NZ’s draft Digital Accessibility Standard: the good, the bad, and the fugly
A critical look at the draft Digital Accessibility Standard, proposed to be implemented by the NZ Government Digital Delivery Agency.
Table of contents
- Introduction
- The good
- The bad
- The fugly
- It doesn’t apply to health
- It removes all accessibility requirements for 1 year
- It reduces the number of mandated organisations
- The Standard contradicts itself on PDFs
- “Must” versus “shall” confusion nightmares
- The definition of “web page” is wrong
- We’re consulted on a non-existent standard
- Accessibility statements are not machine-testable
- The timelines are far too slow
- NZ Govt Procurement Rules will be out of whack
Introduction
The NZ Government released a draft Digital Accessibility Standard (“Standard”) for consultation. This post will give my view on it.
I’ve previously written about two parts of the proposal:
- A digital accessibility standard that (maybe?) forgot the health system
- EN 301 549: the accessibility standard that isn’t accessible
The draft standard effectively just mandates EN 301 549, a European standard, which adapts and supplements WCAG to cover a broad range of ICT, including physical devices, kiosks, the web, non-web documents (i.e. Word docs), and software such as phone apps. It even extends to two-way video/audio communication systems.
Anyway, let’s start critiquing the draft Standard.
The good
Good: extending beyond just web pages
The new Standard improves upon the old Web Accessibility Standard that applied to government agencies in one central way: it applies beyond the web to non-web documents, software, and hardware. That’s a huge and important positive change.
Good: audio description in video is mandatory
Audio description in videos helps blind and low vision people understand what is going on in the video by embedding narration. Platforms like Netflix and Apple TV have done this for several years, and the NZ Govt has been slow to follow suit. This has been my biggest beef with the Web Accessibility Standard for a while: audio description was never a requirement. The latest update to the Web Accessibility Standard did inch towards it by making audio description a should requirement (and a must for high-stakes information), but it shied away from making it a must requirement in all circumstances.
The Web Accessibility Standard intentionally weakened WCAG, which wasn’t a good idea. I’ve audio-described videos I produced myself and it ain’t exactly rocket science. If a (legally) blind man like me can edit videos proficiently, then surely professional government video contractors can handle it too. You’d hope.
EN 301 549 is clear: WCAG Success Criterion 1.2.5 Audio Description (Prerecorded) is a shall requirement. Nice.
Good: live captions are mandatory
Again, the old Web Accessibility Standard made live captioning a should (for non-high-stakes info) rather than a must. The draft Standard fixes this.
Good: it makes plain language a requirement
I was a staunch advocate against the repeal of the Plain Language Act — it even led to one of my greatest achievements — finally being mentioned in a Wikipedia article! 😀
I am genuinely very happy to see the plain language requirement in this draft Standard. It’s the perfect way to ensure it remains an important part of running government.
Accessibility standards often forget or lack focus on people with learning disabilities. This is primarily why I advocate for plain language.
Good: accessibility statements
I think it’s super cool to require accessibility statements on ICT. This will ensure all agencies have to at least think, for a bit, about how they make their ICT accessible.
Good: HTML-first publishing
The Standard makes HTML documents a “should” requirement. Given PDF is a cursed format that needs to meet the same demise as Adobe Flash, this was excellent to see.
Adobe Flash and PDF have a lot in common. They have both been a scourge on accessibility for disabled people, and yet government agencies keep publishing in PDF. We need strong leadership to put this format out of its misery.
We need the Steve Jobs “Thoughts on Flash” moment for PDF.
My one suggestion is to add a clear negative requirement against PDF, and direct towards better formats. For example, “PDF should not be the only available format of information.” Agencies should be considering how they can adapt their publishing process to allow HTML-first processes. I run a fairly complex website and I never need to reach for PDF, it’s unnecessary.
The bad
Bad: descriptive text transcripts no longer required
As WCAG SC 1.2.3 Audio Description or Media Alternative (Prerecorded) can be met through either audio description or a transcript, a transcript is no longer required to meet the Standard.
The WCAG SC that requires transcripts is Level AAA, which is WCAG SC 1.2.8 Media Alternative (Prerecorded). An easy way to fix this problem is to simply specify that SC 1.2.8 is required for time-based media.
This could harm people who rely on transcripts — particularly the deafblind community.
It will also impact AI, as it will have no way to easily ingest video content as text, which could result in lower quality AI responses about government services. Won’t somebody think of the robots?! 🤖
Bad: “only parts that change” contradicts EN 301 549
The Standard applies immediately to “only the parts that change” for pre-existing ICT.
EN 301 549 uses WCAG’s conformance model, which includes Full Pages and Complete Processes. If only a part of a page is changed, or a part of a process, how does this reconcile with the draft Standard’s wording?
The Standard must operate at the same level of “resolution” as WCAG’s conformance model.
Bad: the Standard should not contain the mandate
I am thinking about the Standard’s future potential application to Crown agents, and even private businesses. Or, the potential use of the Accessibility Standards Bill (if it passes).
If the Standard includes the range of agencies it applies to within its normative text, it makes the Standard useless as a mechanism that can be later applied to other organisations (without publishing separate versions).
I think the “mandate” — i.e. the mechanism used to make it mandatory and the scope of organisations it applies to — should be stored outside the Standard’s text itself. It could include a non-normative note to signal the agencies the standard currently applies to, with links to a separate page that demonstrates the mandate(s) in effect.
This will then allow, for instance, a Crown Entities Act directive to expand the Standard’s mandate to Crown agents at a later date. Or, the Accessibility Standards Act could use the Digital Accessibility Standard and apply it to a range of organisations outside Public Service Act s. 10(a) entities.
Bad: it applies nonsense requirements to government agencies
The “Core accessibility requirement” feels woefully undercooked to me.
It basically just applies the entirety of EN 301 549 to every agency.
EN 301 549 contains clauses that don’t really make sense for a government agency to comply with. Some clauses are intended for suppliers who build and sell ICT.
For example, requirements like EN 301 549’s clause about “Installation instructions” are not really for the buyers of ICT; they’re for the suppliers.
This will require a careful consideration of the intended audience of each of EN 301 549’s clauses to ensure they apply appropriately to agencies, in the right ways. I suspect the EU Web Accessibility Directive has probably already figured this one out, through things like Annex ZA.
The fugly
Fugly: it doesn’t apply to health
I’ve already written extensively about this: A digital accessibility standard that (maybe?) forgot the health system.
Basically, because the draft Standard only applies to Public Service Act s. 10(a) entities, it does not apply to Crown agents like Health NZ, ACC, Kāinga Ora, etc.
This is a huge problem and needs to be resolved.
Fugly: it removes all accessibility requirements for 1 year
The Standard currently replaces the Web Accessibility Standard on its start date.
The Standard specifies that pre-existing websites must meet the new Standard after 1 year.
This means, logically, there is a 1 year window where govt websites have literally zero accessibility requirements upon them (except for new modifications).
This is caused by the draft Standard lacking an appropriate savings clause, which helps to ensure accessibility requirements from the Web Accessibility Standard are contiguously joined up with the new Digital Accessibility Standard’s effective dates.
Fugly: it reduces the number of mandated organisations
The list of organisations that the Standard applies to is extremely important.
The old Web Accessibility Standard applies to every Public Service department, NZ Police, NZ Defence Force, Parliamentary Counsel Office, and the NZ Security Intelligence Service.
The draft Digital Accessibility Standard only applies to Public Service Act s. 10(a) entities, so essential government agencies like the police, and the Parliamentary Counsel Office (who publish our laws!) will no longer have an accessibility requirement from a central standard. The NZSIS survives, though — it’s a Public Service Act s. 10(a) entity — accessible spies, yay! 🕵️♂️
Ensuring the publisher of our laws (Parliamentary Counsel Office) has an accessibility requirement is incredibly important; it’s an assault on basic democratic rights to remove accessibility requirements on such an organisation.
A State cannot, in good conscience, remove accessibility requirements from the publication of our laws.
Fugly: the Standard contradicts itself on PDFs
The Standard proclaims that HTML-first publishing is a “should” requirement. That’s wonderful. I agree with that.
Where it becomes a disaster is the Standard’s core accessibility requirement: EN 301 549. EN 301 549, as a document, is an unmitigated dumpster fire. It’s a non-accessible PDF.
It’s logically ridiculous to base an entire accessibility standard on PDF. It excludes the very population it aims to serve.
I have written in-depth about this problem: EN 301 549: the accessibility standard that isn’t accessible.
Fugly: “must” versus “shall” confusion nightmares
The Standard uses words like “must” and “should” to indicate different strengths of requirements.
But EN 301 549 uses words like “shall” and “should”.
The “shall” (in EN 301 549) and “must” (in DAS) seem roughly equivalent; however, “should” (in EN 301 549) and “should” (in DAS) are defined in very different ways.
This will lead to confusion for implementers of the Standard.
Fugly: the definition of “web page” is wrong
The draft Standard claims to use EN 301 549’s definition of “web page”.
However, it actually uses the Web Accessibility Standard 1.2’s definition.
Fugly: we’re consulted on a non-existent standard
The draft Standard cites EN 301 549 version 4.1.1.
Version 4.1.1 does not exist. We have not been provided with a link to an appropriate resource to review the intended document that builds the vast majority of the Standard’s requirements.
Fugly: accessibility statements are not machine-testable
Since accessibility statements are now a requirement, it would make sense if the standard specified basic requirements to ensure their existence and contents can be automatically monitored via a system like the Centralised Web Accessibility Checker (CWAC).
For example, you could specify that a standard path like /accessibility-statement/ (for websites) always leads to the statement itself.
Mandatory headings within the statement could be specified to ensure each statement covers the right areas. One important example might be a “last updated” heading, so that the staleness of statements can be monitored.
Without a testable mechanism, the statements will not be broadly successful.
Fugly: the timelines are far too slow
The Standard applies:
- after 1 year for websites and non-web documents
- after 3 years for non-web software
- after 5 years for all other ICT Products and Services, including hardware.
We’ll be waiting well into the 2030s before these requirements kick in.
This is ridiculously, unacceptably slow.
Agencies have already demonstrated they are comfortable with not complying with pre-existing accessibility mandates. Just look at the CWAC data. They’ve had 20+ years and are still broadly failing to pass even automated checks.
We cannot afford intentional delays in accessibility requirements. Disabled people deserve an accessible society today. Not in 5+ years. Delaying it like this is fundamentally unethical, in my view.
So my philosophical point is: what is the point of delaying it? What does that actually achieve, besides slowing things down? Who does it serve?
The Human Rights Act already makes disability discrimination unlawful. This stuff is already literally illegal today. The Standard just clarifies how to give effect to human rights which already exist.
Agencies already fail current standards. Therefore, intentionally manufactured delays in this Standard are totally pointless and achieve nothing except delaying the realisation of rights for disabled people.
Fugly: NZ Govt Procurement Rules will be out of whack
The NZ Govt Procurement Rules would be the obvious place to apply EN 301 549’s provisions that apply to suppliers. This is how you ensure it’s in the actual contracts. Currently there is no publicly available evidence that the procurement rules are being considered in tandem with the Digital Accessibility Standard.
If the Digital Accessibility Standard is written to be agency-agnostic (i.e. remove the mandated agencies section and put it in a separate doc), the Procurement Rules could be used to apply EN 301 549’s provisions to Crown agents, among other groups.
— Callum