The Section 508 Refresh Takes a Far Too Expansive View of WCAG 2.0
Tuesday, March 6, 2012 Return to Logical Structures
While applauding the approach taken by the U.S. Access Board in choosing to reference established technical accessibility standards in their December 2011 ANPRM, I feel compelled to register my deep concern regarding the implementation of these regulations in the PDF context, and likely other non-web contexts as well.
I encourage other members of electronic content industries (all of whom are affected, it seems) to make their own voices be heard by filing a comment at the regulations.gov website. My own comment, complete with proposed new wording for the regulations, will go in before the close of the comment period on March 7, 2012. If you agree with me (and if you don't), please feel free to leave a (preferably signed) comment on this page.
What's the problem, you ask? The following paragraph occurs in the Access Board's December 2011 ANPRM (Advance Notice of Proposed Rule Making) on the Section 508 Refresh in Advisory E205.1 and Advisory C203.1. (link to the ANPRM):
"WCAG is written to be technology neutral. While oriented towards web pages which are defined as being delivered using HTTP, it is straightforward to apply the WCAG 2.0 Success Criteria and Conformance Requirements to all electronic content." (emphasis added)
I have two main observations.
- Minimally, the Access Board's alteration of WCAG 2.0's scope from "web content" to “all electronic content” is a dramatic expansion of WCAG 2.0's stated purpose.
- Who is the U.S. Access Board (or W3C, for that matter) to determine that applying WCAG 2.0 to “all electronic content” is "straightforward", and how did they arrive at this astonishing conclusion?
There's little question that WCAG 2.0 was not written with the intent of addressing "all electronic content" - the name of the standard itself is a bit of a giveaway. The phrase (and notion) "web page" suffuses the document. Had the authors actually intended WCAG 2.0 to cover all electronic content it's safe to say that the Guidelines would have received very different inputs from the Committee's more technically inclined participants.
WCAG 2.0's four Principles are indeed probably general and valid for all electronic content, but if WCAG 2.0 had actually been designed with "all content" in mind it would be, quite frankly, a completely different document, certainly at the Success Criteria level. Who knows? That's not the road W3C took when it developed the Guidelines.
By stating that WCAG 2.0 is straightforward to apply to "all electronic content", the Access Board has made an extremely general statement covering a wide range of technologies. Perhaps such application is straightforward for some, but is it really “straightforward” to apply WCAG 2.0 to PDF without additional normative language?
The history of PDF demonstrates that electronic content accessibility, even in the light of WCAG 2.0, is not straightforward at all. WCAG 2.0 is written to be technically neutral for web content, so it should be no surprise that in some cases (PDF being one such), a technology-specific standard is vital to achieving consistent accessible outcomes.
The PDF Reference 1.0, the document detailing the technology of PDF, predated popular awareness of the web when Adobe Systems published it in 1993. While today's PDF 1.7 (now ISO 32000-1) details the mechanisms of PDF accessibility, the PDF Reference itself does not require accessibility features because PDF has many legitimate use-cases, printing systems, for example in which accessibility is irrelevant. The distinction is crucial.
The accessibility mechanism in PDF – the system of marked content and tags – is almost unrelated to the appearance of the PDF on screen and in-print. From an accessibility perspective there are advantages and disadvantages to this fact, but from a technical perspective, the extraordinary variety of ISO 32000-legitimate use-cases means developers need additional specific technical norms to achieve accessibility objectives in a consistent fashion.
Consistency, a basis in “hardcopy” pages, flexibility and self-containment for portability and archiving are core features of PDF and irreducible qualities of the PDF experience.
The last twelve years have proven that accessibility models not structured around these core features are never going to result consistent universal access to PDF content based on a common understanding of what’s required when making and processing PDF files.
I've seen no evidence that implementing ISO 32000-1:2008 in light of WCAG 2.0 alone will result in consistent test or real-world experience for AT users.
In many cases, the PDF Reference (ISO 32000) provides inadequate technical detail to enable developers to generate consistent results when testing for WCAG 2.0 conformance. The observation that PDF developers needed specific normative language in order to reliably test and resolve accessibility technology in PDF was the rationale behind ISO 14289 in the first instance.
The most obvious of the problems is this: PDF developers may select from a variety of technical means when repurposing content. Some use PDF tags for logical ordering, and some do not, with a variety of methods in-between. This failure to agree on the priority and significance of logical vs. content order for accessibility purposes is the root cause of the general perception that PDF accessibility is arcane. The situation remains confused in leading applications to this day.
A notably unfortunate example of this confusion is W3C’s own 2008 Accessibility Support Documentation for PDF, which attempts to map WCAG 2.0 Success Criterion 1.3.2 to ISO 32000-1:2008, Section 18.104.22.168. Unfortunately, this section deals with “page content order”, not “logical order”, the proper subject of Success Criterion 1.3.2.
As the PDF Reference puts it in the first paragraph of this very section:
“The two orderings [logical and page content] are logically distinct and may or may not coincide.” (ISO 32000-1:2008 Section 22.214.171.124.1, paragraph 1)
If PDF is complex or ambiguous enough that even W3C's own report applies Success Criterion 1.3.2 (Level A) incorrectly, how "straightforward" can it be?
PDF/UA solves the problem with normative text stating that tag (logical) order is the organizing model for accessibility, thus setting the stage for consistent tools and end-user experience.
PDF/UA is also necessary simply because the PDF Reference provides no meaningful priority guidance on accessibility features. All aspects of tags in ISO 32000-1:2008 are “may” and “should” rather than “shall” statements. PDF/UA, by contrast, uses stronger “shall” and “should” statements in place of the weaker terms of ISO 32000-1. But that's a subject for another post!
During my testimony to the U.S. Access Board on March 1, I urged the US Access Board to include ISO 14289-1 (PDF/UA) as a referenced Standard in the refresh to Section 508 and wherever standards are set for accessibility in electronic documents.
PDF technology serves vital functions throughout the economy with specific features unlike those characteristically associated of web technologies. The Access Board is right to refer to industry standards in its rule making, but it should operate with caution when applying web content standards to all electronic content.
In the case of technologies such as PDF, the regulations should reference applicable ISO or other normative standards if available. For PDF, that means ISO 14289 (PDF/UA).
In the testimony I submitted to the Access Board I provide technical details and examples.