Strong Silent Types: Evil Robots and Their Way with Words

Keeping track of the North Korea crisis on Twitter

It's not unusual for bellicose rhetoric to emanate from Pyongyang. It's not even unusual for that rhetoric to threaten acts of war against its enemies Japan, South Korea and, of course, the United States of America.

It is somewhat unusual, however, for this rhetoric to unfold in the form of a regular series of statements over the course of weeks. These statements have taken the form of a crescendo which has prompted international condemnation, even warnings from allies like China, and activated the repositioning of US military hardware in case a missile is launched or some other act of 'hot' aggression occurs.

Every time a new statement is made, either by North Korea or one of the states it has threatened, Twitter jumps a little. A small spike in tweets about the region occurs and it's been increasingly hard to stay on top of the latest reports as they arrive, especially since they are occurring several times a day but not so frequently that they are the subject of constant news coverage.

I wanted some way of tracking how events have unfolded on Twitter, in order to be alerted to new developments in the crisis and simply to evaluate how Twitter flexes and reshapes as a medium-term story like this progresses.

The result, after a day or two of hacking together a little web app, is this:

Click the image above to go to the web app!

It's outrageously simple, and very rough around the edges, but it does work. Last night, when the New York Times story about DIA intelligence on North Korean nuclear missile capabilities broke, the spike in Twitter activity was duly registered as tweets on the subject rose to around 100 per minute.

The app checks Twitter every five minutes for the most recent tweets mentioning the words "north", "korea" and "missile". It then calculates, based on the time difference between a collected sample of tweets, a rough estimate of how many such tweets are appearing per minute on the site. The graph below the latest reading (which appears in the coloured box) shows collected incidence ratings over time for the last couple of hours.

The box changes colour as incidence grows and would gradually turn deep red if we entered the region of several hundred, or even several thousand, tweets per minute. Currently, the app would have difficulty tracking numbers above that (e.g. several hundred tweets per second or more) but I'm working on ways to improve its accuracy. In addition, I could potentially increase the frequency of Twitter scans but I haven't deemed this to be worthwhile yet.

It's an experiment, a way of playing around with the Twitter Search API, but I have had some good feedback already and I thought it would be fun to share. Suggestions for ways to improve the app are very welcome so don't be shy!

In the future this model could be adapted to track other news stories or events. But for now, if you're keeping track of the crisis on the Korean peninsula, you may just find it useful.


Photo: "Sinpyong Lake, North Korea" by yeowatzup. Reproduced under a Creative Commons (CC) License.

comments powered by Disqus





The Machine Starts is all over Twitter! Click here to check out a flurry of observations on digital culture and links to the latest and most interesting tech news from around the web.


The Interface and Hyperreality

Interfaces express not that a journey has been eliminated, but that a new one may be created.

Predicting a Riot: What Violence Means for "Society"

Networking, in many senses, gives rise to a new perspective on the London Riots of 2011.

Could you quit the internet?

Does abstinence from the web ever last? Is it even a good idea?

The Computer Virus: Our Cultural Contagion

Computer viruses are not just computer viruses. They spread in pathological as well as technological ways.