforming while negative outcomes are seemingly obvious.

That extra effort is the tell, even if it is not always obvious HAL had to do more work. Heck, sometimes HAL would not recognize it did more work just to avoid implementing what it felt was an ugly design decision.

I have experienced these kinds of efforts

over many years
DATE

of consulting and I figured I would start documenting them.

In the very specific case of HAL the

UI
ORG

designer, its effort is clearly malicious conformance. That is going a step further than I am outlining in this post (since trying to kill the user is generally not a goal), but it is still a genuine concern.

WCAG issue #3529 Clarification for Understanding

2.4.11
CARDINAL

Focus Not Obscured (Minimum) to ensure consistent testing, filed

today
DATE

, seeks to confirm that the focus indicator is not part of the control.

In

July
DATE

, co-editor of

WCAG Alastair Campbell
ORG

noted that including the focus style in the component size would have messed up the calculation for

SC
ORG


2.4.13
CARDINAL

:

Focus Appearance
PERSON

(creating a loop), so the focus style was never meant to be included in the definition of the component:

What is perceived as the user interface component or sub-component (to determine enclosure or size) depends on its visual presentation. The visual presentation includes the component’s visible content, border, and component-specific background. It does not include shadow and glow effects outside the component’s content, background, or border.

Success Criterion
ORG


2.4.13
CARDINAL

Focus Appearance, Note

1
CARDINAL

If the focus style is considered part of the control, then my

first
ORDINAL

example would pass and still be useless to most users.

Probably a good idea to watch for the outcome.

Tags accessibility, rant, standards, WCAG Other Posts