Comments on WCAG 2.0 Last Call Draft -- AG block 1

Commenter: Al Gilman

Email: Alfred.S.Gilman@IEEE.org

Affiliation:

Date: 9 May 2006

Directions

Column 1: Enter the abbreviation for the Document you are commenting on:
- W2 for WCAG 2.0.
- UW(or HTM) for Understanding WCAG document (and for the How To Meet sections of this doc).
- TD for Techniques document.
Column 2: Enter the major section name or Item number of the guideline, success criterion, technique etc that you are commenting on.
- For example "Intro" for introduction "1.1" for a guideline 1.1, "3.2.1" for success criterion 3.2.1, "H-56" for HTML technique number H-56
Column 3: Enter the heading for the section you are commenting on.
- E.g. "Intent", or "Description", or "Examples", etc.
Column 4: Enter the type of comment you are making
- G = General (i.e., applies to whole document)
- T = Technical (i.e., deals with substance)
- E = Editorial (i.e., concerns grammar, wording, formatting, etc.)
- Q = Question (i.e. not requesting any change).
All : are required.
Filling out column 6 greatly increases our ability to understand and address your concern.

Please ensure that the comments submitted are as complete and "resolvable" as possible. Thank you.

1. Document Abbv. (W2/UW/TD)	2. Item Number (e.g. 1.1)	3. Part of Item (Heading)	4. Comment Type (G/T/E/Q)	5. Comment (Including rationale for any proposed change)	6. Proposed Change (Be specific)
W2	glossary	non-text content	G/T/E	A sympathetic reading suggests what you want to mean, but this text does not say it. As rendered, text is represented by glyphs, not Unicode characters. Unicode characters are integer index values into the catalog of characters. And the critical property is the independence of any structure beyond the stream arrangment of the characters, as demonstrated by the fact that the character sequence is effective in conveying the intended understanding. The sequence of characters in the encoded form is not suffucient. This has been recognized and expressed by the "as rendered" language. But you need to make the cognitive effectiveness aspect of the test more overt. When natural language is written down, it is often encoded in alphabetical writing systems or scripts. Other forms of communication such as mathematical notations and symbolic identifiers have re-used the characters, or bits and pieces of these writing systems, which have been atomized into recombinant elements by the invention of movable type. This issue clearly illustrates the superiority of stating separate requirements on the as-rendered representation of the content and on the as-communicated representation thereof.	Try: "Text content is content which conveys its intended meaning when rendered in a sequence of glyphs recognizable as representing the characters from some writing system. "Content where character sequences are used to form a symbolic code to reconsrtruct media or action scripts, such as the Base64 encoding of a GIF format image, or an ECMASCRIPT imperative instruction set, are to be considered non-text content. Likewise, character sets where the glyphs must be presented in a particular two-dimensional arrangment such as ASCII art are to be considered non-text content." Add separate encoding requirement: Text content is to be conveyed from the author's automation to the user's automation [in accordance with \| in a mannter interoperable with] the Character Model for the World Wide Web (CharMod). Consider lifting language from the IMS accessibility metadata documents. IIRC that is where I got this concept.
W2	1.1.1	1st and 2nd list items		It is hard to see how the case handled by the second sentence under the first bullet doesn't include the case stated in the second bullet. The only difference is that in the second bullet the information is asked for in a label, and in the first case in a "text equivalent." In either case the actual requirement is that this minimal information about the non-text context item is required to be available in a way that the user knows it is about the non-text item. The fact that information in a label need not be replicated in a substitutable text object is true here and is a vest-pocket example of equivalent facilitiation.	merge two items under the pattern of "human-understandable text explanation associated with the non-text object by a programmatically determined (i.e. machine recognizable) association."
W2	1.1.1	1st bullet	T	There is too great a leap, here. One either has to both present the same information and fulfil the same purpose or one only has to provide a nominative description of the purpose in text. There is an important middle ground.	Make it more like the priorities in WCAG1: The strongest requirement is that the text alternative accomplish the purpose of the non-text content it is an alternative for. The desired way to do this is to present the same information, but that lower-level agreement betweeen the alternatives is at a lower level of criticality. Even this is not the right requirement because there needs to be recognition that the content can succeed by affording equivalent facilitation i.e. a go-path that succeeds for this user at a higher level of aggregation.
W2	1.1.1	1st bullet vs. 3rd bullet	T	CAPTCHAs, displays that match the "if" clause of the third bullet, can claim conformance under the first bullet by simply saying in the ALT text "image to test if you are a human." It is not just nor do I believe it is the authors' intent to open up this loophole.	Close loophole by repairing the first bullet along the lines of the previous comment.
W2	1.1.1	4th bullet	T	The non-text content must be implemented such that it can be ignored anyway, even if the text equivalent provides full equivalent facilitation. You can't have the video frame-change events capturing the AT's attention, etc. The requirement stated here applies all the time, not only for pure decoration.	Break out into separate requirement on the "as communicated" representation of the content, a.k.a. the "data on the wire."
W2	1.1.1	logic	E/T	The 'if' clauses don't belong in this construction. I suspect what you had in mind was a cascade of elseIf clauses.	Change from OR (a.k.a. one of the following is true) of if A1 then B1 if A2 then B2 if A3 then B3 etc. to OR of A1 and B1 A2 and B2 A3 and B3 etc.
W2	3.2.5	effect, standard set, SC level	T	This success criterion delivers less in the user experience than UAAG 1.0, checkpoint 3.1. UAAG makes this subject to user profiling. Single-switch users, for example, rely on context changes that are animated by the system, not triggered one by one by the user. Low-vision users will come down on different sides of this question depending on how much of the content they can see at once and how much of the session structure they can hold in their head.	Best Make equivalent facilitation (now 4.2) a principle. Include user configuration of the user experience under one of the forms of alternative recognized. State user experience requirements separately; define these by reference to UAAG 1.0. State data-on-the-wire requrements separately. These have two options: turnkey format -- player meets UAAG requirements directly. open format -- format populates machinable content model (c.f. rewritten 4.1) with information and actions that let UA manage and provide this capability.
W2	3.2.5	principle	G/T	This requirement is mis-filed in the current outline. This is a control issue, a matter of keeping actions under the user's command. If it were an orientation issue (principle 3) one could repair by announcing context changes. That is not enough. In the current outline it belongs with Principle 2.	re-flow this requirement under what the user can do, not what they can understand.
W2	3.2.2	obstacle: "beyond moving to the next field in tab order"	T	Exception for moving to next in tab order is not in agreement with customary GUI behavior that the user is accustomed to and will expect.	The baseline should be customary GUI behavior, not next in TAB order.
W2	3.2.2	remedy: "the authored unit contains instructions before the control that describe the behavior"	T	Requiring the explanation to be inline before the control is overly restrictive. If and where the explanation gets rendered is appropriately in the domain of the user's view and verbosity controls. Also "authored unit contains" is too narrow. What matters is that the server delivers the material to the browser.	Replace "authored unit contains instructions before the control that describe the behavior" with "the delivered information about the control includes a human-understandable explanation of the behavior that is machine-understandable to be explanatory regarding this control." Align with requirements in 2.4.4 for links that there be human-understandable stuff bearing required information that is associated with the link or other object by an association that is understandable by the user's automation.

<a name="note1" id="note1">Note 1<a>: