For our first warm-up exercise, we'll be examining a simple time-series dataset: the history of nuclear testing by the eight (declared) nuclear nations. In the first phase of this project we will consider only the total number of test explosions across two dimensions:
- the state conducting the test
- the year in which it occurred
Despite the simplicity of the data, there are a number of ways to slice and group the data. Some things to consider:
- Is each year plotted separately, or are the totals grouped into decades or some other multi-year chunk
- Is time represented by position on the x or y axis? Or by angle? Or using color?
- What are the 'objects' being represented in your diagram? Are they the countries, the years, or the individual tests?
- How will your diagram address the 'lull' that resulted from the Comprehensive Test Ban Treaty in ’96 and the more recent resurgence in testing?
Getting started
Begin your investigation with this simplified dataset collected from the Natural Resources Defense Council's pages on the original and subcontinental nuclear powers.
Download a copy of arms-race.zip which contains JSON & CSV versions of the table below (plus a PlotDevice template script to get started with). This script builds on our work learning the drawing primitives in three major ways:
- setting drawing parameters with structured data rather than hard-coded values
- working with serialized JSON data
- using loops to render multiple ‘instances’ (a.k.a. rows) of a data set using the same procedure each time
Start by modifying the code in the render.pv script and explore some of the different ways to represent the country, year, and test-count values. Make 'snapshot' copies of your script as you modify it so you can build up an array of different approaches.
Next steps
Over the next week, continue working with this dataset and sketch out three different approaches to representing it in a way that tells a story.
Confine yourself to static, non-interactive representations for now, but consider any medium that could represent these values; posters, projections, and physical representations are all just as valid as pixels on a screen.
In at least one of your three variations, bring in one additional variable using data you find on the web and be sure to cite your source. This additional variable could be anything from a timeline of world events to the military budgets of the countries (or even the astrological signs of their leaders). Just make sure you can defend how it is in some way adding useful context to the data.
Feel free to use whatever tools you're comfortable with. Programmatic drawing can be very handy but so can Excel…
Data sources
This spreadsheet on Google Docs has a tidied up copy of the NRDC data and provides both the total number of tests as well as distinguishing between 'atmospheric' or 'underground' tests.
Also consider taking a look at the Wikipedia page listing the various tests and providing links to country-specific pages with background information about the individual tests.
For geographical data, take a look at the materials collected at “Johnston's Archive”. But keep in mind that using a map opens quite a few cans of worms, so be sure to have a clear idea of what you're trying to communicate before diving in...
Supplemental materials
I've assembled a cleaned-up csv file with most of the Johnston's Archive data. If you're looking for information about the naming schemes of the individual tests, yield estimates, locations, dates, etc. it may be helpful...
Lawrence Livermore National Laboratory recently released the declassified 2400 FPS films of many of their atmospheric nuclear tests. Those of you whose projects are dealing with specific tests rather than the summary statistics may find stills from the videos useful in your work. Take a look at their playlist on YouTube.