Final thoughts- just a few comments on the limitations of the History Engine as it was employed in the case of our final exam essay question.

One of our exam essay questions involved using a series of History Engine episodes from select dates as a core body of evidence for assessing the entire history of America in the 19th century. While reading the dozens of episodes for the selected months and years, I was struck by how almost every episode seemed to relate to either slavery, the threat of secession, or the Civil War. I think I would be underestimating if I said that 50% of the episodes could easily be categorized in one of those three topics. This seems astonishing: we were given almost unlimited freedom in the types of primary sources that could be used, but still, the episodes ended up being almost entirely about just a few select topics.

This raises the issue of a self-selection bias in the project: did students write their episodes about slavery, secession, and the Civil War because those are generally the only things that happened in the 19th century, or did students write their episodes about those topics because they had been taught that those were what happened in the 19th century? Suppose a student searches through the Furman archives for an "interesting" primary source to write an episode on. Which will seem more "interesting" and "relevant" to the student, an episode about topics he has been taught are important (slavery, secession, Civil War), or an episode about something unrelated and obscure?

An experiment could probably be done to try and avoid this bias: take a group of students and assign a huge body of primary resources from an obscure time period limited to a small geographic region. Providing virtually no historical context or secondary instruction, have them write episodes based on the primary sources, giving complete freedom in the area of episode selection. Ultimately, you could compare what types of episodes were written by students with what historians have traditionally done for that particular time period and geographic area. Though probably impossible to implement in an actual course, the results could be interesting.