Would Merging/Co-locating of nodes with thick edges make "things" better?
It depends on the "causal" antecedents of these collaborations. For example, in the case of co-location, it is not clear if geographic proximity plays any role in the emergence of interactions. There is something to be said for the view that academics also need to interact face to face. But this is further mitigated by the advent of sophisticated online communication tools. Furthermore, the results are for a particular timeslice (e.g., 2007) with no guarantees that the patterns persistent over longer periods of time. Co-location is a big capital investment that needs to be warranted by a longer term analysis of the dynamics of the system.

Even if we assume that the results are reliable, co-location might actually be detrimental as it might lead to a situation where the rich gets richer and the poor gets poorer: major collaboration hubs would become bigger and more agglomerated while satellites quickly whither away and die. Some may regard this as not a good outcome. It is necessary, therefore to couple the results with a conceptual framework that gives the system's stewards an indication of what needs to be done (if any).