Digital Humanities Part 4: Why We Need the Digital Humanities

This is Part 4 of a six-part series on the digital humanities. Part 1 asked what the digital humanities are. Part 2 surveyed tools for dissertation writers. Part 3 turned to theology and religious studies. This part makes the argument the earlier parts have been building toward. The full list of posts appears at the end.

For three posts now I have been describing the digital humanities. I have defined them, cataloged their tools, and traced what they offer scholars of religion. In this post I want to do something different. I want to persuade you.

Here is the claim. The humanities, the universities that house them, and the publics they serve need the digital humanities. Not as a garnish on real scholarship. Not as a grant-friendly buzzword. As a necessary part of how humanistic knowledge gets made, taught, shared, and defended in this century. I will make that case on five fronts: the state of the cultural record, the publics our scholarship should reach, the way we teach, the institutions we build, and the AI moment we are living through. Then I will take the strongest critics seriously, because a case that only works against strawmen is not a case.

One note on terms before we start. In Part 1 I offered a working definition: digital humanities are a contested ensemble of methods, infrastructures, pedagogies, institutions, and critical practices through which humanists create, preserve, model, analyze, publish, and critique cultural evidence in digital forms. Keep that definition in view. The argument below is not that every humanist should learn to code. It is that the humanities as a whole need people, projects, and institutions doing this work, and need them badly enough that the rest of us should care whether they thrive.

The Record Has Already Moved #

Start with the evidence itself, because everything else follows from it. The cultural record that humanists study has moved. Much of it has been digitized, and an enormous and growing share of it was born digital and has never existed anywhere else.

The numbers are hard to take in. Google has scanned more than 40 million books since its library project began.^[1] HathiTrust, the academic digital library assembled by research universities, holds over 18 million digitized volumes.^[2] The Digital Public Library of America aggregates more than 54 million images, texts, videos, and sounds from American libraries, archives, and museums.^[3] And the Internet Archive's Wayback Machine reached a milestone in late 2025 that would have sounded like science fiction a generation ago: one trillion archived web pages.^[4]

Roy Rosenzweig saw this coming before most of his colleagues did. In 2003 he warned in the American Historical Review that historians "may be facing a fundamental paradigm shift from a culture of scarcity to a culture of abundance."^[5] The traditional training of a humanist assumes scarcity. You learn to squeeze meaning from the few sources that survive. Abundance breaks that training. When the sources number in the millions, close reading alone cannot even tell you what is there. You need methods for searching, sampling, modeling, and visualizing at scale, and you need the critical judgment to know what those methods hide. That combination, method plus judgment, is exactly what the digital humanities cultivate.

The born-digital record raises the stakes further. The correspondence of this century is email and chat logs. Its pamphlets are blog posts and social media threads. Its congregational newsletters, sermon archives, and denominational debates live on websites and platforms that can vanish overnight. The Library of Congress spent seven years attempting to archive the full public Twitter firehose before concluding at the end of 2017 that it could only collect selectively going forward.^[6] Think about what that decision means. One of the best-funded memory institutions on earth looked at a single platform's output and blinked. The record of contemporary life is being created faster than any archive can keep it, on infrastructure that scholars do not control.

If you study religion, this is not an abstract problem. The religious life of the last thirty years happened substantially online. Prayer requests, fatwa databases, papal tweets, deconversion testimonies on YouTube, pandemic-era streamed liturgies. A historian of American religion writing in 2080 about the 2020s will need digital archives the way historians of the Reformation need printed pamphlets. Whether those archives exist, and whether anyone knows how to interrogate them, depends on choices being made right now, mostly by people who are not humanists.

And digitized does not mean transparent. Every digital archive you search is an interface hiding a thousand decisions. What got scanned first and what still waits. How the OCR handled blackletter type, or Arabic script, or a water-stained page. What the metadata calls a "sermon" and what it calls a "speech." When you type a term into a database and get four hundred hits, you are not seeing the past. You are seeing the residue of digitization budgets, cataloging choices, and search algorithms, and unless you know how to ask what the interface conceals, you will mistake the residue for the record. Training scholars to ask exactly that is a core competency of the digital humanities. It is source criticism, the oldest humanistic skill there is, extended to the infrastructure our sources now live in.

That is the first front of the argument. The record has moved, and scholars who cannot work with digital archives, platforms, and datasets are increasingly cut off from the evidence of their own fields. This is not about fashion. It is about access to sources, which has always been the beginning of humanistic work.

What We Cede If We Look Away #

Suppose the humanities collectively decided the digital turn was someone else's business. What would happen? Nothing dramatic at first. Monographs would keep appearing. Seminars would keep meeting. But ground would be ceded, quietly and steadily, on three sides.

First, the interpretation of the digitized record would fall to people without humanistic training. Search engineers, platform designers, and data scientists already make consequential decisions about what parts of the cultural record are findable, how texts are classified, and which materials get prioritized for digitization. Those are interpretive acts. Every metadata schema embeds a theory of what matters. Every search ranking is an argument about relevance. When humanists are absent from that work, the theories embedded in our infrastructure go unexamined, and the record gets shaped by commercial logic alone.

Second, the tools of cultural analysis would be built without us. Text mining, network analysis, and machine learning are being applied to literature, scripture, and historical archives whether or not humanists participate. The question is only whether the people applying them understand the materials. A model trained on digitized newspapers will happily produce findings about the nineteenth century. Whether those findings survive contact with what historians actually know about newspaper reprinting, OCR error, and archival gaps depends on having scholars in the room who know both the math and the archive.

Third, and this one stings, the public conversation about culture would continue drifting away from the academy. People are not waiting for us. They research their genealogies, transcribe old letters, build fan wikis, and argue about history on platforms every day. The humanities can meet that energy with open projects and public scholarship, or it can watch from behind the paywall. The digital humanities, more than any other part of the academy, has spent thirty years building the meeting places.

Which brings me to the strongest evidence I know that this work pays off.

The Proof at George Mason #

If you want to see what the digital humanities give the public, look at one center's track record. The Roy Rosenzweig Center for History and New Media at George Mason University was founded in 1994 by Roy Rosenzweig, a labor and social historian with a conviction that has held up remarkably well. The center states its mission plainly: "We use digital media and computer technology to democratize history: to incorporate multiple voices, reach diverse audiences, and encourage popular participation in presenting and preserving the past."^[7] Rosenzweig died in 2007, and the center was renamed in his honor in 2011. It has now spent more than three decades making good on that sentence.

Consider what has come out of one center at one public university. Zotero, released in 2006, is free, open source software for collecting, organizing, and citing research sources.^[8] If you have written a dissertation in the last fifteen years, there is a fair chance Zotero managed your footnotes. It did not come from a software company. It came from historians who decided that scholars deserved a research tool they could own, extend, and trust, and it has saved researchers an incalculable amount of money and grief. Omeka, the center's web publishing platform, did the same thing for exhibitions and collections.^[9] Small museums, local historical societies, libraries, and classrooms use Omeka to put their holdings online with real metadata standards behind them, no vendor contract required.

Then there is the September 11 Digital Archive, which the center built with the American Social History Project at the CUNY Graduate Center in the immediate aftermath of the attacks. The project collected first-person accounts, emails, photographs, and other digital materials from the public, tens of thousands of items from ordinary people, and in August 2003 the Library of Congress formally accepted the archive, describing a collection of more than 130,000 written accounts, emails, recordings, and images.^[10] Read that again. Two years after the event, a university center had gathered a people's archive of a national trauma and placed it in the national library. No commercial platform did that. No traditional archive could have done it at that speed. It took scholars who understood both historical documentation and the web.

The pattern has continued long past those famous examples. The center also gave rise to Tropy, a desktop application for organizing the thousands of archival photographs researchers now bring home from reading rooms, and PressForward, a tool for curating and publishing scholarship from the open web, both now developed by the center's spinoff, Digital Scholar.^[7:1] Notice what unites the whole catalog. Each tool began as a real problem in a working historian's life, got solved in the open, and was given to everyone with the same problem. That is a research agenda disguised as a software shop, and it is the inverse of how commercial academic software gets made, where the problem solved is the vendor's revenue and the scholar is the product's captive.

Rosenzweig himself named the principle underneath all of this. "Democratized access," he wrote, "is the real payoff in electronic records and materials."^[5:1] The center he founded is the standing demonstration that the payoff is real. Free tools used by millions. Public archives of national significance. Training for generations of digital historians. When someone asks you what the humanities have done for the public lately, RRCHNM is a very good answer, and it is an answer that exists only because a university invested in the digital humanities early and kept investing.

One Scholar, Working in the Open #

Institutions matter, but so do individual examples, and I want to give you one because it answers a doubt I hear often from graduate students. The doubt goes like this: sure, big centers can do digital work, but I am one person with a dissertation and no budget. What difference can openness make at my scale?

Lincoln Mullen is a professor of history at George Mason University and now directs the Roy Rosenzweig Center for History and New Media.^[11] Before he held any directorship, though, he was a scholar of American religion doing something that was still unusual: publishing his code, his data, and his methods alongside his arguments, all in the open, from graduate school onward.

Look at how his two major projects fit together. The Chance of Salvation, his history of religious conversion in America, appeared from Harvard University Press in 2017 and won the American Academy of Religion's award for best first book in the history of religions.^[12] That is the traditional path, done well. But alongside it he built America's Public Bible, a computational study that tracks biblical quotation across millions of pages of digitized American newspapers, identifying which verses were quoted, when, and in what contexts. Stanford University Press published it in 2023 as an interactive scholarly work, a monograph you read by exploring it.^[13] And the entire apparatus behind it, the code that detects the quotations, the data, the site itself, sits in a public repository for anyone to inspect, reuse, or challenge.^[14]

That last part is the point. Mullen's methodological claims are checkable in a way most humanities arguments are not. If you doubt his quotation detection, the code is right there. If you want to run his methods on Canadian newspapers or on the Quran instead of the Bible, you can start from his repositories rather than from zero. He has also released a string of R packages that other historians now use daily, tools for tokenizing text, detecting text reuse, inferring gender from historical names, and mapping historical boundaries.^[15] Each one is a piece of his own research process, generalized and given away.

I will not pretend the openness was free. Publishing your code means letting strangers find your bugs. Releasing your data means someone may scoop the article you meant to write from it. Building tools counts unevenly in tenure files that were designed for monographs, a real inequity the field is still fighting to fix. Mullen made those trades anyway, and the return on them is a career's worth of evidence that generosity is a viable scholarly strategy, not a sacrifice of one. His grants, his press contracts, and his directorship all came to a scholar whose defining habit is giving the work away.

One scholar, working openly, has changed how a field works. Historians of American religion now have a model for computational argument, a toolkit they did not have to build, and a standard of transparency to aspire to. Multiply that example across the disciplines and you begin to see what the humanities forfeit when open computational practice remains rare. This is why we need the digital humanities in the plural, as a community of practice. Openness compounds. Every released dataset lowers the cost of the next project. Every published method teaches the next student. Closed scholarship cannot compound this way, because there is nothing for the next person to build on except the citations.

Teaching in the Open #

The pedagogical case may be the least glamorous front of this argument and the most important. In Part 1 I argued, following Brian Croxall and Diane Jakacki's collection What We Teach When We Teach DH, that the digital humanities are learned through practice, in classrooms and workshops, through modules and metadata exercises and mapping assignments.^[16] Claire Battershill and Shawna Ross make the same point at book length for working instructors, and Johanna Drucker's Digital Humanities Coursebook turns it into a full curriculum.^[17] Teaching is not what happens after the real DH work is done. Teaching is the work, because a method that cannot be taught dies with its inventor.

The clearest demonstration of what open digital pedagogy looks like is Digital Pedagogy in the Humanities: Concepts, Models, and Experiments, published by the Modern Language Association in 2020 after years of open development. The project is organized as a set of curated keywords, 59 of them, each introduced by a scholar-curator and supported by concrete pedagogical artifacts. Syllabi, assignment prompts, exercises, lesson plans, student work. More than 500 artifacts in all, and each one is provided in a modifiable, openly licensed format so you can take it, adapt it, and use it in your own classroom this semester.^[18]

Sit with how different that is from the way teaching knowledge usually circulates in the academy, which is to say, barely at all. Most syllabi live and die on a single campus. Most brilliant assignments are never seen by anyone but the students who complete them. Digital Pedagogy in the Humanities treats teaching materials as scholarship, curates them with the seriousness of an edited volume, subjects them to open review, and then gives them away. The project was even reviewed in public before publication, with the draft keywords posted for open commentary.^[18:1]

If you teach, you can feel the force of this immediately. A new instructor assigned a digital methods course does not have to invent it alone. A theologian who wants students to build an online exhibit can start from artifacts that worked elsewhere. The commons model does for pedagogy what Mullen's repositories do for research. It compounds.

Picture the difference in an ordinary classroom. A religious studies seminar spends a unit building a small digital exhibit on local congregations. The students photograph buildings, interview members, write catalog entries, and wrestle with metadata. Is this building a "church" or a "former church"? Whose name goes on an oral history? What does the community get to veto? By the end of the unit they have learned description, provenance, consent, and audience, which is to say they have learned the working ethics of the archive, and they learned it by making something a stranger can visit. Now ask what the same unit teaches as a term paper. The paper is read once, by one person, and filed. I am not against term papers. I assign them. But a pedagogy that never asks students to make public knowledge is training them for a version of the humanities that speaks only to itself, and students can tell.

And notice what makes it possible. The collection is not open by accident. It is open because the digital humanities community made openness a professional norm, argued for it inside a scholarly society as traditional as the MLA, and built the platform to carry it. Which raises the question of platforms directly.

Infrastructure We Own #

Where does open scholarship actually live? For too many academics the honest answer is: on venture-funded platforms that exist to monetize them. Academia.edu and ResearchGate present themselves as scholarly commons, but they are for-profit companies, and their business models run on the attention and data of researchers. The paper you upload becomes bait for someone else's subscription prompt.

The digital humanities community built an alternative. Humanities Commons launched in December 2016 as a project of the Modern Language Association, funded by the Andrew W. Mellon Foundation, offering scholars profiles, group discussion, website hosting, and an open access repository called CORE where work receives a permanent DOI and joins the open scholarly record.^[19] In November 2020 its home moved to Michigan State University, and in 2024 it took the name Knowledge Commons, reflecting a membership that had grown well beyond the humanities.^[20] The network now serves tens of thousands of scholars and practitioners. Kathleen Fitzpatrick, the project's director and one of the most consistent voices for open infrastructure in the academy, describes it as open access, open source, and nonprofit, which is precisely the combination the commercial platforms cannot offer.^[21]

The distinction is not sentimental. It is about governance and survival. A platform owned by its scholarly community answers to that community. A platform owned by investors answers to an exit strategy. When a for-profit network changes its terms, locks features behind payment, or sells, scholars have no recourse and their work is hostage. When Knowledge Commons makes a decision, it does so as academic infrastructure, hosted at a public university, sustained by grants and institutional support, with its code in the open.

If you are reading this as a working scholar, the practical translation is simple. You probably have work that deserves readers and sits where none can find it. A conference paper, a dissertation chapter, a dataset, a syllabus. Depositing it in an open repository with a permanent identifier takes an afternoon and makes it citable forever, and doing that on community-owned infrastructure rather than a commercial network means your afternoon strengthens the commons instead of somebody's quarterly numbers. Small choices like this are how infrastructure survives. Platforms live on participation the way journals live on submissions, and every scholar who moves their work into the open commons casts a vote for the version of scholarly communication they want to exist.

Here is why this belongs in an argument about why we need the digital humanities. Infrastructure like this does not emerge from the humanities' traditional ways of working. It takes people who can think like scholars and build like engineers, who understand peer review and metadata schemas and server costs all at once, and who can persuade learned societies and foundations to fund the unglamorous work of keeping a platform alive. Those people are the digital humanities. Without them, open access remains a sermon. With them, it becomes a place you can actually put your work.

Building Capacity Inside a Public University #

Arguments about the academy can float free of any actual institution, so let me ground this one. Michigan State University shows what it looks like when a large public university decides to build digital humanities capacity deliberately rather than accidentally.

DH@MSU, the university's digital humanities program, describes itself as a community of faculty, librarians, staff, and students collaborating across departments and colleges, with a stated goal to "lead at MSU and on the international stage in enhancing arts and humanities research and teaching through digital methods."^[22] The program offers an undergraduate minor in digital studies and a graduate certificate that requires coursework, pedagogy, and a research portfolio, which means students can leave MSU with digital humanities training credentialed on their transcripts, not just picked up in the hallway.^[22:1]

The capacity runs deeper than the degree programs. MATRIX, the university's Center for Digital Humanities and Social Sciences, has been building digital projects since 1997, with a mission that emphasizes collaborative, sustainable, openly accessible work and a longstanding focus on the African diaspora and continent.^[23] LEADR, a lab run between the History and Anthropology departments, puts digital research methods directly into the undergraduate curriculum.^[24] The university hosts Knowledge Commons, as we just saw. And it recruited Fitzpatrick to direct digital humanities there in 2017, a hire that signaled the institution understood this was a field you invest in with people, not just servers.^[21:1]

Why does the institutional story matter for my argument? Because critics of the digital humanities often picture it as a fad imposed from above, a shiny object administrators chase. MSU shows the opposite pattern. Decades of patient, cumulative capacity building. Labs that teach methods to sophomores. Certificates that survive leadership changes. Infrastructure adopted from a scholarly society and sustained. Fads do not survive decades of leadership changes, budget cycles, and curriculum committees. This has. It is a public university treating digital scholarship as core academic capacity, the way an earlier generation treated area studies centers or university presses. Other institutions can do the same, and the ones that do will be the places where the next generation of humanists actually wants to train.

You do not need MSU's scale to act on the lesson. A small college can cross-list one digital methods course and give it a permanent number. A seminary can put a librarian with metadata training on every digital project from day one. A department can decide that a peer-reviewed digital project counts in tenure review and write that into its bylaws, which costs nothing and changes everything for the assistant professor deciding what to build. Capacity is not a budget line first. It is a set of decisions about what counts, who gets hired, and what students are allowed to make, and every one of those decisions is available to your institution this year.

The Argument the Humanities Need to Win #

Step back from the particular examples and notice the larger fight they belong to. The humanities have spent two decades answering the same question from legislators, trustees, parents, and deans: what are you for? The question is often asked in bad faith, and the best answers, that the humanities teach judgment, that a society without them forgets what it is, are true and rarely persuade the people holding the budget. Enrollments have slid. Departments have been merged and closed. You know the litany. I will not rehearse it further, except to point out one thing about every example in this post. Each one is an answer to that question which does not require the asker to already love the humanities.

A state legislator may not care about historiography, but a free tool used by millions of researchers is legible to them. A parent may not care about archival theory, but a public archive of September 11 that lives in the Library of Congress is legible to them.^[10:1] A trustee may not read monographs, but a platform that saves scholars from predatory publishing costs is legible to them. This is not selling out. The work is scholarship all the way down, and I have spent this whole post showing its intellectual seriousness. It is translation. The digital humanities produce public evidence of the humanities' value at a moment when the humanities are being asked for evidence, and no other part of our fields produces it at the same scale or in a form the public can hold in its hands.

There is a defensive version of this point and an affirmative one, and I want to be clear I am making both. Defensively, yes, digital projects give provosts and deans something to point to, and fields that give their institutions nothing to point to get cut. You do not have to like that logic to plan for it. Affirmatively, and this matters more, public scholarship is what many of us wanted from the humanities all along. Rosenzweig did not build free tools as a budget strategy. He built them because he believed the past belongs to everyone and acted on it.^[7:2] The happy fact is that the same work serves both purposes. It keeps the lights on, and it keeps the faith.

The Skeptics Deserve an Answer #

Now the hard part. In 2016, Daniel Allington, Sarah Brouillette, and David Golumbia published an essay in the Los Angeles Review of Books called "Neoliberal Tools (and Archives): A Political History of Digital Humanities," and it remains the sharpest attack the field has faced.^[25] If you have absorbed anything from this series so far, you owe it to yourself to read the case against.

Their argument, compressed: the digital humanities rose not because of intellectual merit but because the field fit the needs of a university being restructured on corporate lines. "By providing a model for humanities teaching and research that appears to overcome these perceived limitations," they write, "Digital Humanities has played a leading role in the corporatist restructuring of the humanities."^[25:1] The field, they charge, "has often tended to be anti-interpretive," preferring to "archive materials, produce data, and develop software, while bracketing off the work of interpretation."^[25:2] And they note, pointedly, that "Digital Humanities is pushed far more strongly by university administrators than it is by scholars and students."^[25:3]

I want to resist the reflex to wave this off, because parts of it land. It is true that administrators and funders have sometimes loved DH for bad reasons, as a photogenic answer to the crisis of the humanities that asks nothing of budgets for tenure lines. It is true that grant-driven project work can pull junior scholars into service roles without security or credit. It is true that a lab full of deliverables can crowd out the slow interpretive work that has no demo day. Anyone who has watched a university chase innovation theater knows the pattern the critics are describing. It is real.

But the argument fails as a description of the field, and it fails in three ways.

First, it mistakes the vulnerability for the essence. Every part of the university can be bent toward corporate logic, including the traditional humanities, whose adjunctified survey courses are hardly a refuge from neoliberalism. The question is not whether DH can be co-opted. Everything can. The question is what the work itself does, and the record I have laid out in this post, free tools, public archives, open pedagogy, community-owned infrastructure, is mostly a record of decommodification. Zotero replaced software that cost money. Knowledge Commons competes with platforms that harvest scholars' data. The September 11 Digital Archive put a people's history in the Library of Congress. If this is neoliberalism's advance guard, it has a strange way of showing it.

Second, the anti-interpretive charge does not survive contact with the field's own literature. The most prominent development in digital humanities over the past decade has been precisely the growth of its critical wing. Global Debates in the Digital Humanities confronts the field's Anglophone and Global North defaults.^[26] Elizabeth Losh and Jacqueline Wernimont's Bodies of Information brings intersectional feminist analysis to bear on data and computation.^[27] James Dobson's Critical Digital Humanities subjects computational methods themselves to sustained interpretive scrutiny.^[28] David Berry and Anders Fagerjord argue that the field's task is knowledge and critique in a digital age, not tool building for its own sake.^[29] A field allergic to interpretation could not have produced this shelf. The critics wrote as if DH's center of gravity were the software demo. Its center of gravity is now the argument about what computation does to knowledge, which is an interpretive question if anything is.

Third, the field answered rather than flinched, and the answers were often more interesting than the attack. Within days of the essay, responses came from across the field, and the exchange is worth reading as a whole.^[30] Andrew Piper, Juliana Spahr, and Richard Jean So replied in the same venue that they were "sympathetic to the observations that provoked this essay" while showing that other disciplines "have benefited greatly from the merging of critical and computational modes of analysis."^[31] Matthew Kirschenbaum's wry personal essay, "Am I a Digital Humanist? Confessions of a Neoliberal Tool," took the accusation apart from the inside.^[32] The pattern matters. A healthy field metabolizes its critics. The digital humanities took the hardest punch available and got more self-aware, more global, and more political as a result. That is not what a hollow administrative fad does. It is what a discipline does.

So keep the skeptics on your shelf. They are the field's conscience, and the labor critique in particular should be tattooed on every project budget. But the conclusion they wanted you to draw, that the humanities would be safer without the digital humanities, gets the situation exactly backward. The forces they fear, austerity, platform capitalism, the reduction of knowledge to product, are advancing on the humanities regardless. The digital humanities are one of the few places in the academy building the countervailing tools, archives, and institutions. Disarming ourselves is not resistance.

The AI Moment #

Everything above was true five years ago. The rise of large language models has made it urgent.

Machine learning systems are now trained on the digitized cultural record, the same scanned books, archived web pages, and digitized newspapers I described at the start of this post. Companies claim these models understand language, reason about texts, and can stand in for human judgment. Those are claims about culture, meaning, and interpretation. They are, in other words, humanities claims, made at industrial scale by people with no obligation to the evidence, and they demand examination by people trained to give it.

Who is equipped to do that examination? You need to understand how the training data was assembled and what it omits, which is an archival question. You need to understand what a model actually does with text, which is a methodological question. And you need to understand what is lost when statistical pattern stands in for meaning, which is an interpretive question. That combination is the digital humanities' home territory, and the most influential critiques of the current AI regime have come from exactly this borderland between technical fluency and humanistic judgment. Emily Bender and her colleagues' "stochastic parrots" paper, which forced the phrase into the public debate, is at bottom a linguist's argument about the difference between form and meaning.^[33] Kate Crawford's Atlas of AI reads artificial intelligence the way a historian reads an empire, through its extraction of labor, minerals, and data.^[34] Catherine D'Ignazio and Lauren Klein's Data Feminism, published open access, gives readers a framework for asking whose interests any data regime serves.^[35] None of these interventions could have been written by a pure technologist or a pure critic. They required people fluent in both registers.

The same borderland produces the careful middle position the public conversation badly needs. Ted Underwood, a literary scholar who has used computational methods for years, greeted language models neither with panic nor with hype but with experiments, testing what the models can and cannot do with fiction and documenting the results in the open.^[36] That is the temperament the moment calls for. Investigate the thing instead of cheering for it or refusing to look at it.

Religious studies has particular stakes in this examination, so let me name them. Language models are trained on scriptures, sermons, apologetics, and centuries of religious polemic, and they will answer questions about all of it, fluently and with no sense of what they do not know. Ask one about a Bible verse and you get something shaped by exactly the history of quotation and contestation that Mullen's project maps, flattened into a confident paragraph.^[13:1] Communities are already using these systems for devotional questions, and pastoral ones. Who will study what the models say about religion, whose theology dominates their training data, and what happens to religious authority when the answer machine speaks? Scholars of religion who understand how the systems work. That is a digital humanities job description, and the posting is open now.

There is a training argument here too, and it points at your classroom. Students now arrive having used AI systems for years, with no framework for thinking about what these systems are. The instructor who can teach them to interrogate a model, ask what is in the training data, probe where the outputs fail, name who profits, is doing core humanities work, the examination of powerful cultural claims, with the tools of the present. The digital humanities built the pedagogy for this over two decades of teaching students to work critically with data and computation. We need that pedagogy in every humanities department, not because AI is the future of the humanities, but because the humanities are how a society thinks clearly about AI.

And if the humanities do not claim this ground? Then the interrogation of AI's claims on culture will be left to the industry's own researchers and to whatever regulators can manage. The past decade of platform history tells you how that goes. This is the ceded-ground argument from the top of this post, at higher stakes and on a shorter clock. It is the strongest single reason I know why we need the digital humanities now and not eventually.

None of It Survives by Accident #

Here is the uncomfortable coda to everything I have argued. Suppose you are persuaded. The record has moved, the public projects are real, the pedagogy compounds, the institutions can be built, the AI moment demands us. Every item in that case depends on digital projects continuing to exist, and digital projects are mortal. Grant funding ends. Servers get decommissioned. Software rots. The graduate student who maintained the database graduates. Some of the most celebrated digital humanities projects of the 2000s are already gone, or survive only as screenshots and citations to dead links.

Rosenzweig, again, saw it first. We have moved from a world where the record decayed slowly on paper to one where it can vanish instantly and completely. "We have never preserved everything," he wrote, but "we need to start preserving something."^[5:2] An argument for why we need the digital humanities is incomplete without an argument for keeping digital humanities work alive, and that is where this series goes next. Part 5 takes up preservation directly, and Part 6 looks at minimal computing and the Global Outlook::Digital Humanities community as a path toward projects built to last.

For now, the case stands as follows. The evidence of our fields is digital, and we must be able to interrogate it. The publics we serve are reachable, and RRCHNM proves scholarship can reach them. The teaching commons is open, and it compounds. The infrastructure can be ours, if we build and sustain it. And the machines are making claims about culture that only people with our training, extended by these methods, can properly examine. That is five fronts, one conclusion. We need the digital humanities, and we need them now.

In this series #

1. Digital Humanities Part 1: What Are the Digital Humanities?
2. Digital Humanities Part 2: Digital Humanities Tools for Dissertations
3. Digital Humanities Part 3: Digital Humanities for Theology and Religious Studies
4. Digital Humanities Part 4: Why We Need the Digital Humanities
5. Digital Humanities Part 5: Preserving Digital Humanities Projects
6. Digital Humanities Part 6: Preservation through Minimal Computing and GO::DH

---

1. Google, "15 years of Google Books," The Keyword (Google blog), October 17, 2019, https://blog.google/products-and-platforms/products/search/15-years-google-books/ ↩︎

2. HathiTrust Digital Library, homepage, accessed July 2026, https://www.hathitrust.org/ ↩︎

3. Digital Public Library of America, homepage, accessed July 2026, https://dp.la/ ↩︎

4. Internet Archive, "Wayback Machine to Hit 'Once-in-a-Generation Milestone' this October: One Trillion Web Pages Archived," Internet Archive Blogs, July 1, 2025, https://blog.archive.org/2025/07/01/wayback-machine-to-hit-once-in-a-generation-milestone-this-october-one-trillion-web-pages-archived/ ↩︎

5. Roy Rosenzweig, "Scarcity or Abundance? Preserving the Past in a Digital Era," American Historical Review 108, no. 3 (June 2003): 735-762, available at https://rrchnm.org/essays/scarcity-or-abundance-preserving-the-past-in-a-digital-era/ ↩︎ ↩︎ ↩︎

6. Library of Congress, "Update on the Twitter Archive at the Library of Congress," Library of Congress Blog, December 26, 2017, https://blogs.loc.gov/loc/2017/12/update-on-the-twitter-archive-at-the-library-of-congress-2/ ↩︎

7. Roy Rosenzweig Center for History and New Media, "Our Story," George Mason University, accessed July 2026, https://rrchnm.org/our-story/ ↩︎ ↩︎ ↩︎

8. Zotero, homepage, Corporation for Digital Scholarship, accessed July 2026, https://www.zotero.org/ ↩︎

9. Omeka, homepage, accessed July 2026, https://omeka.org/ ↩︎

10. Library of Congress, "Library Accepts September 11 Digital Archive, Holds Symposium," press release, August 14, 2003, https://www.loc.gov/item/prn-03-142/ ↩︎ ↩︎

11. Lincoln Mullen, "Biography," lincolnmullen.com, accessed July 2026, https://lincolnmullen.com/bio/ ↩︎

12. Lincoln A. Mullen, The Chance of Salvation: A History of Conversion in America (Cambridge: Harvard University Press, 2017). On the American Academy of Religion award, see https://lincolnmullen.com/bio/ ↩︎

13. Lincoln A. Mullen, America's Public Bible: A Commentary (Stanford: Stanford University Press, 2023), https://americaspublicbible.org/ ↩︎ ↩︎

14. Lincoln Mullen, "americas-public-bible" (code, data, and website for America's Public Bible), GitHub, accessed July 2026, https://github.com/lmullen/americas-public-bible ↩︎

15. Lincoln Mullen, GitHub profile and repositories, accessed July 2026, https://github.com/lmullen ↩︎

16. Brian Croxall and Diane K. Jakacki, eds., What We Teach When We Teach DH: Digital Humanities in the Classroom (Minneapolis: University of Minnesota Press, 2023), https://www.upress.umn.edu/9781517915315/what-we-teach-when-we-teach-dh/ ↩︎

17. Claire Battershill and Shawna Ross, Using Digital Humanities in the Classroom: A Practical Introduction for Teachers, Lecturers, and Students, 2nd ed. (London: Bloomsbury Academic, 2022). Johanna Drucker, The Digital Humanities Coursebook: An Introduction to Digital Methods for Research and Scholarship (London: Routledge, 2021). ↩︎

18. Rebecca Frost Davis, Matthew K. Gold, Katherine D. Harris, and Jentery Sayers, eds., Digital Pedagogy in the Humanities: Concepts, Models, and Experiments (New York: Modern Language Association, 2020), https://digitalpedagogy.mla.hcommons.org/ See the project description at https://digitalpedagogy.mla.hcommons.org/description/ ↩︎ ↩︎

19. Knowledge Commons, "Our History," Sustaining the Commons, accessed July 2026, https://sustaining.hcommons.org/about-this-site/our-history/ ↩︎

20. Lauren Collister, "Infra Finder Spotlight: Knowledge Commons," Invest in Open Infrastructure, August 28, 2024, https://investinopen.org/blog/infra-finder-spotlight-knowledge-commons/ ↩︎

21. Michigan State University College of Arts & Letters, "Kathleen Fitzpatrick," faculty directory, accessed July 2026, https://directory.cal.msu.edu/kfitz ↩︎ ↩︎

22. DH@MSU, "About," Michigan State University, accessed July 2026, https://digitalhumanities.msu.edu/about/ On the graduate certificate, see https://digitalhumanities.msu.edu/certificate/ ↩︎ ↩︎

23. Matrix: Center for Digital Humanities & Social Sciences, "Mission," Michigan State University, accessed July 2026, https://matrix.msu.edu/mission/ ↩︎

24. LEADR (Lab for the Education and Advancement in Digital Research), "About," Michigan State University, accessed July 2026, https://leadr.msu.edu/about/ ↩︎

25. Daniel Allington, Sarah Brouillette, and David Golumbia, "Neoliberal Tools (and Archives): A Political History of Digital Humanities," Los Angeles Review of Books, May 1, 2016, https://lareviewofbooks.org/article/neoliberal-tools-archives-political-history-digital-humanities/ ↩︎ ↩︎ ↩︎ ↩︎

26. Domenico Fiormonte, Sukanta Chaudhuri, and Paola Ricaurte, eds., Global Debates in the Digital Humanities (Minneapolis: University of Minnesota Press, 2022), https://dhdebates.gc.cuny.edu/projects/global-debates-in-the-digital-humanities ↩︎

27. Elizabeth Losh and Jacqueline Wernimont, eds., Bodies of Information: Intersectional Feminism and Digital Humanities (Minneapolis: University of Minnesota Press, 2018), https://dhdebates.gc.cuny.edu/projects/bodies-of-information ↩︎

28. James E. Dobson, Critical Digital Humanities: The Search for a Methodology (Urbana: University of Illinois Press, 2019). ↩︎

29. David M. Berry and Anders Fagerjord, Digital Humanities: Knowledge and Critique in a Digital Age (Cambridge: Polity, 2017). ↩︎

30. dh+lib, "Neoliberal Tools (and Archives): A Political History of Digital Humanities," response roundup, Association of College and Research Libraries, May 5, 2016, https://acrl.ala.org/dh/2016/05/05/neoliberal-tools-and-archives-a-political-history-of-digital-humanities/ ↩︎

31. Andrew Piper, Juliana Spahr, and Richard Jean So, "Beyond Resistance: Towards a Future History of Digital Humanities," Los Angeles Review of Books, May 11, 2016, https://lareviewofbooks.org/article/beyond-resistance-towards-future-history-digital-humanities/ ↩︎

32. Matthew Kirschenbaum, "Am I a Digital Humanist? Confessions of a Neoliberal Tool," Medium, 2016, https://medium.com/@mkirschenbaum/am-i-a-digital-humanist-confessions-of-a-neoliberal-tool-1bc64caaa984 ↩︎

33. Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell, "On the Dangers of Stochastic Parrots: Can Language Models Be Too Big?," in Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Transparency (FAccT '21), March 2021, https://dl.acm.org/doi/10.1145/3442188.3445922 ↩︎

34. Kate Crawford, Atlas of AI: Power, Politics, and the Planetary Costs of Artificial Intelligence (New Haven: Yale University Press, 2021), https://yalebooks.yale.edu/book/9780300264630/atlas-of-ai/ ↩︎

35. Catherine D'Ignazio and Lauren F. Klein, Data Feminism (Cambridge: MIT Press, 2020), open access edition at https://data-feminism.mitpress.mit.edu/ ↩︎

36. Ted Underwood, "Using GPT-4 to Measure the Passage of Time in Fiction," The Stone and the Shell, March 19, 2023, https://tedunderwood.com/2023/03/19/using-gpt-4-to-measure-the-passage-of-time-in-fiction/ ↩︎