As can be seen from Figure \ref{275909}, most nodes in the tested networks have a very low betweenness score, indicating that there are only a few nodes through which many paths pass. This means that there is a definite bottlenecking phenomenon that is occurring in the tested networks, indicating a few nodes control most of the flow through the network, i.e. that that a few nodes can control the flow of information.
We also see that the betweeness edge distribution for the two networks we see as the best collaboration networks have smaller variance, signifying that most flow is unconstained but there are a few nodes that see slightly more flow. 

PageRank 

We see when evaluating the PageRank centrality distribution (Figure \ref{480483}) that our two good collaborative networks have a  mean and mode much lower than the other networks. This further reinforces the idea put forward by the betweenness centrality that there are a few important nodes but these nodes do not dominate all flow in the network, but may nudge the result significantly. In addition, visual inspection (Figure \ref{835332}) showed that similar bottlenecking structures existed within all the analyzed networks, showing that flow through the networks had similar dynamics.