Making a One-Sided Case for Digitization

Digitization is awesome and its issues are solvable

During our HIST 696 class this week, we discussed the advantages and disadvantages of source digitization.¹I know we also talked about AI and if I don’t talk about AI in this blog post it will be the second week in a row I focus on the theory portion of class rather than practice. But I plead your sympathy. My background is in tech and I have already had a lifetime’s worth of AI discussions. I cannot bring myself to willingly talk about AI for an entire blog post. I found myself a little surprised that the discussion focused mostly on the disadvantages, though I suppose this might have something to do with the arguments against digitization are less obvious and more interesting than the arguments for digitization. As Blaney and Siefring discussed in their paper “A Culture of Non-Citation,” there still remains a persistent culture of sentiments against digital sources in the humanities, and so it is important that we also remember the positives of this process.²Blaney, Jonathan, and Judith Siefring. “A Culture of Non-Citation: Assessing the Digital Impact of British History Online and the Early English Books Online Text Creation Partnership.” Digital Humanities Quarterly 11, no. 1 (December 1, 2016). https://doi.org/10.63744/rme2max8ep42.
Therefore, for the sake of laying it all out, though, I decided I wanted to write about a few reasons I love digitization, and how our discussions of its limitations can be used to inform solutions to those problems.

#1: History is for everybody

In Sarah Maza’s 2017 book Thinking About History, she notes that history is one of the only academic fields which still consistently has laymen operating at its highest levels. As Maza notes, the Pulitzer Prize for history has gone to number of people without history degrees in the 21st century alone, including English professors, journalists and bankers.³Maza, Sarah. Thinking About History. United Kingdom: University of Chicago Press, 2017, 2. It’s a beautiful fact about history, and as digital history is increasingly something accessible to research and contribution by the general public via blogs, Wikipedia, and social media, the benefits of digitization for our field will only multiply.

Keep in mind – digitization covers more than just primary sources! Under the umbrella of source digitalization we can digitize secondary sources and tertiary sources to allow for even greater access to the humanities among everyday people.⁴Yes, technically, if you want to go by the book, you may be limited by copyright here. But also, long live Aaron Swartz.

#2: The vast majority of sources are easily digitizable

One argument that came up in class against digitization is that there are many sources that digitize poorly. Lena brought up a fascinating example of a collection she had worked on, where the pictures available failed to show that the objects emitted light! It’s a fascinating example of an area where a simple picture fails to tell the whole story of an interesting historical object.

The vast majority of our sourcing as historians, though, digitizes much better. How often when reading a history text do you see a physical object being cited rather than an archival paper, an image, even a video or audio piece, all of which transfer smoothly to digital versions?⁵Surprisingly, I cannot find any statistics on this. Between written documents/images/video/audio, surely it’s greater than 99% of citations though, right? Let me know what you think. And yes, you lose certain aspects of a document – its feel, its smell, its weight, its size – by not being able to touch it, but only rarely do factors like these make it into the final product. There’s a reason historians spend so much time in archives taking pictures, and less reading sources. And besides, when we do feel the urge to go into the archives:

#3: Digital sourcing has a very low opportunity cost

For that small percentage of sources, which do not digitize well, digitization does not need to tell the whole picture. One of the advantages of digitizing is allowing researchers to find the sources they need, and this is where our discussion of metadata comes in. Metadata can inform researchers both that a piece of data exists, and a good description can help inform a decision to come into the archives and take a look themselves.

Digital hosting is (relatively) cheap, with robust databases available at prices in the range of just hundreds of dollars per year, which isn’t even considering the fact that most institutions already pay for some form of hosting. If all digital sourcing does is tell someone they need to check out an archive, it has achieved something positive at very low opportunity cost.

#4: We can build out infrastructure to solve many problems

I found the example Zine from Michelle Moravec’s article to be a very interesting example of a problem we could theoretically mostly solve by building out better infrastructure.⁶Moravec, Michelle. “What Would You Do? Historians’ Ethics and Digitized Archives.” On Archivy, April 14, 2016. https://medium.com/on-archivy/on-april-at-the-organization-of-american-historians-i-participated-in-a-roundtable-discussion-563213dc94a. Beyond taking cues from other fields, like journalism, and potentially censoring the name, or reaching out the subject (if they are alive) to ask for their preferences, there are technical methods for obscuring objects like the hypothetical zine from people who are not coming to the archive for research. For example, all major search engines respect robots.txt or <meta name=”robots” content=”noindex”> tags to instruct them not to index a certain page in their results. For crawlers or LLMs, it’s reasonable to be concerned that bad faith actors may ignore agreed upon technical standards.⁷Shilov, Anton. “Several AI Companies Said to Be Ignoring Robots Dot Txt Exclusion, Scraping Content without Permission: Report.” Tom’s Hardware, June 21, 2024. https://www.tomshardware.com/tech-industry/artificial-intelligence/several-ai-companies-said-to-be-ignoring-robots-dot-txt-exclusion-scraping-content-without-permission-report. We still have a technical solution, though! Captcha-like gates, which completely block off a page from loading on the server layer until a challenge is solved, can prevent sensitive data from being seized. Even simply locking the content behind account creation is a simple way of defeating automated threats.⁸The reason I say Captcha-like instead of just Captcha is that LLM browser agents are steadily getting better at defeating captchas, and we’ll need better ways to defeat them soon. If you’re interested, here’s a good paper on it: Ousat, Behzad, Nikita Turkmen, Lalchandra Rampersaud, Dillan Bailey, and Amin Kharraz. 2026. “Broken Gates: Re-Evaluating Web Bot Defenses in the Age of LLM Agents.” Version 1. Preprint, arXiv. https://doi.org/10.48550/ARXIV.2607.18659.

In class it was brought up that digital sources can degrade – and this is simply another issue of inadequate digital humanities architecture. The example brought up in class was lossy compression, which in general should not be something and archive has to worry about, as images should be getting re-compressed regularly, but to the extent changes are made to an image and it is recompressed, this is what diff-centric architecture is for. Git has been an industry standard for some time for tracking versions, and something which we can easily implement to cheaply track file versions across an archive.⁹For those unfamiliar/curious: https://en.wikipedia.org/wiki/Git Community based resources such as the Internet Archive and open source VM-emulators are also constantly being improved in order to help address other types of potential digital source degradation, namely link rot and digital obsolescence.

And no, digitization is not perfect. S.L. Ziegler is right when they note that we must be cognizant of how we use data from people in the past, as to not become “grimy.”¹⁰Ziegler, S.L. “Open Data in Cultural Heritage Institutions: Can We Be Better than Data Brokers?” Digital Humanities Quarterly 14, no. 2 (June 19, 2020). https://doi.org/10.63744/jfyb9zym5am3. Beyond that, it takes a great deal of labor to digitize a database, labor which we must not take for granted. But the downsides that are present in digitization are mitigable, and it’s good to remember that when we discuss the potential threats to digitization, we do so to refine better methods and practices, not to discourage the practice altogether.

• 1
  I know we also talked about AI and if I don’t talk about AI in this blog post it will be the second week in a row I focus on the theory portion of class rather than practice. But I plead your sympathy. My background is in tech and I have already had a lifetime’s worth of AI discussions. I cannot bring myself to willingly talk about AI for an entire blog post. ↩︎
• 2
  Blaney, Jonathan, and Judith Siefring. “A Culture of Non-Citation: Assessing the Digital Impact of British History Online and the Early English Books Online Text Creation Partnership.” Digital Humanities Quarterly 11, no. 1 (December 1, 2016). https://doi.org/10.63744/rme2max8ep42.
  ↩︎
• 3
  Maza, Sarah. Thinking About History. United Kingdom: University of Chicago Press, 2017, 2. ↩︎
• 4
  Yes, technically, if you want to go by the book, you may be limited by copyright here. But also, long live Aaron Swartz. ↩︎
• 5
  Surprisingly, I cannot find any statistics on this. Between written documents/images/video/audio, surely it’s greater than 99% of citations though, right? Let me know what you think. ↩︎
• 6
  Moravec, Michelle. “What Would You Do? Historians’ Ethics and Digitized Archives.” On Archivy, April 14, 2016. https://medium.com/on-archivy/on-april-at-the-organization-of-american-historians-i-participated-in-a-roundtable-discussion-563213dc94a. ↩︎
• 7
  Shilov, Anton. “Several AI Companies Said to Be Ignoring Robots Dot Txt Exclusion, Scraping Content without Permission: Report.” Tom’s Hardware, June 21, 2024. https://www.tomshardware.com/tech-industry/artificial-intelligence/several-ai-companies-said-to-be-ignoring-robots-dot-txt-exclusion-scraping-content-without-permission-report. ↩︎
• 8
  The reason I say Captcha-like instead of just Captcha is that LLM browser agents are steadily getting better at defeating captchas, and we’ll need better ways to defeat them soon. If you’re interested, here’s a good paper on it: Ousat, Behzad, Nikita Turkmen, Lalchandra Rampersaud, Dillan Bailey, and Amin Kharraz. 2026. “Broken Gates: Re-Evaluating Web Bot Defenses in the Age of LLM Agents.” Version 1. Preprint, arXiv. https://doi.org/10.48550/ARXIV.2607.18659. ↩︎
• 9
  For those unfamiliar/curious: https://en.wikipedia.org/wiki/Git ↩︎
• 10
  Ziegler, S.L. “Open Data in Cultural Heritage Institutions: Can We Be Better than Data Brokers?” Digital Humanities Quarterly 14, no. 2 (June 19, 2020). https://doi.org/10.63744/jfyb9zym5am3. ↩︎