Building A Bridge from Digital History to Writing

November 19, 2014

Digital History (DH) is in a “spaghetti phase”, with practitioners experimenting in many formats to see what sticks. Experimentation is good, but when do we start asking where it’s going? Digital projects based on limited data offer interesting curiosities, but once DH reaches a critical mass of concentrated data with a useable ontology, vastly more opportunities arise. When that happens, DH has the potential to transform the mechanics of history, even as it honors its core substance. 98% of undergrads are not history majors, and traditional research in paper archives isn’t in their futures. But evidence-based historical analysis still can be if digital platforms originate from and contain good primary and secondary sources.

Envisioning History (EH) is actively working toward this goal by utilizing sophisticated DH tools to improve undergraduate writing. DH’s power lies in its ability to organize data in tabular and visual forms, and we should use it for what it does best. Dots on maps and network diagrams are good. Visualizations help gain a broad sense of series of events or relationships and are highly appealing to generations more accustomed to consuming information on some type of monitor, cell phone or visual-oriented device. But even the best visual learning environments must be translated into effective writing in order to have transformative effects.

EH illustrated the imperative of effective writing as a product of insightful analysis at a recent demonstration at major university. Using our database and the Palantir analysis tools students were asked to write a ½-page “President’s Daily Brief” (PDB) giving President Roosevelt an appraisal of the situation in the Pacific a few days before the Battle of Midway. The EH database includes many excellent primary sources such as the CINCPAC War Diary and others to provide the evidence. The exercise ran during the mid-semester exam period, so we could assign no outside work; the trial was run solely in class.

Writing a half-page PDB based on large amounts of information requires stripping the issue to its essence. Real PDBs are written by analysts deeply steeped in the subject matter and who’ve attended a 5-week PDB writing camp. Assigning STEM and Humanities undergrads to dig into a sea of source documents and pull out the “nuggets” is achievable with DH tools, however, if the ontology is effective and they know how to use the tools. Our demonstration was impacted by the lack of “digging” outside class, but even so it helped us see where the students struggled as well as what worked. We were most interested in how the tools helped them make the leap from facts to evidence-based analysis.

The week before the scheduled classes several students volunteered as “Tech Mentors” (~1/3 of the classes); they were given EH-Palantir accounts and completed the EH online training. The teams were also directed to an Envisioning History online video that tells the Midway story.

Monday was spent with orientation and exercises in the platform for all students in both class sections.  On Wednesday the instructors began with a description of the PDB format (Title; Assessment; Evidence; Differing Opinions) and handed out the assignment guidelines that include a real PDB from 2001. This was followed by a demonstration on using the software to examine details of the Doolittle Raid on Tokyo. The demo featured a structured approach to extract and analyze the information:

Step 1 starts from the Map with all events between 12/1/41 and 5/31/42. This is created with a single query. The Map’s Histogram categorizes all events based on the ontology (Battle of Midway shown); one click offers a set of further drill-down options for any category;

Figure 1 - Map showing ~2,200 events from Dec. 1941 - May 31st, 1942 with the Battle of Midway related events selected (in yellow on the Map) and the Histogram dialog box offering additional analysis options.

Step 2 builds a “folder” of all the related events using one click from the above dialog box options;

Figure 2 - Folder showing 30 Battle of Midway-related events that occurred before May 31st indexed chronologically in the left panel; the main panel shows details of the first event with links to multiple source documents at bottom right.

Step 3 is to drill down into the primary and secondary source documents behind those events to obtain a broader context.

Figure 3 - One of the source documents (an NSA Special Research History) from the event shown above.

Each step in this “Histogram/Folder/Source” sequence is a simple matter of a few clicks. Students were given the remaining 30 minutes Wednesday to search and explore the data for the period to May 31, 1942. On Friday teams worked independently to develop their PDB submissions with the Instructors engaging and drawing their attention to key points. When teams found important “nuggets” of information, the Instructors pointed this out and other teams scrambled to find the same points.

Midway was extraordinarily dramatic. With a smart phone any student can know in a few seconds how it turned out, and at 70 years remove it’s tempting to assume the outcome was a foregone conclusion. Having students suspend that knowledge and imagine themselves in the basement at Admiral Nimitz’s headquarters late into the night on May 31st is a challenge, but it forces them to understand the essence of the situation and to realize how uncertain the outcome was, especially in the conditions in May 1942.

Manipulation of the Palantir tools is straightforward for tech-savvy undergrads. However, a short foundation lesson on the analytical purpose quickens their acclimation to the environment. We determined that more repetitive and structured exercises with the Histogram/Folder/Source sequence would improve confidence their abilities.

With limited time available for digging, we coached students through the mechanics of extracting the evidence, and the software definitely assisted them in quickly isolating the important points. Using these steps, the data available can be visually organized to assist in constructive writing. Before the class, Envisioning History constructed a shared “template” for students to add to their Graph palettes. The instructors demonstrated how to download the template and how to visually sort events into the various “buckets” to organize their thoughts (shown). In this palette, events are shown as boxes with national flag icons. The Histogram allows sorting based on the ontology with one or two clicks, and then the icons are dragged into the buckets. Once the sorting is done, the next step was to work through the underlying source documents, extract the best evidence and write the PDBs.

Figure 4 – Template filled with the 30 events (the small icons with respective national flags)

One issue that arose—and it was mainly a failure of the instructors to caution the students against it—was overuse of a tool called the Snippets Helper that allows drag-and-drop building of documents from the source documents, with footnoting done automatically. That was too much temptation for some, and it resulted in a few submissions that were strings of quotes with little analysis. This is an easy issue to fix in the future.

The written assignments were turned in (students hit “send”) at the end of Friday’s class. Of the 10 team submissions, several were especially good efforts given the time the teams had to scrutinize source documents and develop their evidence-based analysis. 

While this demonstration had limitations, we think we’re on the right track with the Histogram/Folder/Source sequence for extracting data and the template for visually organizing thoughts. More work is needed, though, and in the spring 2015 semester we’ll run a full-semester class with opportunities to refine and improve training of students and instructors and to develop additional structured approaches to help students move from facts to analysis.

In future iterations each student would have their own user account, as well as time outside of class to explore source documents. We’ll build in more repetitive exercises early on, and we’ll provide more time for students in civilian universities who are not familiar with a military culture.  We still believe that the small team approach has value, especially for writing PDBs. Real PDBs are very heavily edited, so it’s appropriate that teams build these as assignments. The interaction and discussions within and among teams seemed to be the greatest factor in making the leap from structure to analysis.

Comments and ideas welcome.

Sources:

All of the sources are contained in the EH database. For those without EH database access, they can also be found at: