Draft: Evaluation of incidents in cyber conflict [Graphic]

The following graphic at the bottom of this page is a visualization of the impact of cyber incidents that is a work in progress. There seems to be fairly substantial interest in the contents of my little black book of projects (Sam’s Skunkworks of cyber conflict) so I thought I’d throw this up to give people something new to look at early this week. Here is a summary of what you’re looking at. One problem is trying to design a visualization strategy that gives more information than is put into the data. The principle here is to try and graph in three vectors the severity of an incident, the role from (non-actor to state actor) of an entity and the relative conflict status (peace to all out everybody agrees war). You can try and heat map such a study in Microsoft Excel but it loses something. The following image depicts the study effort and design.


Click on image to make larger

As you can see the image leaves a lot to be desired. Currently I’m working up an IRB (university speak for approval to poke and prod people) that will allow me to move this from concept to reality as a scientific study.


The methodology suggested for this is to identify 25 incidents over the last decade (since January 1, 2000) that can be evaluated. Each incident will have  paragraph describing it and the incident will have a Likert Scale following it. The three questions will be Severity 0 to 9, Entity 0 to 9, and Conflict 0 to 9. Two polar examples will be chosen for each of the axis and will not be part of the set of 25. The examples for each axis would be agreed upon by a few experts, and largely selected based on the experts in the field agreement.

A group of 30 experts will be selected to examine the 25 elements and make their choices. It is not expected that they will all agree but there are methods to visualization that will depict disagreement and agreement.

Portraying the results

The final graphic will depict something similar to the following. Using JMP9 or other stools like SPSS the data analysis can be done fairly easily (once the entirety of the methodology has been chosen). The resulting graphical depiction should allow for a visualization of clusters (inherent in any study due to selection bias) for sure. As a prototype study though it could lead to an analysis strategy which is actually being tested. Can you categorize events based on these three factors as a tool for understanding patterns in the events? Though an interesting conceptual visualization exercise the actual hypothesis is more like “Does there factors define reliably the spectrum of cyber conflict” or something like that. This is one method of determining expert agreement that could lead to better analysis.



Click on image to see larger


Once again these are just items that I’ve been scribbling for awhile and simply haven’t had time to get a team together to try and build and analyze. The life of a scientist without graduate students <insert large grin here>. I’d like to know if this is of any value to people as I see lots of blog hits but no <this is crap> or <this is awesome> feedback. I figure keeping all kinds of ideas to myself “might” make me money in the future, but I really doubt it. Scientists in America don’t make money we spend it.Yes some of these ideas are half baked. That is kind of the point. If I had finished them they’d be journal articles and behind a pay wall not a blog post I did while waiting fort he weather to clear so I could go for a run.

Leave a Reply