TED Staff 2014

August 21, 2014 by alex

In the 1960’s, Georg Nees created some of the first computer-generated art. His piece Eight Corner contained figures produced by connecting eight randomly-generated points. The cryptic beauty of the image struck me from the first moment I saw it, and I wanted to create a similar work.

Georg Nees

While Nees is quite well-known as a pioneer of generative computer art, there are not many resources online disussing the Eight Corner work shown here.

I found one Tumblr page on Eight Corner which contains the following description:

To produce the graphics, I used a drawing board controlled by a punch tape and a digital computer producing the pilot tape. Each graphic has random parameters. The program for each graphic repeats generative fundamental operations so that the mere repetitions, the aesthetic redundancy, produce the random parametric values of the aesthetic improbability of the graphic during each repetition.

Rule for 8-corner: Distribute eight dots inside the figure square and connect them with a closed straight edge line.


I was fascinated by the figures Nees created from random data. I wanted to produce similar graphics, but from non-random data. I thought it would be interesting to derive random-looking figures from bodies of text. Essentially I wanted to create a hashing algorithm, where the hash is a pictogram instead of text string.

Each line of text in the input produces a figure. Like in Nees work, each figure has eight connected points. Unlike Nees', mine has no randomness. The same text input will always produce the same graphical output.

The earlier points influence the placement of later points, so while the figures are not random they are significantly influenced by previous figures. The same text will always produce the same figures, but the same set of words in a different order will produce different figures.

I used the code I created to create one figure for each employee at my company, TED Conferences. The names are arranged in order of hire date, so each figure influences the shape of figures representing people who started later.

The code to produce this image is available on GitHub and rubygems.org.

Other Versions

I’ve also uploaded a few alternate versions which give a bit more information about how the figures are constructed.


  1. A red dot indicates the initial point.
  2. The first line is solid black.
  3. Each subsequent line is slightly lighter.
  4. The final line is nearly white.


As stated, the shape of earlier figures (names) in the document influence the construction of later figures. To see the effects of this interdependence between figures, this version of the poster switches the first and second names in the list. This is the only change from the original.

Sometimes the difference is minimal, and sometimes an entirely new figure results. An individual’s contribution to a group is not a fixed quantity.


This is another view on the impacts of interdependence between figures. This version of the poster renders each figure as it would appear if it were the only figure being generated. (Each figure is un-influenced by those around it.)

As in the re-ordered version, the change is sometimes modest and sometimes dramatic. The same person shows different sides of themselves in a group and on their own.

☙ ☙ ☙