Licensing status of APCA in Visual contrast of text

[thibaudcolas]

Apologies if this is explained somewhere that I couldn’t find. I’m also not that knowledgeable about software licensing, patent, contract, and copyright laws.

On Visual contrast of text, there is mention of how the new guidelines will be based on an algorithm called APCA. In the corresponding How-to, under Resources, there is a link to an APCA test tool at https://www.myndex.com/SAPC/.

I’m interested in adding APCA calculation into tools that evaluate contrast – however it’s not clear to me under what terms I’m allowed to do so, based on reading that resource’s licensing terms. As far as I understand, this says all code is licensed to the public as AGPL, which is too restrictive for my liking, and only W3 has a BSD-style license.

---

It’s really beyond me to understand which licenses may or may not apply – would it be reasonable for the specification / how-to to link to a resource that documents the APCA algorithm with a more permissive license / public domain dedication? Or even better, provide a reference implementation with a more permissive license / public domain dedication. Specifically something that doesn’t restrict derivative work and redistribution like AGPL.

As I understand, since that existing implementation is licensed to W3 under different terms, this could be as simple as W3C re-publishing the same implementation but distributed under a more permissive license.

[Myndex]

Hi thi @thibaudcolas

I am sorry about any confusion here.

The files named apca-w3 will eventually be licensed to the W3C once we get to the release stage. This is indicated several places, including the GitHub repo, which is the only canonical location for the public beta code.

APCA-W3 REPO

---

That should provide you clear details for everything.

In short, if the file has "apca-w3" in the file name, it is intended for the W3C.

Please only use the GitHub repo for source code. The SAPC site you mentioned is a development site with experiments and other fun resources, but it is NOT canonical and should not be used as a source for code. This is in part why that site has a stricter license.

Use only the APCA-W3 named files from the official GitHub repo or the apca-w3 npm package please. There is a separate discussion tab at the documentation repo for problems or questions.

Why Did I Make It Slightly More Complicated ?!?

This is beta and undergoing active research and development. I do not want incorrect or non-canonical or proprietary code elements to end up in the wrong place, such as an app, with no way for me to verify that all is correct with implementations.

I am trying to be very OPEN about the process, but this is an early stage and thus must restrict certain things to prevent something incorrect from becoming embedded somewhere. Also, not everything being created is going to be part of the standard. For example, the base algorithm is the essential technology that is licensed, but the experiments, other apps, and other copyrighted materials on the Myndex Technologies site are not.

Please let me know if you have any questions on this or on implementation. I am here to help!!

ALSO: Where did you get the /SAPC/ link?? Is it in a page in WCAG 3? We need to fix that if that is the case.

The correct, public facing stable version is in the /APCA/ directory, here:

https://www.myndex.com/APCA/

Nevertheless, for CODE and data, please use the GitHub repo I linked to above.

Thank you, and again sorry if there was any confusion.

Andy

Andrew Somers
W3 Invited Expert
Myndex Color Science
Inventor of APCA

[thibaudcolas]

Hey @Myndex, thank you for getting back to me. It’s very exciting to see the improvements this brings to contrast checks!

I think the gist of the confusion to me is "licensed to the W3" (emphasis mine). As far as I understand, if it’s only the W3 that has a specific license (regardless of which), I as a member of a public would have to abide by AGPL.

If you say:

In short, if the file has APCA in the file name, it is the W3C three clause BSD license.

… then I interpret that as those files being licensed to the public as W3C 3-clause BSD License, which means I can then create my own integrations with licensing of my choosing.

---

I got the /SAPC/ link from the official Visual contrast of text how-to, under the "Resources" tab. it’s listed as APCA tool.

As a resolution to this, do you think it would be appropriate for this list of resources to include a link straight to the base algorithm / your reference implementation, in addition to the non-canonical / proprietary online tool (which I imagine is useful to lots of people as well)?

I appreciate this is still early days, but considering how bad the current WCAG contrast calculation is, personally I’d really welcome tooling not waiting for WCAG 3 to be a standard in order to transition to more appropriate contrast checks.

[Myndex]

Hi @thibaudcolas

… then I interpret that as those files being licensed to the public as W3C 3-clause BSD License, which means I can then create my own integrations with licensing of my choosing.

Yes

We license to W3, then they subsequently license to the public via a fairly permissive license. I suppose I am being pedantic here (I happen to love the law if Sosumi rings a bell, LOL).

---

As a resolution to this, do you think it would be appropriate for this list of resources to include a link straight to the base algorithm / your reference implementation, in addition to the non-canonical / proprietary online tool (which I imagine is useful to lots of people as well)?

It does, on THIS link: https://www.w3.org/WAI/GL/WCAG3/2020/methods/font-characteristics-contrast/

BUT...Hmmm... Okay, this is useful information, indicating a navigational issue in terms of accessing relevant resources.

---

I appreciate this is still early days, but considering how bad the current WCAG contrast calculation is, personally I’d really welcome tooling not waiting for WCAG 3 to be a standard in order to transition to more appropriate contrast checks.

You... and apparently the rest of the world. !!! I hoped this might grow into something big, but I am surprised and nevertheless happy at the positive response. I never really thought of myself as a "math guy", more of a creative type... BUT it seems I'm a math guy by default because I happen to like math and, and with my love of the law, I guess that makes me an outlier^squared... LOL. The fact the old 2.x used math that literally does the opposite of what it is supposed to (and claimed to) still baffles me to no end. But it is fixed now, and hopefully wide adoption will improve readability across the web.

The tool is working and "available" to use now, though the WCAG 3 as a standard is not yet "official" and laws for some nations may vary, as they still mostly cite the old 2008 WCAG 2.0 (not even 2.1 !!) — it is such a massive paradigm shift, I've been told APCA cannot be applied to WCAG 2.x, but in Silver/WCAG 3 we have a clean slate and a fresh start. Also while the APCA based guidelines are demonstratively superior, I cannot comment on any nation's legal status in this regard.

Please let me know if you have any questions regarding implementation or use.

Thank you

Andy

[thibaudcolas]

Thank you! I think this is pretty clear for me now. I’ve only just begun reading through the WCAG 3 draft, and hadn’t seen Method: Font characteristics contrast. I’ll let you / others be the judge of whether this issue should be kept around so as to improve how those resources can be found, or whether the current links are enough.

For WCAG 2 contrast, yes, I have yet to spend enough time with APCA to make a decision but my plan was going to be to:

• Keep on following WCAG 2 AA for legal compliance.
• Encourage everyone I work with to also use the WCAG 3 guidance to validate any color & font combination.
• Potentially (still not clear in my mind) stop aiming for WCAG 2 AAA in the occasions I would have, and instead aim for an APCA score of "5".

I think what will help me a lot with understanding this as well is having a tool that computes both scores at once – so I can get a feeling over time of where the biggest discrepancies are. For example, whocanuse already has fields for font-size and weight (well, "Bold"), so it doesn’t feel like that big of a stretch to have APCA on there side by side with the WCAG 2 measure.

Anyway, I’m looking forward to spending more time with this. Thanks again for the help!

[jspellman]

I am asking the W3C Staff Contact to weigh in on W3C resources around licensing.

[michael-n-cooper]

In the AG charter it states "This Working Group will use the W3C Document license for normative deliverables and the W3C Software and Document license for informative deliverables." The document license is fairly restrictive but allows limited quoting; the software and document license allows more broad excerpting and repurposing.

[ChrisLoiselle]

@jspellman This is ready for survey. This issue is around licensing. Appears to be closed from a response / question aspect.

[ChrisLoiselle]

DRAFT RESPONSE: This issue is around licensing.

We believe the questions have been answered by Michael C by his comment #266 (comment) .

I hope this answers your question, if it does not, please feel free to follow-up.

Thank you,
Chris

[kepstin]

The W3 Software and Document Notice and License states that in addition to the license text there, you must also include:

• Any pre-existing intellectual property disclaimers, notices, or terms and conditions.

Can you please clarify what, if any, pre-existing intellectual property disclaimers, notices, or terms and conditions apply to the APCA code examples & tests as included in the WCAG specification?

The license in the apca-w3 repository includes some additional terms - e.g. prohibiting some use cases, arbitrary license termination, prohibition against modification - that would prevent the code examples & tests from being used in many applications - including web browsers! Do these terms still apply to the code as published in the WCAG specification?

My understanding is that the W3 Document License does not include a similar statement regarding the text of the document, which means that a separate implementation of the APCA algorithm written based on the description of the algorithm in the document text (rather than by porting the example code) would not have these additional restrictions applied, is that correct? (Any limitations of use of the APCA name/trademark would continue to apply, of course.)

[Myndex]

Hello @kepstin

I just now saw your post while reviewing some old threads, and I do apologize that it slipped by my view, so please allow me to answer, albeit late.

The W3 version of APCA (by whichever name it may have at some future date) intended for guideline use, will have the appropriate permissive license, once WCAG 3 moves to the recommendation process.

Right now, APCA and related projects are in an extended public beta for trials and proving in an open way. We had some unfortunate occurrences with individuals creating very faulty plugins and claiming to be "APCA", while providing bogus results. Because of the problems and misunderstandings these events caused, and the importance of ensuring that instances in the wild are providing accurate and useful results, we had to be become slightly restrictive in order to remain open with the public beta.

The alternate, to lock up and close the project during development, was not something that I personally consider acceptable. Since the beginning of the process with wcag thread #695 in April of 2019, the development and testing has been a matter of open and public view. Not only am I on record stating that there will be the appropriate permissive license, we have been issuing more permissive licenses on request as needed.

If you have any questions, you can contact me direct or open a thread at the APCA Discussion Forum.

Thank you for reading