Since I didn't notice mention of it, and at risk/hope of prolonging this fascinating discussion:  these are issues the high-energy physics community has grappled with, moreso recently due to the Large Hadron Collider.

There may be cultural, bureaucratic, and engineering (in that order?) examples to be drawn. For some perspective, they are dealing with data on the order of GB/s from just one instrument (well, when it runs...):

http://www.scientificcomputing.com/build...
http://lcg.web.cern.ch/lcg/

They are sharing data - and collaboration/pipeline/analysis software - on an unprecedented scale out of necessity due to the complexity of the undertaking.


Considering the relative human impacts, it is imperative for med/neuro-imaging to accelerate similar steps.


Regarding Luis argument somewhere along the line that "data collection must be valued", this is something HEP deals with too... HEP is in the crudest terms divided between "theorists" and "experimentalists." The former get perhaps more glory, but the latter certainly get more money. Being an outsider to HEP, I'm sure there are plenty of problems I don't know about --- but to a layman they seem to have found some ways to make things work despite not infrequent >100 author papers. Granted, it's an imperfect analogy, and I doubt the division in imaging can - nor should - ever be quite so sharp.