2. Interpretation

The network analysis – especially upon consideration of the stable partnership network –reveals a structure of concentric layers.

1. The outermost layer is made up by the "Isolated", i.e. organisations that have no stable partnership within the network's giant component. About 22,000 out of the 31,000 organisations that participated in FP7 (70%) are in this category. Most of these participated in only one project, dipping their foot in the water but deciding not to dive in after all.

2. Next in, we find the about 1,700 "Intermediated" (5%). These organisations are part of the FP7 giant component, but they are connected to it via exactly one stable partner, which brokers their participation to the FP7 scene. These organisations find themselves constrained in the sense of Burt (2001). Many of them seem to be companies.

3. Closer to the center, about 7,300 "Full participants" take part in the giant component, and have 2 or more stable partners (25%). This makes their participation in FP7 more resilient: should they lose one of the stable partnership, they can still connect to the action through the other one.

4. At the center, about 700 organisations (2%) are "Intermediaries": each is the only channel for at least one of the Intermediaries to participate.

5. At the center of the center, one finds a highly connected 80-core with 143 organisations (0.5%), the FP7 Death Star. The most connected organisations have over 1,000 stable partners each. This group is not very diverse: it seems to consist entirely of public research centres and universities, with few or no companies and third sector organisations.

In one sentence: a small, highly cohesive group of a few hundred large public research centers and universities sits at the center of a stable partnership network with 9,000 nodes, in its turn surrounded by about 22,000 isolated nodes. We have been unable to interpret univocally this result. This configuration could be exactly what the proponents of FP7 wished for, with large research institutes acting as hubs and companies acting as spokes; or it could signal insufficient peer-to-peer collaboration across participants. The power law-like structure of the degree distribution is consistent with the idea that having won FP7 funding in the past increases the likelihood that an organisations will obtain more: again this could equally be a feature or a bug.

Spaghetti Open Data is grateful for any intepretation offered.