The People v. PDF

Table of Contents

1. Accessibility and the PDF
2. Accessibility of PDFs
3. Readability and the PDF
4. Security and Versioning
5. Better Practice
6. Recommended format by content type
7. References

1 Accessibility and the PDF

The portable document format (PDF) is a digital file format “used to represent a document in a manner independent of the application software, hardware, and operating system used to create it” (Bienz and Cohn, 1996, p. 19). A PDF should look exactly the same on any computer or device. Sometimes, that is ideal. Consistency in appearance is noted as a strength of the PDF as a file format, but it comes at significant cost to accessibility. PDFs may be barriers for users of assistive technology (e.g., screen reader or text to speech software), users who need to increase the font size, users who prefer to invert the colours, or anyone reading on a phone.

Accessible content is essential to equitable participation, democratic access to information, and inclusion. In 2022, the number of Canadian first-year university students with a disability was 31% of the total student population and by 2024 35% of graduating students had a disability (Canadian University Survey Consortium, 2022, 2024). The exact nature of the barriers those students encounter is rightly confidential, but according to Statistics Canada, in 2023 5.2 million Canadians had a disability related to print material (McDiarmid, 2023). These figures suggest a substantial number of students may struggle with reading. As PDFs are used to distribute reading material, articles, assessments, and many other types of content—and because PDFs are largely inaccessible—PDFs present a significant impediment to access and participation for many students. Specifically, blind and low vision users, users with cognitive disabilities and/or reading impairments, and users with disabilities impacting motor function may find PDFs difficult or impossible to use.

Ensuring that all students can access and participate is an important tenet of education. For those that are unmoved by this sentiment, the Accessible Canada Act (2019) and the Accessible British Columbia Act (2021) govern a wide range of accessibility concerns, including education and digital spaces. One of the simplest steps to increase equitable access (and meet legislative requirements) is to drastically reduce the number of PDFs used at Langara.

“For the rights of the disabled to mean anything in today’s world, they must be extended to cyberspace no less than to parking spaces.”

(“A More Accessible Internet,” 1999)

While use cases continue to exist for the PDF, much of the format’s appeal is undercut by more modern, adaptable, and user-friendly formats. The following article will explain the issues with PDF as a digital file format, the difficult process required to make PDFs accessible, how PDFs are not user-friendly, and some suggestions for better practice to reduce the number of PDFs we create and use. Reducing reliance on PDFs is an extremely low-effort, high-reward step we can take to make Langara more inclusive, equitable, and accessible.

While the PDF excels at consistently reproducing the look and layout of a document across devices, this is not always ideal. The PDF is great for its intended use, but not necessarily for what it is often used for.

2 Accessibility of PDFs

Technically, a PDF is a set of instructions to reproduce content virtually (on-screen) or physically (in print) (Darvishy, 2018). The issue is that the vast majority of PDFs are improperly created as images of the content which are not machine-readable or accessible.

To create a PDF each character, shape, and colour is stamped onto the page. For example, in a PDF the word ‘quote’ is not a word to a computer. Rather, it is a collection of shapes that make up the letters ‘q u o t e’, but not the letters themselves. That system is what makes it possible for a PDF to look identical across devices; the PDF is instructions to tell the computer to ‘print’ q u o t e. As such a collection of shapes have no meaning to a computer (and in turn assistive technology), they need something to define them. Tags are an additional layer of coded annotations that explain each stamped piece of content. Tags make PDFs machine-readable and allow PDFs to be searchable, to be highlighted and copied, and to be navigated by screen readers and read aloud by text-to-speech (Faulkner, 2017). Every piece of information including titles, headings, paragraphs, links, images, formulas, tables, and other semantic elements need to be individually tagged. Tagging PDFs requires significant expertise, time, and patience.

2.1 Making Accessible PDFs

To be PDF/UA compliant, a PDF must have tags, logical reading order, sufficient colour contrast, logical heading structure, alternative text for images, and meet other criteria. Barriers to creating accessible PDFs include lack of awareness about accessibility (Pradhan et al., 2022), lack of understanding of accessibility guidelines (Schmitt-Koopmann et al., 2022), and a misunderstanding of the impact inaccessible content has on readers (Caffrey-hill et al., 2021). Making PDFs accessible requires expert knowledge (Williams, 2018) due to the complexity of the PDF format (Pradhan et al., 2022). Additionally, the significant investment of time (Schmitt-Koopmann et al., 2022; Pradhan et al., 2022) and lack of motivation to create accessible content (Kumar and Wang, 2024) contribute to the difficulties in making accessible PDFs.

Given limited digital literacy, time, and resources, the process of remediating inaccessible PDFs may be beyond many creators (Chee et al., 2022). In an academic context, the decentralized nature of colleges and universities (Lazar, 2022), the increased demands on instructors (Penchev and Todoranova, 2025), and the barriers to faculty training (Bong and Chen, 2024) all present complex challenges to creating accessible content. By Adobe’s own admission (albeit in an advertisement for their AI-powered cloud remediation service), “tagging PDFs can often be slow, manual and error prone” (Adobe Developers, 2023).

Manually creating or editing tags is not intuitive (Kumar and Wang, 2024) due in part to the complicated nature of PDF editing and remediation tools (Pradhan et al., 2022). The most common software to make PDFs accessible is Adobe Acrobat Pro which is a paid software, of which only the more expensive option has the features necessary to create accessible PDFs. Adobe Acrobat has been described as not usable for inexperienced users, lacking an undo function (meaning an error can destroy hours of work), frustrating to the point of abandonment, having a complicated visualization method for tags, lacking a clear workflow, and having an unclear interface (Bigham et al., 2016; Rajkumar et al., 2020; Schmitt-Koopmann et al., 2022).

Once tagged, the Reading Order must be (re)verified and the Content containers (refer to Figure 2) may need to be reorganized to ensure nothing in the tagging process (which does not have an undo function) broke the visual layout of content. No research was found acknowledging these additional barriers to making PDFs accessible.

2.2 PDF Accessibility Statistics

Due to the difficult process required to make PDFs accessible, few are. There are, according to Adobe’s own boasting, 2.5 trillion PDFs, or 312 PDFs for every person on earth (Still, 2020). An examination of 200 disability studies articles published as PDFs between 2009 and 2013 found that 95.5% were not tagged, 99.5% did not have a logical reading order, and 0 had a consistent heading structure (Nganji, 2015). However, 75% of the articles had an equivalent Hypertext Markup Language (HTML) version. An extensive study of 11,397 academic papers published between 2010 and 2019 found that only 2.4% met all basic accessibility criteria (Wang et al., 2021). A review of 20,000 papers published between 2014 and 2023 found that only 3.2% of PDFs satisfied all accessibility criteria while 74.9% failed to meet a single accessibility criterion (Kumar and Wang, 2024). Of 2500 PDFs in open-access repositories published between 2018 and 2022, only 11% had both tags and logical heading structure (Darvishy et al., 2023). A 2023 review of 18,627 PDFs found that 91% did not have tagged headings and 58% did not have tags (Iannuzzi et al., 2025). 8000 PDF articles from Springer, Elsevier, ACM, and Wiley were reviewed in 2024 and 0 met the PDF/UA standard (Pierrès et al., 2024). Even Adobe acknowledges that of the “trillions of PDFs out in the world… more than 90 percent of those PDFs today are at least partially inaccessible”(Yau, 2023). The number of inaccessible PDFs is staggering and grows every day.

“If our digital memory is going to be in PDF, we must ensure that it is accessible to all persons.”

(Turró, 2008, p. 25)

2.2.3 Conversion

Recognizing the challenges to making PDFs accessible and to address the trillions of inaccessible PDFs, some organizations are converting existing PDFs to HTML rather than emphasizing remediation efforts. HTML is not guaranteed to be accessible but has a higher base level of inherent accessibility and is much easier to make accessible compared to PDF. Many of these processes to convert PDF to HTML utilize an algorithm developed by the PDF Association (PDF Association, 2019). Callas Software partnered with the PDF Association to develop an overlay that allows users to view PDFs as HTML and make use of the adjustments offered by HTML such as structure tags, font customization, and custom contrast. These adjustments, according to the PDF Association, offer “a more flexible reading experience” and “make the document accessible for people with visual disabilities or dyslexia” (PDF Association, 2017). Similarly, TransPAC is an overlay for PDF forms which allows users to complete an HTML version of the form which then auto-populates the PDF form. As HTML is generally more accessible than PDF, this avoids the barriers to users and saves creators time making PDF forms accessible (Uckun et al., 2020). Pdf2htmlEX, MaxTract (Sorge et al., 2020), and Scia11y focus on converting science and math articles to HTML (Wang et al., 2021). NgPDF converts PDF/UA compliant PDFs to HTML (Moore, 2023) while DocAccess provides an accessible HTML conversion of PDFs in real time. There does not appear to be any research assessing the accuracy of these products.

ArXiv, an open-access archive for nearly 2.4 million scholarly articles, receives most of their submissions as TeX (a digital typesetting program) and PDF. To address the accessibility of their repository, arXiv has implemented automated conversion from Tex and PDF to HTML. ArXiv notes that the conversion prioritizes function over form, and some HTML content will not look the same as the PDF generated from Tex, while acknowledging the enhanced accessibility and better adaptation to user devices that HTML offers. The submission process offers authors the opportunity to review the converted content before posting (arXiv, 2023). Similarly, many major publishers and databases now offer articles and content in numerous formats, not just PDF.

This is the inflection point. Almost no one creates a PDF. For example, a document is created in Word or TeX and then exported to PDF for sharing purposes. As more and more people and organizations come to understand the problem with inaccessible PDFs and the difficulty in making PDFs accessible, there are now considerable efforts to convert existing PDFs to HTML. The PDF is an unnecessary intermediary, taking a cut (in time, frustration, and exclusion), that we would all be better off without.

3 Readability and the PDF

Even when a PDF is made accessible, the format is not user-friendly for readers, users on mobile devices, and users of assistive technology.

In contrast, modern web content is responsive to the viewer’s screen size. For example, a web site will reflow content into a column that fits the width of the user’s device without cutting off content or requiring horizontal scrolling. In 2025, 62% of all web traffic is on mobile devices (StatCounter, 2025). You may not think students are using their phones for school, you may not want them to, and you may not like that they are, but they are. PDFs are not responsive and do not reflow to suit the user’s screen size. The problematic nature of PDF’s inability to reflow is so universally understood, Adobe released Liquid Mode in 2020 to combat the issue. Adobe posits: “If you’ve ever read a PDF on your smartphone or tablet, chances are you’ve probably gotten frustrated by the amount of pinching and zooming you had to do” (Adobe, 2020). This acknowledgement undermines Adobe’s own position that PDFs “work anywhere with any device you have on hand” (Adobe, n.d.). Liquid Mode uses Adobe Sensei AI to analyze a PDF and attempt to reflow the content into a device-responsive layout (Adobe, 2020). A 2023 study found that Liquid Mode on Android and iOS and Reflow view in Acrobat Reader on Windows each displayed the same document in different ways (Cavazos, 2023). Additionally, Liquid Mode only works in the Acrobat application, again undercutting the presumption that PDFs work on any device.

The inability of PDFs to natively reflow impact someone who requires font to be larger. Refer back to the video and note that when zooming on HTML content the text became larger, but each line became shorter. That’s reflow. On a PDF, a reader can zoom in to make the font larger, but that then requires horizontal scrolling to see all the content; a PDF does not reflow. This requires constant horizontal scrolling back and forth, which is awkward, can cause someone to lose their place, and excludes users with physical disabilities. Reflow is essential for viewing content on mobile devices and people who need larger text.

3.1 “People Prefer PDFs”

As PDFs are so ubiquitous, we may be operating on the assumption that people like them. However, research overwhelmingly indicates that people avoid PDFs. In 2013 an Australian government agency found that 99.3% of users chose web pages over an equivalent PDF with that percentage rising to 99.5% in 2015 (Wild and Craddock, 2016). A 2014 study of World Bank documents found that 31% of PDF reports are never downloaded (Doemeland and Trevino, 2014). A 2017 study of university students found that students accessed a website version of a template 15.2 times more often than the comparable PDF version (Valentine et al., 2017). Students prefer lecture slides and notes to be provided in PowerPoint or Word as opposed to PDF, as the former are more easily manipulated (MacCullagh et al., 2017). A 2018 article noted that a company’s website had 178 page views for every PDF downloaded (Craddock, 2018). A 2019 study of American university students found that nearly 70% preferred educational material in HTML, compared to 25% preferring PDF content (Noyes, 2019). A 2023 study found that students overwhelmingly prefer web content over PDFs to distribute course notes (Bhute et al., 2023). PDFs are not preferred by readers, do not work well on smaller devices, and are difficult to make accessible. Reflecting on the continued reliance on the PDF is not only wise, but imperative.

3.2 Customization

One of the benefits of PDFs is the ability to reproduce consistent layout and styling. This can be useful for visual layouts like posters and infographics or when including complicated or non-standard text passages as well as non-alphabetic scripts. However, this rigidity may be a hindrance for some readers. Font style, size, and background colour are important factors for digital reading (Divya and Mohamed, 2020). Sans-serif fonts in sizes between 12 and 14 points are shown to be the most readable styles (Morrell & Echt, 1997; Lin et al., 2013). Customizing the settings of a device (such as increasing font size, changing default colours and contrast, and using tools like magnification) can make technology easier to use (Wood et al., 2024). Ultimately, different customizations, font size for example, work best for different people. Accounting, and allowing for, individual user preferences allow readers to consume content in ways that are easy and convenient for them (Kim, 2009). While universal design has a popular appeal, digital interfaces offer the possibility of personalized usability and optimized accessibility features through individual design (Peissner and Edlin-White, 2013). One of the possibilities of digital content is the ability to convert information into potentially infinite configurations and representations of the same data (Manovich, 2001) “represented in accordance with the needs and desires of different users and contexts” (Ellcessor, 2010, p. 304). PDFs, even a fully PDF/UA compliant file, do not afford these customizations (Ganner et al., 2023). The font style, size, spacing, and contrast set by the author are locked by the PDF format, as intended. Many other formats, such as HTML or Word, allow users to customize their reading experience.

Allowing readers to tweak their reading experience to suit them has a positive impact on user experience. The PDF, by design, does not afford readers this personalization. Providing content solely in PDF as a rigid, one-size-fits-all format removes user autonomy and does not allow for the uniqueness of individuals. This makes reading harder for everyone and prevents some users from engaging with content at all.

4 Security and Versioning

4.1 Security

One supposed benefit of the PDF is security. This assumption is based on the fact that some creators don’t know how to edit a PDF and assume no one else can either (Craddock, 2018). PDFs are not automatically secure and must be locked using specialized PDF software such as Adobe Acrobat Pro. PDFs are only as secure as the protections applied, and those protections are easily bypassed. A simple online search reveals numerous tools that can bypass PDF security layers. In fact, even a secured PDF can be duplicated (and subsequently edited) by taking a screenshot of the PDF. On the other hand, HTML content requires no additional steps to be secure (a web server would have to be hacked to modify HTML content) and is always the most current and relevant source of information (Wild and Craddock, 2016; Craddock, 2018). Creating content in Brightspace will always be the most secure, and most current, option.

4.2 Versioning

Once downloaded by a reader, a PDF becomes a time capsule, potentially out of date the instant it was uploaded (Oxley, 2020). Many PDFs are created, uploaded, and abandoned (Collins, 2020). Any subsequent updates to the information contained in the PDF will not be available to anyone that has previously downloaded a copy. PDFs are harder to maintain, track, and update (Williams, 2018; Ricablanca, 2020). Managing updates of PDF content is lengthy and awkward (locate source document, make edits, ensure PDF is accessible, upload new copy, update any links to previous copies, locate and archive previous copies) and even if older versions are located and archived there is nothing to be done about previously downloaded copies that exist on users’ devices (Williams, 2018; Nielsen and Kaley, 2020a; Brinn et al., 2024). PDFs are not dynamic, not secure, and rarely accessible.

5 Better Practice

The simplest solution to ensuring accessible, equitable access to digital material is to offer content in multiple formats. The ideal solution for many content types is to offer HTML with PDF as an optional download where potentially useful (Aalbersberg, 2013; Nganji, 2015; Nielsen and Kaley, 2020a; Ricablanca, 2020; Disabilities, Opportunities, Internetworking, and Technology, 2022). However, creating and maintaining multiple formats places additional burden on already time-constrained instructors. Managing separate HTML and PDF versions requires updating content in multiple locations, ensuring consistency across formats, and potentially learning new technical skills. This additional workload can be particularly challenging given the increased demands on faculty (Penchev and Todoranova, 2025) and limited resources for training (Bong and Chen, 2024).

Starting with HTML can eliminate the need for multiple formats entirely. If content is created, and can be shared, in a more accessible format from the start there is no need for alternate formats, remediation, or users being unable to access content (Wild and Craddock, 2016; Bowes III, 2018; Nielsen and Kaley, 2020a). But, as is, alternate formats, remediation, and exclusion exist and the numbers are rising exponentially. Academic accommodations are often the only available option but require significant time and resources. Additionally, accommodations require no changes to pedagogy, procedure, or culture and simply treat barriers as they arise and requires the student to adapt their experience “to the dominant logic of classroom pedagogy” (Dolmage, 2017, p.80). Avoiding PDFs may reduce the reliance on, and the number of, academic accommodations required for students to participate and succeed.

5.1 HTML

HTML is code that defines the content and structure of web content. HTML is a markup language which specifies the structure and formatting of a document, the relationships among its parts, and cues for its appearance. Where tags, as discussed in section 2, are an additional layer that must be added to a PDF, HTML includes tags and logical reading order by default (Lazar, 2021; Lewis, 2022). HTML tags are added automatically by ‘what you see is what you get’ editors such as those in Brightspace, PebblePad, WordPress, and other common web platforms. HTML has inherent semantic meaning for text, headings, lists, tables, images, and other content types.

In this document, for example, ‘5.1 HTML’ is tagged as <h3> which means it is a heading level 3. <h3> provides semantic meaning to assistive technology and cues the cascading style sheets (CSS) to style the text as a heading level 3. In a PDF, ‘5.1 HTML’ would just be the collection of shapes ‘5 . 1 H T M L’ and would require tags to have any meaning. As discussed in section 2.2, very few PDFs are tagged and the process is frustrating and complicated. So much so, that, as discussed in section 2.2.3, there is a significant trend toward converting existing PDFs to HTML.

That is not to suggest that HTML is inherently accessible. In fact, WebAIM compiles an annual report on the accessibility of the million most visited home pages and there are scores of errors per page (WebAIM, 2025). There is significant work to be done on web accessibility. The inaccessibility of web content causes significant barriers and frustration for many. However, all but the most egregious failures on a web page will not be as exclusionary as an untagged PDF (of which roughly 90% of 2.5 trillion PDFs are) which is essentially just a picture with no meaning to users of screen readers and text to speech tools and is nearly unusable by those requiring larger or other customized font. Advocating for HTML implies creating accessible HTML, which is not automatic but is much easier to achieve than in PDF and has at least basic levels of accessibility by default (Mankoff, et al., 2020). If you’d like to learn about creating accessible content in Brightspace start here to address Brightspace accessibility issues.

The research (section 3.1) shows people prefer HTML. Additionally, as discussed above, PDFs are a set of instructions to visually (or physically) print content with the exact same layout and appearance regardless of device. That means, by design, a PDF cannot be customized to suit reader preferences or needs. Customization can significantly enhance the readability of content and HTML lends itself to a wide array of customizations and tweaks (section 3.2). The PDF is not suited for many of the uses thrust upon it. As awareness of PDF inaccessibility—and the significant time to remediate PDFs—grows more and more jurisdictions are suggesting avoiding PDFs.

Post-secondary institutions are moving toward limiting the use of and reducing the number of PDFs in favour of HTML. The University of British Columbia and New York University suggest careful consideration of whether a PDF is necessary. The University of Arkansas, Concordia University, Florida International University, Penn State, and Princeton all recommend avoiding creating PDFs with an explicit call to use HTML instead. Harper College, Kentucky Community & Technical College System, NC State University and Wake Forest University all recommend eliminating the use of PDFs and preferring HTML or other document types. Appalachian State University, the University of North Carolina School of the Arts, and the University of Texas San Antonio recommend converting PDFs to HTML or remediating them to be accessible. The University of Texas at Austin, the University of Waterloo, the University of Wisconsin-Madison, and Yale all emphasize the use of HTML. Chico State University and Williamette University stress that PDFs are only to be used for content that is to be directly printed, not read on a computer with the latter advocating for HTML in place of PDF. The University of Washington recommends alternatives to PDFs where possible, while Toronto Metropolitan University states if a PDF cannot be avoided, it must be accompanied by an accessible HTML version. The University of San Francisco states the PDFs must be accessible, noting the difficulty in doing so, while recommending HTML and other accessible formats.

The 23 institutions listed above do not constitute an exhaustive list of all post-secondary institutions providing guidance on PDF use/avoidance. Almost certainly, more institutions recommend avoiding or limiting the use of PDFs. The above consists of publicly facing content and many institutions keep technical documents in institutional repositories which are not indexed by search engines or available to the public.

Public bodies increasingly recommend HTML as the default format for publishing information. The UK and Welsh governments recommend HTML with strict caveats for PDF use (Williams, 2018; Welsh Government, 2024). In Ontario, Postsecondary Education Standards suggest PDFs should only be used if an “accessible alternative format is also simultaneously available” and indicate that any PDF used must meet PDF/UA standards (Postsecondary Education Standards Development Committee, 2021). The Government of British Columbia recommends that “publishing content on a web page rather than in a PDF should be the default.” While they acknowledge the need for a PDF may exist, “such as printing,” PDFs must be made as accessible as possible and “all PDF documents posted to the web should also be available in HTML” (Government Communications and Public Engagement, 2024). Avoiding PDFs as often as possible will reduce the need for alternative formats and remediation of content. Don’t invent a barrier.

“The best PDF will ever achieve is what HTML delivers. All it can do is catch up.” – Dr. Jonathan Godfrey

(Brinn et al., 2024)

5.2 How to create web content instead of PDFs

Providing content in HTML does not mean creators must become fluent and comfortable writing HTML. Modern content and learning management systems (Brightspace, PebblePad, WordPress, Drupal, Canvas, etc.) use ‘what you see is what you get’ editors that mimic the experience of writing in word processing software. Most web platforms support pasting content from a document into their page building tools to create a web page (refer to video 4 for a demonstration).

Learn how to make accessible Brightspace pages, accessible Word documents, accessible PowerPoint documents, and accessible PebblePad content. Additionally, learn how the WordPress Block Editor works and how to make accessible content in Pressbooks.

6 Recommended format by content type

There is no single best format, as different formats have been developed with different content and circumstance in mind. The problem is that the PDF became a catch-all for everything and, as we now know, the PDF has some problems. To provide guidance to more appropriate formats, the following lists common content types accompanied by a recommended format with a brief justification as well as a suitable secondary format when applicable.

While some of these recommendations may not match creators’ current workflows, the additional time and expertise required to make accessible PDFs warrant consideration and reflection on current practice. Recall that even if you have the time and expertise to make an accessible PDF, that file is not user friendly on mobile devices or usable by those that require larger font sizes. So, these recommendations may be outside some people’s comfort zones but will ultimately save significant time, prevent frustration, and immensely improve access and the reading experience for all.

Do not worry if you have more than your share of PDFs (312 per person, remember?) Do not try to convert every single one overnight. Rather, as you update your content or start a fresh course shell, start by ensuring your most essential information is in HTML or Word and then work your way through your old PDFs (odds are most of your PDFs started as Word documents, so if you still have those you’re halfway done). And, before you select File > Export > Save as PDF, I implore you to stop and consider if your content absolutely needs to be a PDF. Avoiding PDFs will make a significantly better experience for all your learners. Refer to the above recommendations for format suggestions. If you aren’t sure, email me lmcknight@langara.ca, I would love to talk about it.

It’s the people v. PDFs, which side are you on?

References