What are people doing with H5P and new DOJ regulations?

Dear Community:

Apologies for cross-listing. I have a quandary that has been plaguing me the past month that I was hoping to get feedback from the community about.

As many (hopefully most/all) of you know, the new Dept of Justice accessibility guidelines are coming down the pike and set for next Spring. Briefly, these guidelines argue that academia should have all their active content set up to WCAG 2.1 accessibility standards with some added penalties if not. In the time of increased federal scrutiny over academic activities, this is not something that should be ignored. Judith Sebesta gave a nice webinar of this last month for OEG/CCCOER (New Accessibility Requirements for Web Content and Mobile Applications for Public Entities – CCCOER).

While this has implications for many different aspects of edtech (including anyone creating, distributing, and adoption OER), my immediate question is about H5P interactive assessments. While popular, the vast majority of h5p interactives are not up to WCAG 2.1. While some of the assessments can be with proper workarounds (see our accessibility guide for details - H5P Accessibility Guide | LibreStudio), most creators don’t create WCAG-compliant assessments. Furthermore, many H5P assessments cannot be made WCAG-compliant.

We cannot ignore this issue since campuses are NOT going to encourage or even all the adoption of OER with potential legal penalties from the DOJ. Given the ubiquity of H5P assessments in OER, what are people thinking about doing to address this issue?

These are ideas floating in my head:

  1. Excise all h5p assessments from OER to avoid the issue - use a different technology.
  2. Build a converter to convert the H5P assessments into better tech that can be made accessible with aria and other coding capabilities (e.g., QTI in LMS or other tech like our ADAPT homework system). This would take a bit of effort to do and if QTI is the protocol, then not all problems are possible to be converted since it has limited capabilities.
  3. Fork the H5P tech into something separate with control, but then operate outside of the H5P network for updating question types.
  4. Work with D2L, the new owner of H5P, to bring those questions that are possible to WCAG-compliance
  5. Build an alternative system in the H5P presentation system that provides students with an WCAG-compliant alternative (we do this with ADAPT). This would be painful to implement too.

There are benefits and detraction for each of the options above and none are simple and painless to implement. What is the community thinking about in regards to this specific question. I appreciate any/all feedback. We don’t have a lot of time to work around this requirement (Spring of '26).

Regards,

Delmar

I find it very strange that THIS government is taking this stance on accessibility.

Heather

Heather M. Ross, BA BEd MEd

Educational Development Specialist

University of Saskatchewan

Gwenna Moss Centre for Teaching and Learning

Ph: 306-966-5327

Find open textbooks and other open educational resources on:

http://open.usask.ca

I don’t know the full history of the regulation, but it was crafted before the Trump administration started. And while one may not expect this administration to pursue an equity/universal design mandate, the option for giving the federal government a new cudgel to attack “unruly” campuses may be attractive the administration.

1 Like

Thank you, Delmar, for bringing this issue to our attention. You have made some excellent suggestions.
Delmar’s take on the fact this was started way before the current administration is accurate. It is also dangerous for any education institution to try to predict what this administration will or won’t do.
And, while we’re on the topic of making resources easily available to students, how about making them available to students who don’t have internet access?

Delmar:

I think you are right to anticipate that accessibility could be weaponized in the ongoing attempts to delegitimate and dismantle education. Of course I am a huge supporter of accessibility, but don’t want to see it used to further that misguided agenda.

Of your options I’m thinking 4 seems like the best first move. I know some people at D2L and could reach out if that seems helpful.

True on all fronts. The aspect of offline access has been something dear to our (LibreTexts) collective hearts for years. We even had proposals out for expanding both textbook and homework offline in the past (unfunded). Of course wtih accessibility in mind for both (to keep with this thread’s theme). We should discuss this as a different time or perhaps have a panel/webinar on the topic.

Nate:

Yea. As Dan mentioned, we cannot predict with any accuracy what the federal government is going to do. But, preparing for the worst is a sound approach. I would love to touch base with D2L about their plans (if any) on this important topic. Feel free to reach out to be off-forum with any contact details you are willing to share.

While clearly important for the US, the existing accessibility rules/laws north of the border should have prompted this discussion years ago. So I am perplexed why so much momentum exists in H5P integration without a clear approach to handle its poor accessibility capabilities. Eitherway, we have to deal with it now…

Thanks for your feedback.

Regards,
Delmar

1 Like

I didn’t see a message from Dan, so perhaps that was off list. I now see Dan’s message here in the web interface. Not sure why it doesn’t appear via email.

I actually think it is possible to predict what this federal government will do — the only question is how it will be worse, faster, and more than one’s prediction.

I will reach out to my D2L contacts in the spirit of addressing the need collaboratively rather than assigning blame for past inaction.

1 Like

Are the LibreText findings up-to-date? I know that H5P Group do rounds of accessibility improvements now and then, for instance (cmp. e. g. their latest release note). Maybe the picture is not as dire as it looks like based on what I read in this post? And H5P Group talk about WCAG 2.1 conformity themselves, and I’d assume that it’s in their own interest to keep that promise. So I assume they’d be open for discussion.

But it is not only D2L/H5P Group that you’d need to talk with, but other content type contributors, too. And there are some that have “retired”, leaving their code out in the open - just waiting to be adopted and improved by someone else (Find Multiple Hotspots, Find the Words, Image Sequencing, Image Pairs, Advanced Fill in the Blanks, Personality Quiz at least).

And have you thought about pooling resources? There’s a German research group dealing with H5P accessibility. NDLA in Norway also is highly invested in this topic.

1 Like

Thanks for a juicy topic, Delmar. I recommend watching the recording of the CCCOER webinar by @JudithSebesta it was sobering and informative together.

The DOJ regulation is dated April 2024 but worth noting that the 2026 deadline is for institutions larger than 50k students; small organizations have a 3 year time frame (direct reference link).

Option 1 seems overly aggressive, especially since the LibreText guidelines indicate 10% of the H5P content types meet WCAG and 50% will with “workarounds”- which are well summarized on the LibreTexts resource..

Kudos to the Academic Senate for California Community Colleges (ASCCC) that started critically analyzing accessibility in 2020.

It does seem most sensible to work with, apply some pressure on D2L (and as Oliver notes, creators of other content types) to address these issues. That its not been done although raised for quite a few years leads one to speculate that it’s extremely complex.

And I bet its even a challenge to get summaries of all H5P content types used in an OER or across an institution, unless housed in something like LibreStudios. For Pressbooks, there is my favorite secret trick- just add /h5p-listing to any Pressbook title to get at least a browsable view of all included H5P.

e.g. you see a lovely title like Biology and The Citizen https://uen.pressbooks.pub/biology1010revision/ and just add to the end H5P listing – Biology and the Citizen

I recall I did submit a Pressbooks suggestion to perhaps generate some kind of report on the kind and count of H5P content in a title. I have seen some data shared a bit back from Steel Wagstaff that listed a frequency count of the most used H5P content types across Pressbooks Directory (sharing a static copy from May 2024)

There is some complexity when you consider the different between single purpose content types (fill in the blanks, drag text, and the compound content types like Course Presentation, Interactive Video, Branching Scenario that include other content types within.

Lastly, I worked on a small project with BCcampus and Pressbooks aimed at outputting printable versions of H5P in PDF exports for the situations where learners use print offline (the URL replacement is of no value).

Oi, one more- yes the accessibility features is one thing, but I have seen in a lot of H5P content that creators do not often use or enter metadata.

2 Likes

Thanks for the shout out, Alan! I want to point out that the deadline for compliance is contingent on the total population the entity serves, NOT the institutional enrollment. Very few public colleges in actuality serve a population of less than 50,000 – that would be a very small community college district. And the additional time for compliance would not apply to any public state universities, since there are no states with populations less than 50,000.

2 Likes

Thanks for catching that, Judith.

So what can institutions expect? Is there are process or timeline that DOJ is going to be reviewing? And how will the DOJ assess compliance? Or is it that’s when they will take complaints to investigate?

1 Like

Thanks, Alan, for acknowledging a lack of internet access as an Accessibility issue. Too often, people without internet access get treated as if it’s their fault that they’re not connected 24/7. There’s also the issue of those individuals and groups and settings where being connected is not the best idea. I think security issues will be making that group increasingly larger.
If your classroom is not connected to the internet, it becomes easier to know what’s AI produced and what’s not. There might be lots of teachers who would like more control over the content available.

In my analysis of the regulation text, beyond the deadlines for compliance, there is no stated timeline for review. Enforcement apparently will occur, as you mention, primarily through complaints and lawsuits alleging non-compliance, not DOJ actions. On a positive note, I think it is important to view this as a move from being reactive to being proactive when it comes to accessibility. Instead of primarily responding to individual students’ accommodation requests, colleges and universities (and other public entities) will need to make content accessible from the outset.

Alan:

There is a lot of digest in your reply. I will try to be succinct (although rarely sucessful).

The DOJ regulation is dated April 2024 but worth noting that the 2026 deadline is for institutions larger than 50k students; small organizations have a 3 year time frame (direct reference link).

Judith addressed this below. There is both an ethical and, now, legal rational to address this. The fun and utility of h5p has blinded many to its accessibily liabilities.

Option 1 seems overly aggressive, especially since the LibreText guidelines indicate 10% of the H5P content types meet WCAG and 50% will with “workarounds”- which are well summarized on the LibreTexts resource..

It was a base option to consider. To be honest, this is the recommendation I give to LibreTexts authors wanting to make interactive texts. There is too much trouble with h5p - poor accessibility and no security are the biggest. I feel that investing so significantly in h5p will cause a massive headache in the near future (or financial loss) and probably reduced OER adoptions.

To clarify, the ratings we gave are more an indication of how much effort was needed to make WCAG compliant assessment than if they are or are not accessibile. Nicholas (our A11y master) was concerned that people would construe the green-labeled assessments as accessbile “out of the box” and that is incorrect. While many of the assessments can be made accessible, most authors don’t make accessibile h5p interactives and many of the “cool” h5p interactives that people really like cannot be made accessible (an alternative is needed - how we handle it in ADAPT).

Kudos to the Academic Senate for California Community Colleges (ASCCC) that started critically analyzing accessibility in 2020.

Yes. They were the first people to point out to me how flawed H5P was. We (LibreTexts) used the same A11Y evaluator; Nicholas really rocks.

It does seem most sensible to work with, apply some pressure on D2L (and as Oliver notes, creators of other content types) to address these issues. That its not been done although raised for quite a few years leads one to speculate that it’s extremely complex.

I am unsure how complex it would be. The decentralized approach to making/distributing/embedding h5p question types makes it a complicated process (although D2L may have plans to address this) but in the end of that it is just a HTML5 container (JS,HTML,CSS) so it isn’t rocket science to address. It does take effort and the first step in that effort is to recognize this is needed - hence my desire to kindle a discussion.

And I bet its even a challenge to get summaries of all H5P content types used in an OER or across an institution, unless housed in something like LibreStudios. For Pressbooks, there is my favorite secret trick- just add /h5p-listing to any Pressbook title to get at least a browsable view of all included H5P.

Well, there are benefits to a centralized approach like LibreTexts and eCampus Onterio vs. a decetranlized approach like Pressbooks or LMSs. If the solution to the problem is fixing the question technology and not the questions themselves, then either approaches can be updated without much effort so existing question can operate in an a11y manner. I don’t think this will be case though. The solution will likley require re-editing/re-building questions individually (perhaps with the assistance of AI) and a centralized approach will be 100x faster/better/easier to address since in decentralized approaches each cloned text/course has a copy of the embedded h5p containers.

This may be a good case study about the benefits and limitations of decentralized approaches vs. centralized approaches. But, that is a different conversation for a different day.

Thanks Delmar, this is exactly the kind of talk I like to see happening here! And what is “succinct”?

I’m no H5P advocate or zealot, just had some experience on a project looking at using it for pressbook activities now a few years back with BCcampus. Thus I am also far from the current front lines. But I support what Judith urges to be proactive, create firstly accessible content.

So what are the alternatives that offer similar features that are fully accessible? I know you have mentioned ADAPT. I admit not having dived deep enough to know, but might guess it provides a certian set of interactive types- H5P is a broad swath of tools.

And finally, only 4 paragraphs here- what you describe as “H5P assessments” these are formal assessments that are scored, right? The “security” issue is that a learner might delve into the web source and parse out the answer keys stuffed in the code? There is a wide variety of H5P use that is outside that scope, e.g. activities for self practice, reinforcement, that is not assessed (ie. Branching Scenarios for Case Studies, though likely that content type is an accessibility problem too.

Hoping to hear from more folks out there. Thanks Delmar!

Oliver:

I just noticed your comment. Those are good questions. We do update the question types on our platform from time to time (I should do it this week- thanks for the reminder). We can review the questions again for WCAG, but it took a bit of time to address last time. Is there an errata page of updated question logs somewhere to make this easier?

I know I checked a few questions recently and there weren’t any updates to the a11y issues of those problems (I don’t know which ones off the top of my head). However, in the hypothetical and unlikley world that all works fine with h5p from a tech perspective, that still means there are still issue:

  1. with the questions that cannot be made accessible (e.g. drag and drop) and
  2. issues wtih questions that authors didn’t build the questions with good a11y work-arounds (rarely the case from my experience).

#2 requires manual editing (AI or human) on each clone of each question, which is immensely time-consuming at-scale (nearly impossible) for a decentralized approach (unlike LibreTexts and eCampus Onterio and some of the German sites).

#1 requires introducing an accessible alternative to use with a bit of tech to build and deploy. We have that in our ADAPT homework platform for those wanting to use H5P. But for non-LibreTexts community that just directly embeds these inaccessibile questions into their texts, this won’t work easily.

Regards,
Delmar

There is too much trouble with h5p - poor accessibility and no security [highlighted by myself] are the biggest.

I’d like to learn more about this.

There’s https://h5p.org/post-hub-releases which lists (all?) changes to content types. And there’s https://help.h5p.com/hc/en-us/sections/7555999389341-News, but it’s a separate page per release.

I am pretty sure that there have been a11y improvements that are not covered by the report. Just one example that comes to my mind: Dictation in fact allows to add a delay since January 2023 (the report dates to September 2022).

The software itself cannot guarantee full accessibility, of course. Authors can find a way to break it - but isn’t that an issue with all tools?

2 Likes

Sure.

Since h5p is client-side tech, it isn’t good for summative assessments like homework although formative is perfectly fine (as Alan mentioned). Students can hack into the console of browsers to see the answer, deactivate the internet to get the answer and turn it back on, or (sometimes) download the h5p files for offline dissection. I often meet adopters (potential and active) thinking that h5p can be used as a secure homework system and get shocked when I point out the different ways h5p assessment can be compromised (i.e., h5p is not secure). The community should emphasize better that h5p isn’t homework (unless used for low-stakes) although it may look like it. However, this will conflict with some companies’ business models of course.