1.1 The Transformation of Historical Scholarship

This Element explores how this technological transformation has unfolded and what its impact will be. Yet it is not a ‘how-to’ guide.Footnote ¹⁰ This Element rather explores how this digital turn is changing historical scholarship and practice. The public – and even some historians – can sometimes see historical scholarship as ‘objective’. From this perspective, archives are understood as passing along to historians stories from the past from which they write history.Footnote ¹¹ Yet historians are of course influenced by their working conditions. Do they read documents on a screen or not? Can they access funding to travel to the archives that they want to? Do they have technical ability and knowledge to understand the knowledge systems they are using? Can they understand how to conduct keyword searches properly, do they grasp the underlying constraints of optical character recognition (OCR) in the database they are searching, or grasp what was or what was not digitized? Do they have children, constraining working hours or travel opportunities (making databases an especial godsend)? Are they in the developing world and accessing a website too overbuilt for their tenuous internet connection? All these mundane questions shape historical scholarship.

This transformation is neither wholly negative nor positive. Few revolutionary shifts are. Gains include rapid access to sources, quick fact-checking, the ability to search over decades and continents of historical sources, and – significantly – the ability to spend more time with family, teaching, and other duties and less time in faraway archives scribbling in notepads. The democratizing potential of these shifts cannot be ignored. Similarly, historians can use keyword search to amass large corpora of information. They now generally operate on new, larger bodies of information. Yet there are losses. Notably, historians can lose an understanding of historical context when keyword-searching directly to sentences or individual documents removed from their broader context. Historians are also using information retrieval systems that are not understood, the contents of which are in turn shaped by digitization bias. Similarly, this new scope of digitized research brings advantages and disadvantages. Search brings us many more results, but we still tend to evaluate research claims on a pre-digital level of sourcing.Footnote ¹² As historians now systematically explore thousands of articles with algorithms, perhaps our norms need to change and now require a half-dozen or more ‘hits’ to rise to the same level of significance we might have looked for in a pre-digital period. The way in which we evaluate scholarship via peer-review and scholarly assessment also need to take this new information ecosystem into account.

Sitting at a computer all day also represents significant change in how historians approach their research questions. We are perhaps losing the experiential knowledge of a place at a time when parts of our profession are underscoring the importance of community and place-based research. Yet there is no going back; the digital genie is out of the bottle. Whether this transformation is positive or not depends in no small part on the critical ardour with which historians approach their digitally transformed world.

Through careful and critical use of digital technologies we can ensure that this transformation is a net gain. Issues of context present the biggest challenges. Most digital systems are not designed to reveal context and are instead focused on keyword search. With better training and conscious digital research methods, however, we can countenance this to some degree. This may also require a collective change in the ways in which historians approach the level of citation needed to establish a scholarly claim, thanks to the larger amounts of information that we all increasingly operate on. With more user demand, more platforms can adopt an approach which facilitates context-aware reading, an approach adopted by the Internet Archive. Digital historians have also pushed the literature forward on the provision of context, informing primary documents through network analysis and trends.Footnote ¹³ Experiential knowledge also raises questions, requiring a deeper conversation about what we as a profession value. As a self-governing profession regulated primarily through peer review, we can collectively choose the direction that we want historical scholarship to go and what our values are.

This will require a transformation of our profession in four main ways, explored in this Element’s conclusion. These are a recognition of digital literacy’s importance, valuing interdisciplinarity, a prioritization of methodological discussions, and changing how we train future historians to incorporate new and emerging technologies. Despite the challenge before us, historians have a good foundation, especially vis-à-vis our use of context, by which to rise to these challenges.

1.2 The Structure of this Element

This Element brings the conversation around the transformation of historical scholarship together into a sustained micro-monograph. Its goal is to equip historians to be self-conscious practitioners in a digital age. The Element will do so through three substantive sections after this introductory one, followed by a substantive conclusion.

The second section, ‘Libraries and Databases’, explores how aggressive digitization, especially of newspapers and microfilmed resources, has created massive exploratory databases. Historians need to think about their construction and explore these platforms consciously. Algorithmic bias and selective digitization practices have comprehensively transformed how historians parse information. How has everyday technology transformed the work of historians, from 1930s microfilm to twenty-first century databases? We have gained dramatic access to primary sources, but historians need to ask questions about what has been digitized, and what has not? How has something been digitized? What are the impacts of copyright on these repositories?

The third section, ‘Archives and Access’, explores the long, intertwined existence between archives and historians with special focus on digital technology and source mediation. Historians are expert users of archives, although growing estrangement has led scholars increasingly to consider the interactions between the two parties as ‘interdisciplinary’ encounters.Footnote ¹⁴ This section thus explores how technology has changed the relationship between historian, archivist, and archives. What has the impact of partial collection digitization been? Online finding aids? Digital photography? Over the last two decades, historians are spending less time than ever before in archives, yet never have we had such powerful tools and platforms at our disposal.

The fourth section, ‘Publishing in an Interdisciplinary Age: From Journal to Social Media’, explores the changing relationship between historians and their audiences. When historians think of ‘digital history’, many think of digital public history, thanks to the historical profession’s long lineage of public-facing engagement.Footnote ¹⁵ Public scholarship has taken the shape of CD-ROMs, exhibit sites, ‘memory banks’, social media, and engagement on Wikipedia. Yet traditional career progression – hiring, and in North America, tenure and promotion – compel scholars towards traditional markers of career success as embodied in certain publication types (especially traditional books). How has technology changed publishing? This section briefly explores new formats, the changing approaches towards idea circulation, and the potential for interdisciplinary engagement.

Historians have rarely been transparent enough about many of the above topics: we treat scholarship as finished products, and methods are too often relegated to footnotes or informal discussions.Footnote ¹⁶ We could all do better history by reflecting on the ongoing technological transformation that is changing how we research.

1.3 Are We All Digital Historians Now? Digital History and the Digital Humanities

Scholars have been exploring how to explicitly use new and emerging technologies for humanistic and historical research since the 1950s and 1960s, under the auspices of what we today call the Digital Humanities or more specifically Digital History. The relationship between these two DHs is complicated. Digital History owes its lineage both to the Digital Humanities and also to currents and trends within the broader historical profession, particularly public history.

The Digital Humanities, broadly defined, explore the intersection of technology and the humanities.Footnote ¹⁷ In many ways, as Adam Crymble has explored at length, the Digital Humanities grows out of a digital literary studies tradition (by way of the Text Encoding Initiative and humanities computing scholars).Footnote ¹⁸ Historians have traditionally been underrepresented within the broad scope of the Digital Humanities and its earlier intellectual approaches.Footnote ¹⁹

Digital History can be expansively defined as the intersection of historical scholarship with new and emerging technology. In practice, this can be broken into two subfields. First, some Digital Historians use technology to reach new and different audiences with new media, continuing a lineage of historians using new media to carry out the mission of social and public history. An understanding of the democratizing potential of technology has been foundational to this approach.Footnote ²⁰ Secondly, other Digital Historians have used technology to do historical scholarship.Footnote ²¹ These scholars in part emerged from earlier approaches to quantitative history and historical demography – Social Science historians – as well as a smaller group of historians influenced by the more literary-focused computational studies approaches.Footnote ²²

Today, Digital Historians engage in a wide variety of scholarship. They may write programs or leverage computational platforms to assemble sources en masse, such as from records housed at the Internet Archive or the Library of Congress. Data are analysed by extracting features (i.e., word frequency, detected items in images, place names) before being subsequently analysed and visualized in a variety of ways.Footnote ²³ Other Digital Historians challenge conventional norms of scholarship, exploring new methods of scholarly communication. While it is impossible to do justice to this field in a few sentences, it is a vibrant subfield adopting explicit new methodologies and approaches.

Yet what about other historians who use computers to do their work but who do not fall into the above categories? Historians now all use databases, run keyword searches across millions of documents, and take digital cameras to turn scholarly documents into electronic files to be analysed at home. Are we now all digital historians?

There is a fruitful distinction to be made between Digital History – the delineated field of study bounded by academic journals, conferences, and pedagogical approaches – and ‘digitized history’, or the broader transformation prompted by technology. We are not all Digital Historians (my capitalization is deliberate). But we all engage with digitized sources and workflows. It is unwise to silo historical engagement with technology as a subfield given its sweeping impact on the entire profession. A focus on Digital History can make the rest of the historical profession think that ‘we are not digital historians’, as they fire up their web browsers and research over the Internet. We are all digital now.

1.4 The Digitally Aware Historian

Let me close this introduction on an optimistic note by imagining a digitally aware historian who takes the content of this Element to heart. What kind of work would a historian who was fully cognizant of the work that technology was playing in mediating their work do?

They begin to research their topic in a newspaper database. But rather than haphazard keyword searches, they instead look at which newspapers and years are included in the database – and which ones are not. They then explicitly consider selection bias (why was this newspaper digitized but not that one) and do some contextual research on the period’s newspapers. Perhaps at the end of the day they believe that the accessible advantages of the digitized newspapers outweigh the limitations of their selectivity bias, and accordingly write a few sentences to that effect in their introduction. This sets the tone for their engagement with sources throughout their manuscript. The historian thinks about what is present and what is missing. For an event, they go page-by-page for a few weeks before and after the event, developing a contextual sense of the source as well as to ensure that the OCR has not missed salient keywords. Scholarship is still assisted by databases, but every step is deliberate and thoughtful. When a button in the interface is clicked, it is done so deliberately. The database is no longer a black box.

The time then comes to work with other digitized primary sources, and many of the same questions come to the historian’s mind: what’s there? What’s not there? They respectfully email an archival colleague and ask these questions. The sources that are used are deliberately framed and contextualized. They understand their sources and are self-reflective about their use.

The historian then goes to the archive. They anticipate a follow-up archival trip, to follow theoretical rabbit holes that will inevitably arise when looking through their digitized photographs at home after their trip. As they leave the archive, they offer to share photographs with the archive as well in case it might help. After finishing their research, they tweet about their work, think about primary sources, and are transparent in their writing about argument, method, and approach.

Maybe they will become a Digital Historian. This could entail adopting new analytical lenses, moving away from the typical approach of ‘close reading’ to one of distant reading. Alternatively, they could draw on metadata to visualize interconnections between sources, or challenge norms of scholarship by publishing a database, a map, or another novel form of scholarly communication. New frontiers may then present themselves. But most historians will not go down this route. Instead, what we have seen in the above hypothetical of a digitally conscious researcher is one that pays heed to the mediating influence of technology on their work. The researcher understands that the way by which sources are mediated has had profound impact on how they are read, understood, and contextualized. In short, this is a digitally-aware historian, actively using technology rather than being shaped by it.

2.1 The Microfilm Revolution

Perhaps the best parallel to our contemporary moment is the early-twentieth century move to store documents on microfilm. Indeed, microfilm in the 1930s raised utopian hopes around universal access to all knowledge, as well as its long-term stewardship and preservation.Footnote ²⁵ Yet reception by historians was mixed. While many historians recognized the value of having sources conveniently microfilmed, others complained about eyestrain and research difficulties.Footnote ²⁶ An understanding of this earlier moment, especially given the roots of digitization in earlier microfilming projects, lays the foundation for thinking about libraries and databases today.

Microfilms are scaled-down documents on film strip. Documents are reduced to roughly 1/25th of their original size, and then viewed through microfilm viewers which both illuminate and magnify them to original (or larger) size. Microfilm is a beautiful system for information retrieval. Indeed, of all the rooms in a modern research library, the highest volume of analogue information is found in the microfilm room. Entire print runs of government documents, newspapers, print repositories, and dissertations from around the world are available for immediate consultation. A run of the New York Times from its 1851 founding to present would fill a small warehouse. In microfilm format, it occupies a shelf or two.Footnote ²⁷ Importantly, properly stored microfilm reels have a long lifespan, and they can be accessed with any combination of magnification and illumination.

Despite this power, in an age of instant search-and-retrieval, microfilm seems increasingly antiquated. Priceless cultural heritage sits on microfilm reels, inaccessible by contemporary standards. They are separate from where we expect to find information: the Internet. In 2021, after our library had been closed to on-site access at the University of Waterloo for a year owing to the global pandemic, I asked our library staff how many requests they had received to use the microfilm machines. The answer: one student, who happened to be doing her doctorate with me. Among over 40,000 students and over 1,300 faculty members, only one request for this treasure trove of information had been made. Times have certainly changed.

Microfilm originally promised democratic access to all information.Footnote ²⁸ While a nineteenth-century technology, it was in the early twentieth century when – thanks in part to the work of technologist Robert C. Binkley – microfilm began to be seen as a scholarly solution to the problems of needing to expensively travel to access scarce source information.Footnote ²⁹ In the 1920 and 1930s, the American Library of Congress microfilmed millions of documents held by the British Library, and brought them back to Washington, DC. An American researcher could now do research without crossing the Atlantic. During the Second World War, microfilm was used to facilitate transatlantic correspondence, as well as to courier secrets.Footnote ³⁰ Microfilm hit a conceptual pinnacle in 1945, when Vannevar Bush – the American inventor and then head of the United States’ Office of Scientific Research and Development – articulated his microfilm-driven ‘Memex’ machine in a famous Atlantic Monthly article.Footnote ³¹ The Memex, conceived in part to harness the revolutionary potential of microfilm-based information, would inspire the idea of hypertext – foundational to today’s World Wide Web.Footnote ³²

In other words, the advantages of making information accessible were clear. You can see what this looked like in Figure 1, a collection of digitized microfilm reels held by the Internet Archive that illustrates the production quality and scope. Researchers could go to their local university library or a major public library and consult millions of otherwise-inaccessible documents. Microfilm is a beautiful analogue system and led to a moment of massive excitement: the idea that mass amounts of information could be universally accessible.Footnote ³³ But what happened to the promise of microfilm? Two factors help explain why it fell a bit short of utopian projections.

Figure 1 The Internet Archive periodical collection

(https://archive.org/details/periodicals)

First, microfilm is difficult to use. If you have not used a microfilm machine before, imagine watching a long streaming video that you want to find a particular scene in. However, you can only press play, fast-forward, and fast-rewind. You might fast-forward too quickly, then you need to go back, and then forward, until you find the content you are looking for. Now, of course, imagine that these are thousands of individual pages that you are skimming through, and you can see how this would quickly frustrate. With more modern digital microfilm readers, while they are generally more pleasant to use, content is also often taken out of focus while one fast-forwards and rewinds, making it difficult to read headlines when the reel is in motion.

This problem was in part what underpinned Bush’s conceptual Memex. He imagined the revolutionary potential of being able to unlock the power of microfilm by quickly navigating to and displaying material of interest. The Memex envisioned the rapid retrieval and projection of microfilm within a user’s desk – one would make an inquiry and the material would whirl – and pop! – immediately appear.Footnote ³⁴ Yet these rapid retrieval systems for microfilm never came to fruition, now having largely been eclipsed by the digital turn.

Secondly, and perhaps as a portent of what would come later with digitization, microfilm was largely commercialized by a handful of big corporations. ‘All’ of the world’s information was not microfilmed in one utopian push, but rather microfilms were increasingly produced so that they could be sold to libraries – an important distinction. While early microfilm pioneer Robert C. Binkley saw the democratizing potential of microfilm – facilitating cheap, widely distributed copies and lowering publishing costs – salesman Eugene Power saw the commercial potential of microfilm and envisioned the market for selling entire periodical runs to libraries.Footnote ³⁵ Internet Archive founder Brewster Kahle notes that this ‘transformed small libraries into holders of collections that only the largest libraries could dream of’.Footnote ³⁶

Commercial imperatives and historical legacies dictated choices; however, something especially important today as yesterday’s microfilms often serve as the foundation of today’s mass digitization projects. This is perhaps best encapsulated by Kirschenbaum and Werner’s warning that the digital is a ‘frankly messy complex of extensions and extrusions of prior media and technologies’.Footnote ³⁷ As Stephen H. Gregg has shown in a recent Element on the history of the Eighteenth-Century Collections Online, this important database (published by Gale) owes its own provenance to decisions made earlier. It was a 1970s digital book cataloguing project that led into a 1980s–2000s microfilming initiative which then evolved into today’s online database.Footnote ³⁸ The same is true of periodical collections. Consider the case of Victorian periodicals. What we have today was shaped by forces including initial accessibility decisions, the impact of the London Blitz, and which were selected for microfilming in the postwar period. All of this together decides what is digitized today.Footnote ³⁹

The commercial and scholarly landscape for microfilming has naturally transformed in the digital age. While a boon for libraries, Power’s company – University Microfilms – was later acquired by Xerox, and then Bell & Howell, to become publishing behemoth ProQuest (in turn acquired by information conglomerate Clarivate in 2021). Databases became ProQuest’s priority, microfilm became yesterday’s technology, and the company’s massive array of microfilmed cultural heritage was neglected. Microfilm’s transformation from revolutionary technology to mundane background object helps obscure the pivotal role it played in democratizing and facilitating access to knowledge.

Even in the twenty-first century, the ability to go into the microfilm room and scroll through thousands of pages of documents continues to be a boon for scholarship. While microfilm revolutionized access by bringing documents to the scholar rather than requiring them to travel to the archives themselves, it did not dramatically change the ways in which scholars worked with documents. Microfilms are read page-by-page. A scholar advances the reels forward rather than flipping pages, but still has eyes on all of the information as it flows by. In this respect, microfilm was a less radical transformation than the keyword-search-based databases that followed.

Microfilm illustrates that the mediation of historical sources through technology has a long history: projects became feasible thanks to microfilm, and undoubtedly sources were selected depending on whether they had been put onto reels. The historical profession constantly engages with and is shaped by new and emerging technologies. In most cases, after an initial flurry of debate, new technologies become commonplace and unremarkable. Few historians had spirited debates about the role of microfilm by the middle to the late twentieth century – we just used it. Microfilm became part of the background of scholarship, machines used in the library basement without thinking, even as the reels profoundly changed the ways in which scholars chose and researched topics. The selection bias in what was microfilmed also played an important and often unrecognized role in shaping scholarship as it would serve as the basis of many digitization projects. Just because it became commonplace does not mean that its revolutionary impact was muted.

2.2 The Digitization Explosion

Two major features arrived in the late 1990s and early 2000s that dramatically changed the relationship between historians and primary sources. The first was the advent of optical character recognition and searchable historical databases, much of which was created by scanning microfilm en masse. The second was the advent of personal digital photography, which transformed archival trips from lengthier in situ experiences to short gathering missions followed by the actual reading of photographed documents at home. Combined, these two factors represent dramatic change in the day-to-day research workflows of historians.

Optical character recognition, or OCR, was a critical factor for allowing the use of digitized primary sources to proliferate. OCR made large bodies of historic documents accessible in new ways, ensuring that these large collections of text would become more discoverable than non-digitized ones.Footnote ⁴⁰ One could now keyword-search across hundreds of years and thousands of pages. OCR arose in a context of needing to make sense of large bodies of documents, such as in corporate legal discovery. Enterprising individuals realized that this technology could be used in other domains such as searching decades of historical newspapers. Throughout the 1990s, pilot projects were carried out (at Yale and Cornell) which explored the scanning of microfilm.Footnote ⁴¹ In 1999, however, R. J. (Bob) Huggins saw a business opportunity in scanning digitized newspapers.

Huggins, a Canadian entrepreneur, founded his company Cold North Wind to scan entire microfilm reels, run them through the OCR process, create a searchable full-text index, and make them accessible to fees-paying customers.Footnote ⁴² In 1999, high-speed internet access was not widespread, a precondition for sending larger images between locations. Huggins, however, anticipated the widespread revolution in high-speed internet access in the developed world and thus came online at just the right time. Thanks to this, the Toronto Star apparently became the first fully searchable and accessible digitized newspaper in the world.

Compared to Project Gutenberg’s earlier volunteer efforts, or even the digitization efforts of pioneering digital scholars and library-based digitization projects, Cold North Wind adopted a different approach. It represented a shift away from internal teams digitizing material themselves towards external vendors doing the work for them. If the roots of mass digitization had been laid in volunteer projects like Project Gutenberg or scholarly grant-funded activities, the story would have been different (engaging in hypotheticals: digitized holdings would be more accessible but rarer). Similarly, the use of OCR meant that high-resolution images were necessary, as the unreliability of automated transcription meant scholars needed to look at the original and not just a text representation. This made high-speed Internet essential for access, exacerbating an internet access divide between Global North and South as well as between well-served urban areas and the less-connected rural. Finally, as noted, the use of microfilm as the foundation of these projects is also significant. Given that certain material was microfilmed over other material, reflecting both the scholarly interests of past generations as well as commercial imperatives, these earlier biases were written into contemporary databases.

Furthermore, access to massive repositories soon facilitated scholarship but led to a patchwork of vendors (ProQuest digitizing these resources, Gale a different set of documents). Silos formed, antithetical to the broad approaches that historians tend to take with sources. It also made digitized culture especially vulnerable. While independent companies like Cold North Wind were a boon for scholarship, commercial success was elusive. Huggins was not alone in facing difficulties around sustainability and monetizing access to digitized culture: Google also shuttered its Google News Archive in 2011 after only three years. One exception to this has been the increasing role played by genealogy companies such as ancestry.com in digitization. Moving beyond census rolls to newspapers, in 2012 ancestry.com spun off newspapers.com, providing access to millions of digitized articles. While their core business may have originally been genealogists looking for birth and death notices, the OCR layer opened up hundreds of periodicals to researchers. Over at ancestry.com itself, a large array of material is digitized, moving beyond traditional census data to information such as ship and prison registries.

It all comes at a cost, however. To safeguard their intellectual property and investment, these growing bodies of digitized sources would be hidden behind legally enforced password-protected paywalls. What would be a minor frustration for institutionally backed scholars became a major barrier to those without institutional affiliation. External library borrowing status often did not include access to large and expensive databases, exacerbating the divide between richer and poorer universities. Microfilm had been a one-time purchase, open to all who could physically attend the library. In other words, the consumption of historical sources is now closer to a streaming model (think Netflix or Spotify) rather than a one-time purchase (like a DVD or music album). These databases brought ongoing subscription fees and additional barriers such as the necessity to have a high-speed internet connection.

Indeed, sites like ancestry.com or newspapers.com are somewhat democratic in that they allow individual researchers to subscribe. This makes them accessible in a way that the large publishing organizations and their institutional focus are not. Yet in some cases, the genealogical use case can collide with the historical one. The UK-based company Find My Past, for example, digitized the 1921 UK Census with The National Archives. They charge $4.90 USD for each image and $3.50 USD for each page transcript.Footnote ⁴³ This works for a user seeking a few relatives, but is unusable for any historical research of a larger scope.

The other factor which led to a dramatic increase in digitization was the advent of cheaper digital photography. While early digitization projects – such as Valley of the Shadow or Who Built America? – digitized (and in many cases transcribed) documents for broad use, most digital photographs taken of historical sources would soon become proxies for photocopies and disappear into private research collections.

In both respects, the ramping up of digitization which slowly began in the early 2000s would accelerate by the end of that decade as search engine behemoths contemplated what they would gain by having print material appear alongside born-digital results in search engines. Companies felt that it would enhance their search engines, help increase their user share, and have a ‘wow’ factor as well.Footnote ⁴⁴ Soon there were three book digitization and search coalitions – one led by Google, another by Microsoft, and a third by Yahoo! While Google emerged victorious, with the other projects being abandoned by their corporate partners (the Microsoft-led Open Content Alliance lives on in the Internet Archive’s Open Library project), the forty million Google Books titles represent a vision of what mass digitization could make possible.

There are three components of Google Books that have influenced how historians work: snippet search, Culturomics, and HathiTrust. The ability to search across forty million titles using ‘snippet’ search, when the book is under copyright but a few lines around a keyword result can be shown in many cases, facilitates fact-checking and snap decisions on whether a book is worth exploring in full.Footnote ⁴⁵ Additionally, works not in copyright can be downloaded in their entirety in several formats including PDF and ePUB. Secondly, the high-profile Culturomics project, launched in 2010 to fanfare, provides relative word frequency popularity over centuries of a substantial subset of the Google Books corpus and has increasingly become a staple of conference presentations and scholarship.Footnote ⁴⁶ Thirdly, HathiTrust emerged out of Google Books scans and illustrates the potential of a fully featured digital library.

Sitting in the Global North, of course, means that it can sometimes seem as if barriers to access have been uniformly lowered. Yet while Huggins had correctly predicted broadband access revolutionizing database access for North American academics, much of the world still lacks connections necessary to access these resources. Ever-growing technological requirements, from up-to-date software licenses to cutting-edge hardware, add to these barriers. The minimal computing ethos, driven by a pioneering group of digital humanists, aims to ask us to think of ‘computing done under some set of significant constraints of hardware, software, education, network capacity, power, or other factors’, with implications for increased access, less environmental impact, and ultimately for the development of more impactful digital projects.Footnote ⁴⁷ All platform designers could benefit from this approach.

2.3 The Implications of Digitization: What’s In and What’s Out

The sheer amount of digitized materials, however, obscures the reality that not everything is digitized. Most of it is not. Digitization is a resource-intensive process, both in terms of the scanning process and also in describing and making resources discoverable. Undescribed and thus inaccessible data are nearly useless. There are also considerable costs to preserving data in perpetuity. Accordingly, any user of digitized resources needs to ask: what has been digitized? What has not?

What is digitized? The holdings of well-funded institutions, from the Global North, are overrepresented. Even affluent institutions must make hard choices about what to process, balancing user interests and institutional priorities.Footnote ⁴⁸ Digitization thus proceeds unevenly and tends to favour richer countries and more popular collections. Conversely, specialist collections in less-resourced institutions tend not to be digitized. As archives aim to enhance the discoverability of their collections, they are integrated into search engines. Sources appear in Google results, and are cited in Wikipedia articles, with ripple effects on visibility. Critical reflection is necessary when thinking about things that are not digitized.

A 2013 exploration of these questions made a case study of Canadian newspapers and how relatively often they appeared in dissertations on Canadian historical topics between 1997 and 2010. The goal was to measure if a given newspaper’s usage increased after being digitized, and whether newspapers that were undigitized were conversely mentioned less. The findings were stark. In 1998, a pre-digitization year with 67 dissertations, the Toronto Star appeared 74 times. In 2010, once the paper was online, it appeared 753 times across 69 dissertations. Controlling for sample size, this was a 991 per cent increase. Similar trends were found with the Globe and Mail, another paper that was digitized early. However, the non-digitized Montreal Gazette and Toronto Telegram decreased by 16 per cent and 72 per cent respectively.Footnote ⁴⁹ This was the ‘Matthew Effect’ of digitized resources in action: the rich (digitized) got richer in terms of use, whereas the poor (un-digitized) got poorer. It was clear that Huggins’ and Cold North Wind’s digitization efforts had substantially reworked the ways in which Canadian historians carried out their research. Given the discoverability issues attendant with early-2000s OCR (a word accuracy rate of below 90 per cent was likely), historians were using tools they did not really understand and allowing it to reshape their scholarship.Footnote ⁵⁰

Historian Tim Hitchcock made a similar but broader point in 2013. He argued that historical work was being transformed without accompanying reflection:

As Hitchcock noted, the forces that we are seeing in the use of digitized newspapers are present in other places, such as Google Books views.

Finally, the uneven landscape of digitization is also influenced by copyright law and policy. The United States is a global behemoth with long (and occasionally growing) copyright terms.Footnote ⁵² There, 96 years must elapse between the publication or release of a film, book, or other copyrighted work before it enters the public domain. This profoundly impacts what is available in open repositories. Public-domain work can quickly proliferate across the Internet Archive and other repositories, whereas copyrighted material is secured behind controlled digital lending walls, paywalls, or is just plain accessible. Just as a digitized newspaper is more citable, material in the public domain is also more discoverable and thus more citable. We perhaps see this effect in citation patterns, where there is some (but not overwhelming) evidence that open-access papers receive more citations.Footnote ⁵³ In other words, historians need to understand mediating forces. This becomes more apparent as we dig into the details of digital objects themselves.

2.4 The Text Layer

With the overall contours of what has been digitized (and what has not) having been established, the next question is how historical documents go from being primary sources to being put into a database and made discoverable by keyword search. This question also has significant impact on historical research.

There are several ways that primary sources are made accessible for full-text search. First, they can be manually transcribed by a human – this works well for typewritten as well as handwritten documents, although it naturally cannot scale as easily due to resource limitations. There are only so many words a person can type in a day, and it is monotonous, detail-oriented work. In some cases, this has taken the shape of volunteers transcribing items they deeply care about. With Project Gutenberg, individuals selected books that were of personal interest and transcribed them. Starting in 2000, Gutenberg added an additional layer of crowdsourced proofreaders.Footnote ⁵⁴ Volunteers are finite and are more inclined to pursue projects of personal interest.Footnote ⁵⁵ More recently, a second form of volunteer labour has arisen: crowdsourcing. This is most successful in the case of corpora that might be most useful to a community if it has been made fully searchable in its entirety. For example, census records or immigration files benefit the genealogical community writ large if made accessible. Crowdsourcing has also enabled large-scale translation projects. For example, volunteers are translating over 70,000 French-language articles found within the Enlightenment Encyclopédie into English (in an interesting point of continuity, the original Encyclopédie was itself collaboratively written by over 140 different authors).Footnote ⁵⁶

Yet crowdsourcing brings ethical concerns such as whether it is fair to create large corpora without paying people for their time, and it skews towards volunteer interests.Footnote ⁵⁷ Finally, a project can pay for transcripts. If done property, a project can create an impeccable resource. The Old Bailey database, a repository of nearly 200,000 English court cases, used ‘double-entry rekeying’ to create part of its corpus. Two typists type, and if they diverge, a third individual comes in to resolve the conflict.Footnote ⁵⁸ This process creates nearly flawless transcriptions, but at high cost.

Alternatively, as we have seen in newspaper digitization, projects can turn to algorithms such as OCR. OCR, developed in the late 1970s and continuously refined, has a wide array of applications, from reading license plates to processing thousands of pages of corporate documents. Cold North Wind’s application of OCR to historical newspapers was only the beginning of a process which would see these algorithms applied to historical books, microfilm reels of vast arrays of documents, and beyond. Today, OCR is an active area of vibrant research throughout the digital libraries and information retrieval communities. While accuracy has improved, at the scale deployed in historical repositories, even a few infrequent character-level errors (a ‘n’ being misidentified as an ‘m’ for example) has dramatic impact. Simon Tanner has outlined what a seemingly exceptional 98 per cent success rate really means. This is a reasonable if optimistic figure when working with microfilm:

While some platforms allow for users to correct OCR errors that they find (such as the National Library of Australia’s Trove platform), most repositories do not allow you to make corrections, or to even see the underlying raw text that is being searched.Footnote ⁶⁰ OCR mediates text, adding an interpretive layer. In Figure 2, we can see an example of ‘good’ OCR – with high accuracy; in Figure 3, we can see an example of ‘bad’ OCR, where the algorithm has been confused by noise in the scan and tight, early twentieth-century periodical columns.

Figure 2 An example of high-quality, successful OCR from the Internet Archive

Figure 3 An example of lower-quality OCR from the Internet Archive

As in so many computational domains, machine learning – a form of artificial intelligence – suggests new approaches for automated text recognition. This promises the expansion of keyword search into historical domains previously imagined beyond its scope, such as the Medieval and Middle Ages. The Transkribus project shows how machine learning can help computers parse handwriting, previously largely beyond OCR’s scope due to its lack of standardization. A scholar ‘trains’ Transkribus to understand the writing of a particular scribe or author’s handwriting. In this way Transkribus learns how, for example, one author writes the letter ‘a’ versus the letter ‘b’, as opposed to how another author might handwrite those characters.Footnote ⁶¹ After several hundred pages of teaching Transkribus how to read handwriting, a ‘Handwritten Text Recognition’ model is trained and can be used on future pages. This technology is already being put to fruitful use. For example, a team has trained Transkribus to recognize Michel Foucault’s handwriting and is now creating a searchable repository of his profligate handwritten corpora.Footnote ⁶²

Understanding how the text layer is constructed is thus critical for a scholar to understand how to thoughtfully use a database. Is it double-entry text? In theory, it will be nearly 100 per cent accurate. OCR from the 1990s? A scholar will need to do more contextual skimming. Unfortunately, the platform layer often inhibits this.

2.5 The Platform Layer

While some interfaces, such as the Internet Archive, promote skimming and context when exploring collections, most do not. Slow loading times, page-by-page PDF downloads, and results lists all convey that a platform is designed to be primarily searched via keyword rather than systemic skimming or reading. This approach brings risks as the following example demonstrates.

In August 1938, American President Franklin D. Roosevelt visited Queen’s University in Canada and gave a speech pledging American support for Canada should the country be threatened.Footnote ⁶³ In 2021, I was giving a lecture at Queen’s and wanted to use his campus visit as an example of finding information in newspaper databases. To my surprise, a search for ‘Kingston’ and ‘Roosevelt’ in 1938 led to no results in a search of the Toronto Star. Was Roosevelt’s visit not significant? Was the OCR faulty? Digging into the newspaper, hosted on ProQuest Historical Newspapers, produced the discovery that there were no results from August 1938; indeed, the month was entirely missing. Was this a result of late Depression-era austerity, a publication stop? It seemed unlikely, as in early September letters to the editor referred to articles published only a week before. This was worrisome as it was the first event that I had searched for. If a user just keyword-searches their way through this historical collection, they would not know what was missing. A user could interpret the lack of hits as a null response. They would think of it as a true negative, whereas it might be a false one.

Perhaps, as we used to do with microfilm by necessity, we should return to skimming? ProQuest makes this difficult. A user needs to load a page of the newspaper (via drop-down menus, selecting year, month, and then year). The resolution is too low to read, meaning a PDF needs to be downloaded or manipulated in the browser. Click to download. Click to zoom. Read. Click to the next page. Wait for it to load. And so forth (and then, frustratingly in ProQuest, when you need to select the next issue, you sometimes need again to select year, month, and date). ProQuest is designed for keyword search, with the interface compelling you to access data in that manner. Yet the keyword search is a black box. There is no indication of OCR quality or missing data.

These problems are present in non-profit environments as well, mostly due to copyright. Consider HathiTrust, a repository of digitized books and primary sources. Throughout COVID-19, HathiTrust gained recognition for its Emergency Temporary Access Service (ETAS) which allowed member research libraries to allow their borrowers access to digitized books which they themselves had in their (temporarily inaccessible due to campus closures) physical holdings.Footnote ⁶⁴ This was an essential service for scholars who could now read digitally what they could not physically access. In doing so, however, they confronted the reality that skimming on HathiTrust was different than reading a print book. For copyrighted content, there was no download to PDF option. A user again faced load times, albeit far quicker than ProQuest. For slower, line-by-line readers, the interface worked, as it did for those doing keyword searches (although the text layer was again a black box in copyrighted works). Yet skimming was difficult. Researchers were compelled to keyword-search on large bodies of OCRed text, primarily due to copyright. While the interface appears to have improved by the pandemic’s waning days, this still underscores the pressures facing platforms that provide access to copyrighted material.

Enter the Internet Archive. Thanks to its emphasis on public domain materials, the Internet Archive has taken an open approach which empowers researchers to engage with documents and books in various ways. The interface is flexible: one can skim in their web browser, or can download documents either as a PDFs, images, or – importantly – plain OCR text (which is useful to see how messy the text layer is). It is a platform that facilitates research in all of its forms and presents a vision of what could be possible absent copyright and profit motive. It illustrates possibilities.

ProQuest, HathiTrust, and Internet Archive cover a spectrum, underscoring the degree to which copyright and platforms shape our approach to accessing knowledge in the digital age. Yet covering several platforms is important as historians usually require information from across many platforms in a single project. They may locally download from one website, access another through a different paywalled platform, and may be drawn to keyword-search a third. Accordingly, an astute understanding of these interfaces and how they shape our research is critical.

2.6 Digitization’s Impact

The impact of digitization on scholarship is hard to measure. Historians are largely opaque about their methods. This is not due to nefarious motives but rather disciplinary norms. Frustratingly, citation practices cite the source sans mediating platform.Footnote ⁶⁵ For example, historians cite the newspaper article but not how they accessed it. While archival citation is better, historians still often cite documents as if they were found in person even if they were accessed online (a bane for those who digitize documents and seek to measure impact).

Despite this citational opacity, changing technology and digitization has affected projects and research questions. Students and faculty can carry out previously impossible projects thanks to their ability to reach quickly and inexpensively across oceans and time. At my Canadian university, where we have shorter, one-year long master’s degrees, students can carry out thesis projects without physically entering archives, despite needing to base their research in primary sources. Digital photography means that archival trips are quick, surgical strikes.

Despite the unevenness of the digitized source base, however, it has had considerable implication on the kinds of questions that historians can explore. Instead of having to go deep into one or two newspapers, we can go shallow across dozens – bringing the ability to connect disparate points of information together. This has been a boon for genealogists, as they can find traces of an individual across previously disconnected sources. International connections can be quickly drawn as well.Footnote ⁶⁶ New topics of broader scope can now be pursued. For example, one can easily trace the evolution of public sentiment in a dozen newspapers (and, indeed, these are now questions routinely explored by undergraduates in term papers). Yet these newfound powers need to be better paired with interpretive frameworks. Scholars need to think deliberately about where they are searching and which results to draw on, especially as search results can be drowned out by duplicates. They may also lack the understanding of an article’s contextual placement within a periodical.

Lara Putnam examined how source digitization enables transnational and global history scholarship. Previously the domain of senior researchers, who had amassed a body of knowledge after a career of scholarship, such sweeping work is now routinely done by any scholar with pertinent research questions. They can quickly use Wikipedia, HathiTrust, Google Books, or other digitized repositories further to investigate an interesting archival discovery. Imagine: a researcher can sit in a New York City archive, find something of interest that happened in Paris, and quickly Google their way to a few (certainly unrepresentative given digitization bias) sources. Putnam considers these ‘side glances’ as having dramatic impact. As she notes, the ‘impact of such side-glancing – formerly rare, as each glance would have demanded hours or days of effort with no likely return; now quotidian, requiring nanoseconds to search and minutes to read – has been profound’.Footnote ⁶⁷ Ultimately, according to Putnam, while technology ‘has exploded the scope and speed of discovery … our ability to read accurately the sources we find, and evaluate their significance, cannot magically accelerate apace’.Footnote ⁶⁸ In our own fields we know what has been digitized and what has not been, but this level of critical engagement cannot be extended to every field that we incidentally explore on the Internet.

The digital turn is thus transforming scholarship in three respects. The first is geography, as global projects – at least those drawing on repositories in the Global North, given the costs of digitization – are now possible in previously impossible ways. The second is digitization bias, the ‘Matthew Effect’ of historical sources. The third is the transformation in the way in which sources are used, a shift from contextually aware skimming to surgical keyword search. An understanding of these forces can mitigate the negative effects of these changes. These important concepts cannot be left to be magically solved by the next generation of scholars, but rather need to be actively developed. Our current professional apprenticeship model fails to capture technological shifts, revisioning how we train our students.

What sorts of knowledge must historians acquire in order to properly contextualize their use of digitized primary sources? The first is understanding and being aware of algorithmic bias. Scholars, both to be good historians and citizens, need to think critically about the role that search engines play in their work (and life). Why is one website, for example, the first hit on Google (and thus accordingly cited) whereas another is relegated to the ninth page of results (unlikely to be seen)? What has been digitized and how has it been made discoverable? What voices and perspectives are reflected in digitized materials, and which ones are absent? How was a database constructed? In short, on what information is the scholar basing their arguments on?Footnote ⁶⁹

The second factor historians need to understand is source mediation and context. How a document is mediated matters as much as its content. A newspaper article read in its original form, or in a clipping file, or via microfilm, or via ProQuest keyword search are all mediated differently – and that matters. This needs to be built into our citation practices as everyday transparency. At the very least, citing the trails that we follow reveals these decisions and spurs reflection around whether the medium was influencing the way in which historical knowledge was constructed.

Similarly, the role that search engines play in shaping scholarship is significant. As Ted Underwood has noted, this is not just the shifting level of evidence that might be required to sustain a query – it’s rather the fact that a researcher already has a thesis when they enter the keywords with which to uncover. As he notes, researcher’s ‘guesses about search terms may well project contemporary associations and occlude unfamiliar patterns of thought’.Footnote ⁷⁰ This is compounded by the lack of context on search results that historians provide when we cite our findings – these critical dimensions are largely left uncited and not discussed. Was a source a needle in a haystack, or was it chosen to be representative – and from where was it cited? The platform layer needs to be made more visible in historical scholarship. Jo Guldi argues that scholars must adopt a ‘critical search’ methodology. She notes that ‘[c]ritical thinking about the words that supply a digital search lends strength and rigor to our research process … Iterative approaches and multiple tools are essential for controlling for the scholar’s own subjectivity in encounters with the archive’.Footnote ⁷¹ By transparently documenting choices, research is strengthened and made (somewhat) replicable. Finding the right balance between transparency and overwhelming a reader can be difficult. Yet providing context of search results, a sense of the relative prominence of ‘hits’ within a database, and other relevant information helps make scholarship intelligible. Databases and interfaces may be completely different and unrecognizable in five or ten years, meaning a historian must always write with this future audience in mind.

Finally, historians need deeper and more substantial digital literacy skills. We need to dismiss the misleading cliché of the ‘digital native’, and realize that these are actual skills that need to be taught.Footnote ⁷² Unlike our general approach to palaeography and language acquisition, often driven by project-specific needs, the ubiquity of search boxes and digitized documents means that such awareness needs to come early in the historical curriculum (or, arguably, in the base liberal arts curriculum of the twenty-first-century university). By recognizing that all scholarship has been transformed by these forces, we can begin to see that digitized history underpins our contemporary profession whether or not we choose to be Digital Historians.

2.7 Conclusion

Digitized sources will continue to dominate and shape the historical profession. As historians get a deeper understanding of this shift, they will hopefully look back on the first quarter of the twenty-first century as an aberrant period of unreflective digital practice. With more training and attention paid to the mediating influence of platforms, historians need to go down a mental checklist and consider the roles of mediation, algorithmic bias, and context both when they write, read, and evaluate scholarship.

This requires an openness to exploring new forms of scholarship, including moving away from narrative-centric books and articles and thinking broadly about the role of argumentation in the digital age. One of the more important historiographical interventions of the last decade was the ‘Digital History & Argument White Paper’, a collectively written document by a group of twenty-four historians looking at digital scholarship, argumentation, and – most importantly – the role of the discipline. Aimed at digital scholars, many of its central points are essential reading for all historians. Consider:

We need to be conscious and understand how our publications and arguments come together, not simply in terms of content and argument, but mediation. And, once understood, we need to write about it in our work.

3.1 Changing Work in the Archives

Archives have constantly evolved over the last century. In the 1920s, archives were understood through the ‘custodialist’ model, a gendered vision of archivists as the ‘handmaidens’ of history.Footnote ⁸¹ This was an idealistic view of archivists as neutral actors who transmitted documents from the past to the present for analysis.Footnote ⁸² In some ways, this was mirrored the then-prevailing empiricist ethos of academic historians, where a historian was understood as being able to reconstruct the past as it essentially was through hard work and conscientiousness.Footnote ⁸³ Yet, just as this objectivism was always a ‘noble dream’ for historical research, archivists have always had to make choices about what to collect and what not to.Footnote ⁸⁴

The ‘post-custodialist’ archival theory turn of the 1970s explicitly recognized the activist role played by archivists in shaping the historical record. One of the leading theorists of post-custodialism, Wisconsin State Archivist F. Gerald Ham, argued that archives were now in an ‘age of abundance’ and that archivists needed to thus recognize their agency.Footnote ⁸⁵ The early advent of digital records drove this in part, as did the broader growth of bureaucracies. Unfortunately, while archival theory moved into new directions, historians and archivists underwent a professional divorce. In 1975, for example, the Association of Canadian Archivists emerged from its previous professional home as the Archives Section of the Canadian Historical Association.Footnote ⁸⁶ Archivists were correctly conceptualizing their facilities as sites of active engagement and construction, whereas historians tended to see archives as static places. Even today, as Alexandra Walsham has noted, historians can understand archives ‘as neutral and unproblematic reservoirs of historical fact’.Footnote ⁸⁷ This divide forms the context for the transformation of historian’s archival work in the digital age.

3.2 The Siloing of Knowledge

Responding to the high cost of digitization, huge swaths of cultural heritage have been effectively privatized. Any professor at a research university will see this firsthand. A ‘free trial’ will be announced by a university library, or a salesperson will email a faculty member, offering them access to a cutting-edge array of primary documents and vendor-specific platforms through which to analyse them. Yet nothing is free. Costs for these platforms are beyond the reach of individuals, and the goal in many cases is to pressure an institutional library to subscribe to them. Indeed, the faculty member would rarely even know the cost of these platforms. Rather than one-time purchases, these are usually ongoing costs. For example, the Wiley Digital Archives platform has invaluable records such as the British Association for the Advancement of Science, the New York Academy of Science, and the Royal College of physicians (1200–1970).Footnote ¹⁰⁸ Some of these services are very sophisticated. Gale, another vendor, offers their Digital Scholar Lab to carry out transformative digital scholarship – yet only on material to which a library subscribes.

This has led to a siloing of digital archives, problematic given the needs of scholars. Figure 4 demonstrates this in terms of what a historian wants. Yet, with vendor and platform silos, what historians confront is seen in Figure 5.

Figure 4 What scholars want

Figure 5 What scholars get

Each interface is different, leading to an upfront learning curve. And, of course, this is just digitized sources, as there are of course in-person sources to also explore. We have a landscape of scattered silos. This mitigates against both conventional and digitally enabled scholarship, where vendor agreements end up shaping the resources used. Research questions are shaped by commercial interests, as opposed to fundamental questions. Fortunately, there is one inspiring model to look to: the Internet Archive. While I focus on the Internet Archive given its scale and broad utility, we can understand community-run tools and portals (such as the Native Northeast Portal) as offering similar potential.

The Internet Archive has been creating the largest collection of openly accessible digitized material: books, documents, microfilms, and beyond. Since its 1996 founding as the Internet’s memory bank, it has dramatically expanded in scope and size. In 2004, the Internet Archive began to digitize books – working with libraries to add scanned public domain works to their website – and shortly thereafter in 2005 was part of the founding effort behind the relatively short-lived Open Content Alliance (alongside Yahoo! and several research libraries; it was a public domain and open-access alternative to Google Books).Footnote ¹⁰⁹ Indeed, today many cultural heritage organizations use the Internet Archive to either freely host their collections, or contract them to physically scan and host their paper materials. In sum, the Internet Archive illustrates what almost-unfettered access to digital culture could look like.

The big-thinking nature of the Internet Archive is worth underscoring. In 2011, for example, Internet Archive founder Brewster Kahle announced his organization’s bold vision to digitize all of Iceland’s heritage and put it online (while evidently this did not get full Icelandic buy-in, a similar project collaborating with the Indonesian province of Bali – via their Bali Cultural Agency – soon followed with more success).Footnote ¹¹⁰ Today these efforts proceed thanks to a network of regional digitization centres which digitize material at low cost. Importantly, they look beyond the holdings of affluent research libraries in the Global North (one project is exploring how to build web archiving capacity around the world). With an eye towards copyright, the Internet Archive maintains their Physical Archive in California – owning physical copies allows them to legally loan them out. The statistics and holdings are impressive: 28 million texts, over six million videos, and 16 million recordings.

Critically, however, the Internet Archive also enables transformative digital scholarship. Documents can be browsed on the screen like a book and can be downloaded as PDFs to work with locally. Raw text can also be accessed directly (and in so doing can allow scholars to gauge the underlying quality of the OCR), and in some cases can even be downloaded as ePUBs. This leads to a flexible workflow: a user can download all the PDFs or plain text in a collection to work with it in the manner that they want. A scholar wanting to read just a few documents might do so on their web browser. In my case, I carried out a recent research project by using the Internet Archive’s programming library to download tens of thousands of PDF, text files, PowerPoint files, and images; I could then search it using my operating system’s search engine, and stitch them all together to explore much more conveniently than would otherwise have been possible.Footnote ¹¹¹

The ability to bring all this information together, in a variety of formats, forms an important counterbalance to the uneven digital landscape. It is a portrait of what might be possible absent the silos and barriers that we have thrown up between primary source collections and serves as a worthwhile counterbalance to the uneven digital landscape.

3.3 Conclusion

Given the excitement around the digitization of primary sources, it is important to underscore that most documents will remain undigitized by default. Painstaking archival research will likely always be necessary to some degree due to the varying nature of the historian’s craft. Decades from now, one suspects that historians will still be physically traveling to repositories to consult documents. Yet scholars know the limitations of the historical record in our area of expertise, making the great undigitized more worrisome when we ‘side glance’ (per Putnam) to other geographic or temporal domains with which we are less familiar.Footnote ¹¹² A scholar taking a peripheral glance at a relatively unfamiliar period or country may not realize the sheer expanse of what has not been digitized in that domain. Their lack of deep knowledge of a field prevents them from being self-reflective about archival silences. This means that whenever we carry out a database search for primary sources we must always ask: what is and is not here?

The way in which historians engage with archives has dramatically transformed over the last two decades. We consult digitized documents from around the world (unevenly due to digitization bias), and we go to archives and take thousands of photos to study at home. Significantly, we allow research to be mediated and shaped by platforms that impact how we answer and explain historical questions. Technology shapes the way we approach our research process, leading to bifurcated collection and analysis stages. This will accelerate over the coming years, as handwritten text recognition begins to make non-typewritten documents discoverable at scale. This acceleration will proceed unevenly, as contemporary issues and policies shape the record. To rise to these new challenges, we need explicit theorization and training. We will also need new publishing models to share our increasingly digital findings.

4.1 Why Historians Publish

Publishing, broadly defined, lies at the heart of what it means to be a professional historian. This interpretive work defines the discipline. Historians publish for many reasons, none mutually exclusive: a deep commitment to the field, engagement with communities, professional advancement, or (rather rarely inside the academy) monetary gain. Publishing allows a scholar to share knowledge, recover lost stories, reach an audience, and to strengthen their scholarship through critical engagement and conversation. In many cases this means that historians validate and strengthen their scholarship through a process of peer review. These are reasons why historians are drawn to publishers, including academic or trade presses. For the last decade, barriers to online and self-publication have been low enough that almost anybody could just put their own work on the web or self-publish. Yet, most historians realize they are not suited to the task of self-publishing. Self-publishing lacks external quality validation and peer review, necessary for the process of improving one’s work.Footnote ¹²⁰ Many authors also look for the professional recognition that comes with a publisher’s imprimatur.

Secondarily, historians publish because they are members of a self-governing profession that prizes certain types of publications. In other words, we publish not only in the service of history but also in the service of History. Despite advocacy around the expanding scope of what it means to be a successful historian, career milestones remain somewhat narrow and conservative.Footnote ¹²¹ Publishing operates in a prestige economy, where prestige is associated at least in part with the venue or press itself as opposed to the intrinsic value of the work. In some evaluation contexts, where the work cannot all be read by evaluators, the perceived selectivity of a publishing venue provides a proxy for the value of the work. While prestige and quality may be linked at the macro level – there is some truth to more prestigious venues tending towards more rigorous peer review – this link does not always hold true for individual books. In any event, the prestige of a press is not intrinsic, but rather ascribed by professional peers and networks (indeed, the relative prestige of a press varies dramatically between fields and subfields). Finally, publishers have their own incentives, notably the need for marketability (will a book sell well and recoup costs, or – even rarer – make money).

The complexities of publishing emerge at the nexus of author, profession, and press. Understanding the transformation of historical research in the digital age then requires attention to all three of these factors. An author moving on their own without the validation of professional or press recognition risks muting the impact of their work, but presses and professions rely on individual authors to effect change. All these factors are changing in this new age of digitized history.

4.2 Publishing in the Digital Age

Digital dissemination enables the diverse spread of ideas, using mediums such as blogs, social media, online visualizations and databases, online exhibits, and even innovative digital-first presses which lower bars to innovative scholarship. This has dovetailed with increasing attention being paid to the impact of publicly funded scholarship, such as the United Kingdom’s emphasis on ‘public impact’ in the Research Excellence Framework.Footnote ¹²² Yet this can often exist uneasily alongside disciplinary pressures to focus on traditional books and journal articles.

In Canada, historians have seen the transformative impact of this firsthand thanks to the Social Sciences and Humanities Research Council of Canada (SSHRC) funding agency. While other international funding agencies may have calls for knowledge translation or public impact (such as the National Endowment for the Humanities’ Digital Humanities programs, which often stress public engagement), SSHRC now embeds this in all of their funding calls. Historians may still want to publish a monograph as research outputs, but SSHRC’s emphasis on diverse and accessible ‘knowledge mobilization’ compels scholars to explore new methods of reaching audiences.Footnote ¹²³ These are usually digital. Applications are submitted with diverse ‘KM’ plans that often include blog posts, a podcast, or a website to provide intermediate access to research findings in advance of the monograph. While follow-through is somewhat limited owing to the lack of professional recognition for some of these outputs, it demonstrates the degree to which diverse outputs are increasingly at the core of funder’s visions for our (and other) discipline. There is increasing recognition that scholarly outputs should be accessible. Thanks to digital technology, historians are thus actively engaging publics through the web.

4.3 The Transformation of the Journal and Book

A journal reader might be forgiven for thinking that little has changed in the publishing world. Journal articles continue to be denoted conventionally by volumes, issues, years, and page numbers, and in most but not all cases can be downloaded as typeset PDFs. Sometimes the hyperlinks in these articles work, but not always. Indeed, individual journal articles often do not look dissimilar to those articles published decades earlier. Disciplinary pressures have contributed to this situation. Peer review is essential for the professional recognition of scholarship, typeset PDFs lend credence to publications, and evaluators of CVs often want to see page numbers. Similarly, while open access is an increasing part of the landscape, many journal publications remain closed: limited to institutional subscribers or those who might choose to pay for individual articles. ‘Gold’ open-access models, requiring authors to pay article processing charges to make a publication available to all readers, are an awkward fit with the modest funding ecosystem available to historians. These pressures are also present in the case of books: publishers take author manuscripts, transform them into typeset documents, and have them bound and published for sale.

Granting agencies, scholarly foundations, scholarly associations, and individual researchers have been spurring a slow transformation of book and journal publishing. MIT Press, for example, now publishes new monographs and edited collections as open access. They are fully downloadable for free, thanks to a combination of support from the Arcadia Fund as well as fees from participating libraries (who receive access to the otherwise closed back catalogue).Footnote ¹⁵⁰ In 2015, the University of Minnesota Press partnered with the City University of New York’s Digital Scholarship Lab to launch the ‘Manifold Scholarship’ platform. The Manifold platform allows scholars to host rich web-based books – often but not always as a complement to a print book. These rich digital editions facilitate reader interaction with primary sources and images. They allow readers to interact with the author and with each other through in-line annotations and commenting.Footnote ¹⁵¹ While the platform has been used by several publishers, the University of Minnesota Press also works with authors to publish iterative drafts of books. These publicly take shape, allowing authors to engage with readers as a draft is written.Footnote ¹⁵²

Journal and book peer review is also changing. Public participation in open peer review, where drafts are publicly posted and readers invited to comment, either informally through in-line comments or annotations or more formally through comprehensive reader reports, aims to produce stronger scholarship through productive open conversations. This process provides diverse perspectives on scholarship, more so than just the two, three, or four standard peer reviewers that evaluate most articles. This model has been around for a while now. In 2014, writing a Digital History textbook alongside my co-authors Shawn Graham and Scott Weingart, we posted the draft of our draft manuscript as sections were written online. These helped build confidence in our textbook – many eyes made for stronger work – although we did note in a retrospective for Perspectives in History that it took a bit of a thick skin. It was ‘one thing to receive [reviews] quietly in your office when a peer review arrives, but another to undergo the process in public’.Footnote ¹⁵³ Ultimately, our experiment with open review became one of the most rewarding elements of the book.Footnote ¹⁵⁴

Recently, even long-established journals have experimented with these new peer review approaches. In 2020, the American Historical Review hosted an open review of the article ‘History Can Be Open Source’ by Joseph L. Locke and Ben Wright.Footnote ¹⁵⁵ The review process combined two evaluation approaches. The journal editor solicited readers who would provide traditional reports and had those publicly posted. The journal also complementarily hosted an open review platform for paragraph-level comments on both the initial as well as the revised manuscript.Footnote ¹⁵⁶ Anybody could provide feedback on the manuscript. The dozens of comments are testament to an openness towards these new experiments in publishing.

Some recent ‘books’ are beginning to push at the boundaries of traditional publishing, forcing a rethink of what a ‘book’ in the digital age might look like. Stanford University Press has emerged as a leader in this space, rethinking digital monographs not just as enriched PDFs or websites but as fully digital publications. Elaine Sullivan’s Constructing the Sacred: Visibility and Ritual Landscape at the Egyptian Necropolis of Saqqara, winner of the 2020 Roy Rosenzweig Prize for Innovation in Digital History, exemplifies this type of publication. Designed as a digital object first, this argument-driven digital monograph contains videos, high-resolution maps, and even interactive 3D models where you can explore Saqqara, Egypt, and move a timeline slider to explore changes over time. The integral nature of the integrated digital visualizations means that any ‘book’ version of Constructing the Sacred would be a pale imitation.Footnote ¹⁵⁷

Journals are also adopting new forms in the digital age. In 2021, the Journal of Digital History launched articles which were essentially code-driven ‘notebooks’. They combined interactive code snippets and narrative prose. One can interact with a database while reading the article, for example, allowing a user to interact with the primary sources or information on the very same page as the article itself. This approach to ‘transmedia storytelling’ allows researchers to feature their methodological approaches through a hermeneutical ‘layer’, resulting in a new form of scholarship. While still a new journal, their first issue contained articles on text mining in newspapers, Twitter mining to understand public commemoration and computational explorations of Tacitus’ works.Footnote ¹⁵⁸

These new approaches present challenges. Sustainability is first and foremost of these. New forms of dynamic and interactive scholarship require new methods of digital preservation to ensure a book or journal is replayable in a decade or century from now. Workflows designed to preserve print scholarship or static PDFs do not always work with dynamic, interactive projects. At Stanford University Press, a years-long project funded by the Andrew W. Mellon Foundation witnessed a collaboration between the Press, a web archiving project (Webrecorder), and the Stanford Digital Repository to ensure the long-term preservation of digital books. Their project blog underscores the sheer difficulty facing even a very well-resourced team in ensuring the high-fidelity replay of these ‘books’ in the future.Footnote ¹⁵⁹

Past pitfalls illustrate the dangers of adopting digital publishing without sufficient attention to sustainability. In 2003, digital historian Roy Rosenzweig published an American Historical Review article entitled ‘Scarcity or Abundance?’Footnote ¹⁶⁰ Simultaneous to its publication, the journal hosted an online discussion forum. While it was promised to form part of the journal record, when the journal moved from the community-supported HistoryCooperative.org site to Oxford University Press, these discussions disappeared and were thus only serendipitously preserved by the Internet Archive.Footnote ¹⁶¹

The final challenges present in all of this as well are the financial challenges: book and journal publishers rely on purchases and subscription fees to recoup expenses (and make a profit in some cases to enable the cross-subsidy of less-commercially successful works). In the case of book publishers, this extends far beyond producing a print object, but include the work that goes into selection, peer review, marketing, editorial feedback, and beyond.Footnote ¹⁶² Open access requires different financial models to succeed. Perhaps the success of the MIT Press open access approach will spur broader sectoral change.

To end this section on a meta point, let me turn to forms of publications such as the one that you are reading: linear publications that take the shape of a book or journal article, but which take advantage of new opportunities presented by digital technology. There are a variety of these kinds of publications. Short monographs such as Cambridge Elements, Palgrave Pivots, Oxford Very Short Introductions, and beyond; 20,000 to 30,000 words, often consumed digitally rather than via print. This is made possible by the digital turn.Footnote ¹⁶³ While open-access publishing in traditional historical venues, such as books and journal articles, is relatively rare due to its cost and smaller grant agency support in the humanities, it is growing (especially in the United Kingdom where funders are increasingly mandating this). Naturally, we are now seeing the rise of digital editions: dynamic digital versions to complement (rather than replace, in the case of Stanford University Press digital projects) traditional print versions, although experiences have been uneven.Footnote ¹⁶⁴

Challenges aside, it is clear that traditional forms of the ‘book’ and the ‘journal article’ are beginning to be reshaped in new and interesting ways. Perhaps this represents a useful pathway forward for the profession. Such scholarship is intelligible to history departments, external evaluators, and deans, but draws on new and emerging technologies. Ultimately, the digital age will perhaps be most visibly demonstrated in the changing ways in which historians publish and share their research.

4.4 Conclusion

We are in a transitional stage of digital publishing: innovations of novel approaches are combined with the linear argumentation and recognizable form of a monograph or article. It can accordingly make a strong professional impact. Indeed, it is this calculus that lies in the Element you are reading: a linear argument in a recognizable form, but accessed and shared online.

Ultimately, the shape of how historians exchange our findings and engage with the public determines what history will look like in the digital age. It is exciting: historians no longer simply need to turn to a handful of presses and venues through which to publish work in the same typeset, linear fashion. Yet it is also a bit worrying: choice overwhelms, and the variety of mediums means that their reception amongst the public and the profession are uncertain. The final question then is: what can we do to transform the historical profession so that it can meet the challenges of the digital age?