In this document we explore remediation approaches that can be applied after the source layer to deliver more accessible web experiences to end users. Approaches considered include JavaScripted modifications, and realtime HTML and CSS rewrites on content delivery networks (CDNs) as well as in browser extensions—all without writing any of these modifications into the original source. The capabilities addressed range from improving visual presentation to ensuring dynamic content accessibility and automated remediation of common issues. We also discuss the challenges and benefits of automating these adaptations, highlighting the balance between human oversight and machine-driven solutions.
We show that web content can be made more accessible for users with diverse needs, particularly those relying on assistive technologies
by leveraging edge technologies.
We conclude that integrating such capabilities at the edge provides an efficient way to address accessibility gaps and promote a more inclusive web experience.
What seems to be a single web page to most users today is actually anything but that. Today's world wide web is a complex blend of multiple original sources, mediated by server-side code to deliver a cohesive user experience. It is the result of a design and development process yielding experiences that work well for some users but fail for others, especially when W3C standards and best practices are indifferently engineered or inconsistently applied. This leaves many websites and online applications inaccessible to people with disabilities.
The current remedial approach for addressing this accessibility failure requires that developers update the source code to more closely conform with accessibility standards. While we certainly support optimizing accessibility at the source layer, modifying the original code may not always be feasible. It's also often the least efficient way to deliver different personalized experiences to different individual users. Furthermore, it can be viewed as too time-consuming and costly by business owners.
This document explores how various realtime post-source adaptations can deliver more accessible user experiences—and more varied accessible experiences—all without requiring changes to the original source. The diagram below illustrates where the adaptations discussed in this publication fit within a typical server and client data flow:
sourcelayer, though as noted above it's generally many sources, and not a single source.
Edge?
In this document the term edge technology
refers to technologies closest to the user. Often it's the technology directly in the user's hands, though sometimes it may be located on a content delivery network (CDN) server that is actively personalizing content for the user. The terms post-source
and edge
are essentially synonymous. We use the term edge
to emphasize that it is the user's experience that really matters when considering accessibility, rather than where in the data flow remediation is effected.
In the main the classic paradigm of a web domain serving content directly to end users across the Internet has long ago become a historical artifact.
Today's typical web page is often a dizzying composite of multiple content streams injected from various sources, including content unique to the user's locale and even to the specific user herself. Content
delivered to a user today may be a unique one-off
composite that may well change when the page is refreshed. The 2021 Web Almanac notes more
than 20 third party injection streams for today's typical web site, with 10% containing over 90 separate content injections.
The source of the typical website today is actually many sources. Even bespoke web applications aren't always created by a single source. Few developers exclusively use their own code. They rely on libraries, components and frameworks to build their web content.
None of this should surprise us. The fundamental nature of hypertext almost demands the behavior. It illustrates how appropriate the term web
actually is for describing the environment.
This section provides an overview of this document. It describes what feature capabilities are addressed in this document and then how each capability is examined and evaluated.
our analysis of various accessibility related capabilities is organized by the primary coding technology expected for that capability as follows:
We next discuss capabilities related to Management of accessible web service delivery that can be accomplished with edge technologies. This is followed by a discussion of capabilities and edge services we consider aspirational because they are yet to be realized in the web world. Lastly we append a discussion of the value of Human Centered Design to our analysis followed by a review of typical roles and responsibilities that belong to various players in the web world user experience ecosystem.
Each capability we examine is considered from the following perspectives:
This layer of delivered content consists of words, graphics, and rich media objects. In the main it is created using html, both static and dynamic.
Supply missing alternative text for content images.
This capability supplies a small pop-up window that appears on hover to display additional information about an element on the page for users who benefit from enhanced contextualization or clarification.
This capability correctly specifies the primary language of a web page when the declaration is incorrect or missing entirely.
This capability provides for meaningful titles, for pages and/or iFrames within pages, and provides screen reader users a more efficient and meaningful browsing experience.
EDITOR'S NOTE
Expand section content and include href to Media Accessibility User Requirements (MAUR) written.
This capability helps assure that a published PDF document is correctly tagged and marked up sufficiently in order that it can be accessed and more readily understood by people with disabilities, including those who use assistive technology such as a screen reader.
The capabilities discussed in this section are afforded users to facilitate content rerendering utilizing presentational characteristics each individual user finds most accomodative. These are accommodations that allow users to perceive and understand published content and serviceslook and feel a content provider may have chosen and published, the user's adaptations must necessarily take precedence for reasons of accessibility. To put it in other words: The author proposes, but the user disposes.
In the main presentation is achieved with css. We have a strong preference for css in presentation rendering because css is most fungible. Rendering the presentation layer through technologies other than css is highly likely to prove inaccessible and require remediation into css in order to support accessibility requirements as defined by W3C.
This section addresses capabilities related to enhanced focus visibility Isolating Specific Semantic or Interactive Components for Efficient Browsing and Access: methods and technologies used to enhance user experience by isolating specific elements within a digital interface. It enables users to navigate efficiently through content by focusing on particular types of elements, such as headings or buttons.
Some users require a more visible cursor and appreciate the capability that can increase the standard cursor size by up to 400% to ensure the pointer always remains visible to them. This facilitates faster and more accessible navigation through hyperlinks, tabs. and form elements.
Some people require greater clarity regarding which spans of text or graphics in web content are functioning as hyperlinks. This capability facilitates Selecting the text or image that serves as a hyperlink and causing it to be displayed in a different, more perceivable way, to help users identify the hyperlinks on a page more easily.
Some users benefit from a horizontal line that appears below the text they're reading in order to help them keep their place as they read. This facilitates staying focused.
Some users require different fonts or different font characteristics in order to comprehend and interact effectively with web content. This section groups those capabilities.
Not all users are able to read any font a web content publisher may choose. This capability allows the publisher' s content to be reformatted in a user chosen font the user is comfortable using
This capability differs from Fonts section above only in that the proffered fonts are understood to be particularly usable by people living with dyslexia. Typically, they have more distinctive symbols in order to avoid confusion, increased baselines and thinner tops, and a larger x-height.
This capability insures that screen reader users can understand the prior price of an item. Displaying the prior price using strike-through or crossed-out fonts, and then providing the new price immediately next to these does not work for screen reader users. These users will have no idea which is the prior and which the current price without additional content and tagging.
Some users need enlarged, or perhaps diminished fonts in order to read content comfortably. This capability allows increased or decrease font size, by supporting clicking an interactive widget in order to make appropriate adjustments.
Some users require additional white-space between lines of text content in order to read comfortably. Others require adjustment between individual characters in and between words. This capability supports these adjustments by facilitating the modification of text and line spacing to support personalized readability enhancement. As typically provided today, the capability offers three degrees of adjustments for a personalized and more accessible reading experience.
Some users benefit greatly by simply increasing white space in between lines of text.
Some users require pixel-level RGB transformations to comprehend and interact effectively with web content. This section groups those capabilities.
Some users require the ability to stop animations or other moving elements on a web page from playing entirely. Others may wish to see them under direct user control, i.e. as a series of steps.
Some users are impeded when media begins playing on page load. Screen reader users may need playback paused until they've assured themselves they've heard all the information their TTS can provide on any given page. Others will seek to enable captions (and possibly even automated language translation) before playback begins. It is also necessary to support a mute function which can be invoked by touch, mouse, keyboard, or other control mapped for such purposes.
EDITOR'S NOTE
Section content still under development.
Color and contrast modifications such as light mode
, Dark mode
, Inverted Colors, Color Desaturation, and related contrast pigmentation adaptations have proven highly useful
to many people with a wide range of visual impairments, and frankly for most users in various distinct situations. They modify the difference between the
lightest and darkest parts of a web page, reverse colors, or reduce their
intensity.
Configure Once, Use Everywhereprinciple.
In many cases.
EDITOR'S NOTE
Section content yet to be written.
Fully justified text blocks have long been standard in publishing, including in web content publishing. It is achieved by using variable width fonts and adjusting spacing between characters and words in the block of text. r this practice is not beneficial to all users. This capability allows individual users to determine for themselves the alignment characteristics that best accomodate their reading needs.
The capabilities discussed in this section are those content publishers must provide users in order to facilitate accessible user interaction with published content and web services. These are described by the fundamentalWCAG principle: operable
. In the main they are achieved using JavaScript.
Forms are the fundamental interface of the interactive web. Programmatic labels for each data entry field or option selecting widget on any form are critical for many users requiring accessibility support. Furthermore, consistent grouping and form controls on screen location is also critical.
This capability ensures that screen reader users and other users requiring enhanced status support receive clear feedback on the status of a form being completed along with concise error messages. This includes the colors used, the proximity to the form field, and the time-sensitive nature of the screen reader announcement.
WCAG explains in painstaking detail the critical importance to accessibility of structural markup. It is a fundamental requirement and any means to remediate or improve missing structural markup provides an immediate functional enhancement to users whose accessibility accommodations rely on explicit structural semantics.
mashup,strongly suggests a lack of concern for the coherency of the ultimate composite as delivered to the end user. Thus, anything that can be done to introduce greater semantic coherency is of benefit to all users, and especially those users relying on accessibility supporting features such as coherent and explicit structure. Lastly, as also previously noted, emerging strategies to further tailor content interaction to particular user's interaction preferences will rely on good structural markup, and these adaptations will, necessarily, be produced in edge technologies.
shouldarguments must give way to this simple rubric.
Notifications are messages shown to the end-user to alert them to an event, some specific data-point, or to some state. They are crucial for real-time functionality but can create accessibility problems if they disappear on their own (while the user may be otherwise engaged in the moment) or are not announced by screen readers. Designing effective notifications that are both timely and accessible often requires considerations such as user control over timing and the method of dismissal.
augmented realitywhich are potentially very exciting.
This capability provides definitions of words or phrases that may be unfamiliar to some users; provides pronunciation guidance; and provides synonyms which can all especially help users living with cognitive and learning disabilities better comprehend content.
Configure Once, Use Everywhere.
Offer a user-triggered method to This capability changes the language of a web page in real time, by translating the current page into a language selected by the user.
This capability allows a user to hear content spoken aloud by using a cloud (or browser) hosted Text to Speech (TTS) engine. There are now high quality TTS engines and listening while reading on screen is often beneficial for users in general, including especially users living with certain cognitive and learning disabilities.
It is only smart business practice and simple courtesy to support a one button click
feature to reset any user
triggered enhancement and restore the page to its default state. It should be as easy as possible for users to try different settings and undo them readily. Similarly, it should be just as easy to undo any auto-applied transformations and return a page to the settings provided directly by source. Users should be empowered to adjudge for themselves whether any applied overlay transformation actually enhances their ability to interact—and to return the overlay enhancements should they discover they're actually enhancing their experience, contrary to initial doubt!
The capabilities discussed in this section describe additional actionsthe content providers may take in order to shape the key elements or features of their digital content in order to support an intended user persona and journey. As stated above this typically results in a highly interactive experience presented to the user, packed with services supplied by third parties, many of which are added in their efforts to further accommodate, engage and please the end user while meeting regulatory requirements.
This capability helps assure that site content meets user needs as well as the the publisher's content policies. Typical policies will include everyday phrases with racist origins, and gendered or exclusive expressions and terms that are deemed to exclude audiences.
Automated transformations applied based on conditions edge technology discovers in the user's technology.
Algorithmic detection of violations of WCAG 2.x success criteria is considered here as a vector for providing the user a more conformant experience. There is, of course, the other application, where violation detection serves as a step in a source remediation process. We address the former here.
While not the glamorous holy grail
of technology deployment today, human mediation of A-I identified checkpoints, together with the site owner's active participation to clarify meaning and intent, describes an effective approach to accomplish accessibility remediations where they're needed. No other approach delivers equally effective remediations at such scale.
This capability describes an interface in the overlay, available to authorized users, facilitating manual editing of source code that affects accessibility, e.g. alt tags for images, or ARIA labels, often with the assistance of linting, wizards and supplemental materials.
This capability functions very much like the Accessibility editing, by site owner immediately above, except that a third party accessibility coding expert is granted access to perform needed remediations.
Site publishers need to support the full spectrum of accessibility features. On the other hand individual users need certain features and will ignore those they do not need. The types of features individual users need are often group-able into bundles, making it easier to turn on groups of features, e.g. high contrast and larger text size.
Configure Once, Use Everywhere.
Providing an easily located Accessibility Statement has become a widely accepted best practice in the industry. While there is no consensus on what all should be included, it is generally agreed that the formal Statement should define the site's commitment to accessibility of its content and services. Additional content, such as accessibility specific contact data, and accessibility specific help resources are also frequently included.
This capability displays to website users the name (and perhaps the logo) of the primary accessibility provider, who stands behind the accessibility of the site. This may be part of the accessibility statement.
This capability provides a form, or an email link, enabling end-users the ability to report barriers they've experienced on a site. Accessibility-specific reporting is an industry best practice.
The capabilities and edge services discussed in this section are aspirational because they describe development directions we believe edge technologies will take in the coming decade. They will rely on various specifications from the W3C around authentication, security, and privacy in addition to developments in W3C accessibility technologies.
Users now frequently use several devices. Furthermore, users tend to have the same or very similar accessibility requirements regardless of the device they may be using in the moment. It is therefore an emerging goal in web accommodations to support the user in configuring once and having their configured preference propagate across all their devices. Such a cross-Device preference profile should ideally function across all vendors and operating environments.
User can be afforded the opportunity to opt-in to various transformations via interactive personalization menus: The end-user can access a list of available enhancements and enable those they require and then reuse them across all sites visited day after day.
Many users with accessibility needs rely heavily on consistency in the user interface. Yet no standard exists, nor is likely ever to require certain common controls be located consistently across sites. This is very much in the design prerogative of the individual content publisher. And, while individual sites may be fully consistent in where they locate controls on the pages they publish, the user who comparison shops will encounter a different site consistently locating the same controls differently. The publisher has done nothing wrong, but the user is forced to rely on a perfect memory or put up with time-consuming strategies for locating these controls. Consider this example:
Standardizing the location of widely-deployed common components of a web site is now being addressed in the IETF and the W3C. An easily located Accessibility Statement is one expected outcome. While time will be needed for most sites to adopt the newly canonized web address locations once a standard is available, powerful benefits will help drive adoption and end users will benefit..
A compelling largely unmet need exists for legal surrogacy services, whether fiduciary, medical, or probate. Whether to access data stored in the account of someone recently deceased, or in the account of a minor child, or of a parent or social agency client, only a very few major web content service providers have begun addressing this need as of this writing and only for their own systems. This need is particularly acute where print disabilities are to be factored. Most legal jurisdictions today accept such instruments executed only on paper, thus discriminating against persons incapable of managing paper on their own. When provided for persons with disabilities this capability is today increasingly known as Supported Decision Making (SDM)
, and has historically also been known as Guardianship
.
Document format translation is important for most serious users of software, but arguably more so for the user relying on assistive technologies.
Calendar format translation is important for most any user of software, but arguably more so for the user relying on assistive technologies. While industry could long ago have solved this persistent incompatibility across operating environments, they have rather chosen to seek to impose vendor lock-in. This does not serve users who should be free to maintain calendars in the environment of their choice, or even different calendars in multiple environments with full transparency. This is not a technical challenge. Rather, it is a business choice.
Sometimes vendors support accessibility by mapping a series of keystrokes for screen reader users (and others who prefer keyboard commands to mouse clicks) to their own proprietary definitions. Often these duplicate standard command sequences used for similar purposes in widely used software. This capability proposes that overlay technology can remap proprietary command mappings to more commonly known command mappings, thus greatly reducing or even eliminating an unnecessary learning challenge for the end user.
It is often useful to skim through content to determine what, if any of it, might benefit the user by a more careful reading. Good user agent support for such a feature would not only support sequential summarization, but the marking of particular sections to return to for greater study.
Skimming through content to determine what, if any of it, might benefit the user by a more careful reading is just as valuable to users who depend on sign-language, or who utilize Augmentative and Alternative Communication (AAC) symbolic languages, as it for those who read text. As with summation services with text, good user agent support for such a feature would not only support sequential summarization, but the marking of particular sections to return to for greater study.
Our approach to the design and analysis in this document is rooted in the principles often called: Human-Centered
Design (HCD).
As an outlook that places people at its core, we believe HCD aligns perfectly with the goals of accessibility. This method of design prioritizes empathy, collaboration, and user empowerment, which aligns with
our mission to generate accessibility solutions that meet the diverse needs of various human individuals. Our understanding of HCG requires us to minimize technological configuration tasks and maximize each individual's ability to read and interact with actual content—not the application (or browser) chrome.
Among various design terminologies linked with accessibility such as
universal design
and inclusive design,
HCD sets itself apart due to some
specific distinct attributes, such as:
Our design philosophy does not merely aim for inclusivity; instead, it strives to create meaningful experiences that significantly enhance the lives of the people we serve. A successful application of HCD manifests not just in broad-reaching accessibility, but in generating rich, meaningful, and transformative experiences for the user.
Furthermore, HCD endeavors to build a comprehensive understanding of the end user, transcending the confines of mere accessibility. It delves into the emotional, cognitive, and social dimensions of the human experience, with a commitment to design solutions that resonate on a profoundly personal level.
Our commitment to HCD is woven throughout our efforts to ensure accessibility. In prioritizing a user's individual needs and emphasizing their active involvement in the design process, we believe we can create a more accessible and inclusive digital environment. By employing HCD, we hope to deliver a seamless, enjoyable experience for all users, that is more than just accessible - it is truly human-centered.
The capabilities discussed in this document were originally organized under two primary categories with several subcategories under each primary section conforming with client-server architecture, the backbone of all digital transactions and interactions that take place over the web. These roles are still important to note, so we have preserved our analysis in this Appendix. The primary categories are:
The Content Provider, analogous to the server, plays the pivotal role of specifying and delivering diverse forms of digital content ranging from text, images, audio, video, to interactive elements. They curate the experiences that users consume and engage with, thereby shaping the digital landscape. A selection of some primary Content Provider roles:
Role | Description |
---|---|
Author | Responsible for generating the primary content, be it text, script for audio/video, or conceptual outlines for interactive elements. |
Designer | Work on the aesthetic aspects of the content including the layout, color scheme, typography, and visual elements. They also handle the user interface and user experience design for interactive content. |
Product Owner | Oversee the overall content creation and distribution project, make key decisions and coordinate between different teams to ensure the final product meets its intended objectives. |
Developer | Implement the designs into functional code. They work on the website or application where the content will be hosted, ensuring its responsive, functional, and accessible. Some write proprietary algorithms, but most call and configure frameworks and libraries written by communities of coders. |
Marketer | Oversee and implement the marketing strategy for the content. They coordinate marketing campaigns and work closely with other teams to ensure the content reaches the desired audience. Growth marketers utilize data-driven marketing techniques to help the business expand its customer base. They use various strategies and tools to attract, engage, and retain users driving critical website requirements. |
On the other end of the spectrum is the End User, or the client, who experiences this digital content. End users span a wide range, varying in their demographic, cognitive, and physical attributes. They also bring to the table their unique needs, preferences, and accessibility requirements. It is for these users that content providers shape their digital content and experiences, with or without the knowledge that the user' s own hardware and software may render it in ways the Content Provider never imagined.
Content sites are inherently dynamic; they are assembled for the end-user in a mashup, a unique blend of content that can vary significantly from user to user. A selection of some of the hardware, software, tools and technologies that participate in this edge activity include:
Technology | Description |
---|---|
User-Agent (Browser) | Interprets and presents web content to the end user. Different browsers may render web content in slightly different ways due to their different rendering engines. |
Assistive Technology | Devices or software applications that assist individuals with disabilities or impairments. This could include screen readers, alternative input devices, and voice recognition software. |
Browser Extension | An add-on or plugin that extends the functionality of a web browser. Could alter the appearance of web content, block advertisements, or add additional features to a webpage. Includes mashup tools or web services that combine content from more than one source into a single integrated experience. These can greatly modify the original content and how it is presented to the user. |
OS or Software Application | A program or group of programs running on the local operating system. Applications can vary drastically in purpose, from productivity apps to entertainment apps, and can influence how digital content is accessed and interacted with. |
Video/Audio System | Hardware and software components that control the output of video and sound. |
Content Delivery Network | Store copies of web content in various locations around the world to reduce the physical distance between the server and the user, improving site performance. This intermediary, typically invisible to the end user, plays a critical role in delivery and updating of digital content, as well as authenticating users. |
This two-poled digital communication model is not a one-way broadcast, but a dynamic conversation. Feedback from end users can influence and improve the content generated by providers, while content providers continuously adapt their strategies based on user behavior and needs. The feedback loops become more complex as edge technologies come into play.
Third party services not hosted by the Content Provider, providing services in support of the Content Provider's value proposition. Typically invoked by JavaScript or the Content Delivery Network, they meet the planned needs of end users. Accessibility is a complex subject, and approaches differ per role. Suitably trained designers, brand owners, marketing managers, software developers and quality assurance are difficult to hire and retain. While some rare web pages may yet serve up a single experience like a short order cook handing over a hamburger, most are like the food court at a shopping mall with many independent services rallying to the end user's needs. Content providers pack their pages with services such as cookie permissions, chat, payment gateways, user-generated reviews, and most prolifically advertisements. Whether these mashups are accessible can only ever be discovered by evaluating the end user's experience. Content providers today may guess at an end user's actual experience by testing a few carefully chosen scenarios. Large organizations may do better and arrange focus groups that give a glimpse of the actual experience of a few users at one moment in time. However the larger a web presence to be tested is, the more impossible it is to test all possible scenarios prospectively. Edge technologies, on the other hand, are actually well positioned to
identify, quantify, and report on precisely those accessibility challenges
that end users are experiencing. Furthermore, far more end user experiences can be evaluated and reported by edge technologies than
human testing could ever cover. Gathering and reporting many samples to upstream end points is
arguably the most efficient way to spot anomalous patterns (and a great
application for A-I analysis). Source can only impute what the edge can confidently demonstrate. Regardless the testing approach, only the edge knows for sure whether any particular web content experience can be called My Site, Their Service
Having Knowledge and Skills
and What Only The Edge Knows For Sure
.Having Sufficient Knowledge & Skills
My Site, Their Service
.What Only The Edge Knows For Sure
accessible.
The following individual participants of the Accessibility at the Edge (A11yEdge) Community Group contributed to the development of this document: