As an academic, I have two principal goals. The first goal is to teach students and hopefully to have a positive impact on their lives. The second (and it is really a by product of the first in my opinion) is to perform good research and publish papers that advances scientific understanding. Recently, some of our work inspired some artwork on display at University College London. I found this to be incredibly gratifying and I am left wondering if this mosaic will have more “impact” than most of the papers I publish?
First, some background. A few years ago, our research in network emulation led us to consider the characteristics of Internet topologies. We wanted to be able to generate random graphs of varying sizes that still maintained the essential characteristics of original graphs. As is often the case with research, the work led us in unanticipated directions. My former student, Priya Mahadevan,worked with a terrific former Physics PhD at CAIDA, Dmitri Krioukov, to find a series of graph degree distributions that are guaranteed to always converge to the original graph. Interestingly, it appears that capturing the distribution of degree-labeled node triples in a wide range of complex graphs is sufficient to reproduce global graph properties.
We described this approach, called the dK-series, in a SIGCOMM 2006 paper. In a SIGCOMM 2007 paper, we described Orbis, a graph generator that employs a number of algorithms to generate random graphs that reproduce a given dK-distribution. Both papers contained visualizations of the graphs. You can see some of the original pictures at the Orbis web page.
Back to the artwork. The folks in the Computer Science Department at University College London liked some of the depictions of Internet topology and their resemblance to “digital dandelions.” They commissioned an artist to create a mosaic based on these depictions and have it on display in their new building. The Department Head of CS at UCL, Anthony Finkelstein, was kind enough to forward some great pictures. The inscription reads:
Mosaic by Hannah Griffiths 2009
Networking researchers model the structure and growth of the Internet to
predict its evolution. These models reveal the scaling characteristics of
networks, and can help scientists evaluate how new network algorithms and
architectures will perform. This mosaic, which looks like the head of a
dandelion, is based on a map of the Internet generated by algorithms from
computer scientists at UC San Diego in 2007. This map features Internet
nodes – the dots – and linkages – the lines. It is a (mostly) randomly
generated graph that retains the essential interconnectivity characteristics
of a specific corner of the Internet but doubles the number of nodes.