KrackPlot is a network visualization tool intended for social networks, and designed to be quick and easy to run, yet highly customizable. KrackPlot has been under development for many years. The current version, 4.3, is written in java, can read and write many graph formats and create resolution-independent output for Word or Powerpoint. You can download a self-installer for this version below. KrackPlot can also be run as an applet in a browser that supports java 1.4.2 or higher.

Currently, we are researching ways to use motion in network visualizations to highlight changes in the network structure over time.


You can download a self-installing version of the program for windows here. The current version is 4.3, built on 11/02/2006. You can also install an older version from this zip file if you prefer, or don't use windows.


We are currently working on a new version of the manual that covers the more recent features of KrackPlot. You can find brief tutorials of some of these features from the help menu in the running program.

You can find papers and articles about KrackPlot here.

Gallery of images

Here are some recent visualizations of network data created using KrackPlot, along with some explanations of the principle they illustrate, and how they were created.

1. Visualizing flow through a network

Many questions that users may want to explore concerning networks can be viewed in terms of something flowing through a network, obeying direction and capacity constraints on links and nodes. For example, users might explore the overall capacity of infrastructure networks such as roads or gas pipelines or explore their vulnerability to attack. The image below illustrates this with pipeline data in Texas gather from satellite imagery.

The image shows potential flow between a chosen source node in the lower left group and a chosen sink on the right, computed as a max flow with the Ford-Fulkerson algorithm, integrated with KrackPlot. Nodes on the flow are shown in red or orange: red means that no more flow could be sent to the node and orange means that flow could potentially be increased. Nodes with no flow are shown in blue or yellow: blue means that no flow could be sent to the node given the current flow. This means that the large yellow section to the left of the blue section highlights a bottleneck in the flow.

The user can explore this flow interactively by adding or blocking links or nodes and changing the capacity of nodes. Flow is recomputed as soon as a change is made, giving the user a sense of direct connection with the object, even though they may be interacting with an external analytic tool. The flow can also be animated to show how elements that take different paths may move apart and come together. You can see a succession of images that show such an interaction in powerpoint slides here, or download KP and follow the interactive tutorial on flow.

This network has approximately 3000 nodes and the same number of links. Each link may consist of a number of line segments, in this case up to 80. Because the latitudes and longitudes of nodes are given, KrackPlot can lay the network over a map, given as an image with its corners registered.

2. Visualizing agents planning to access an area

This network illustrates working with an agent simulation program. In the Hats simulator, a small proportion of agents may be planning to access a protected area. To do this, they must arrive at the area as a group, and between them have the right set of capabilities to make access. Agents have different native capabilities, and can acquire capabilities from each other in meetings. The goal for the user is to find a group of agents that may be planning to access some protected area, in a simulation containing perhaps millions of agents, mostly benign, and thousands of protected areas.

The image below highlights a group that may be planning to access an area requiring four capabilities. Only agents within a small fixed distance of a chosen area are initially selected interactively in KP, showing a few hundred out of more than 10,000 agents. Groups have been assigned by a separate analytic tool, and are used to color the agents, with agents colored blue belonging only to benign groups. The red agents at the top right of the diagram may be planning to make an access. Agents are linked to meetings, shown as ovals, and colored links indicate that the agent got or shared some capability in a meeting. The meetings have been laid out in time order from left to right, so it is clear that the group acquired all four capabilities through meetings, two of them quite recently. Also, two of the capabilities were acquired through a chain of meetings, behavior not seen elsewhere in the local network.

The image above is taken from a high-resolution image, which can be found here.

3. Visualizing correlations in voting behavior

This image shows correlations in voting between western countries on certain UN resolutions from 1981 to 1985. Positive correlations are shown with solid black lines and negative correlations with dashed red lines. The layout is automatic, with an algorithm built into KrackPlot that optimizes a weighted sum of terms that include minimizing the length of positive links and maximizing the length of negative links. This was joint work with Paulette Lloyd and Cathleen McGrath. More details can be found here.