Fuzzy Times and Time Lines

Published:June 19, 2009 by pmh3g

Pocket_watch_with_chain

Most but not all entries in Encyclopedia Virginia have a time line. (This one, for instance, does not.) At first, a time line seems like a simple thing; it takes a subject and reduces it to days, months, and years, succinctly outlining the life of a poet, or a governor, or a civil-rights activist. Still, non-biographical entries can be tricky. For example, war—and its constituent battles—is confusing. Things happen simultaneously. When historians can agree that an event actually occurred, they can’t always agree when it happened. Civil War battles are a case in point, and on our time lines you’ll find fuzzy labels such as “July 3, 1863, morning” and “April 6, 1865, 2:00 p.m.–night.”

Night?  Morning? Why so vague?

At this point I should back up and without getting too geeky explain how our time line system works. The EV Founding Fathers were wise when they developed the architecture for our website; they designed a system that was flexible. The idea was to make decisions that, down the line, would create opportunities, not close them off. This philosophy shows up even in our time line system.

All the dates and descriptions that come together to form time lines are stored individually in a database. (In other words, we have “July 3, 1863, morning” in our database and not “Gettysburg Time Line.”) This allows us to repurpose those dates and descriptions in all sorts of ways, like the “Today in Virginia” feature on our front page and “Explore Virginia,” which allows you to play with both time lines and maps.

Very cool, believe me, but the code that makes all this magic happen—in order to create one of these time lines out of all those individual times, days, months, and years—has to be able to dynamically and chronologically sort those events. Luckily, computers are good at sorting numbers, but they stumble when trying to figure out words like “morning,” “evening,” and night.” At the moment, our computer just throws its hands up and orders stuff arbitrarily, which is hardly ideal. But if we add more specificity to events—if we tell the computer when exactly morning begins and ends, for instance—we may be misrepresenting the history. We know something happened in the morning, but do we know when in the morning? Or did it last all morning?

The solution, which is still a work in progress, seems to be a combination of exact times for the computer to sort and fuzzy language for users to read.  This would allow us temporal wiggle-room while also reflecting scholarly uncertainty.  Maybe time lines aren’t such simple things after all.

IN ADDITION: Visit Bruce Robertson’s Historical Event Markup Language (HEML) project at Mount Allison University (New Brunswick, Canada) for a more thorough investigation into thoughts on how to assign temporal and geographic representations to historical event.

IMAGE: Pocket watch with chain