Back in the bad old days, biomolecular crystallographers would tend to hold their data for as long as they could, in order to publish more about data that only they had. (To be fair, it might have taken 10 years to get the protein / DNA / virus / etc. to crystallize.) The publications would be scores of pages because they wanted to describe all of the details, and hence the credit.
They would, for example, only publish the backbone structure of a protein as a drawing, and not as 3D coordinates. In response, people would input the 2D coordinates into software that would try to reconstruct the 3D structure. (This was possible because the publication would often be as a split stereoscopic image.)
Finally, in the mid-1990s, the major journals put their feet down and said they would only publish of the coordinates were deposited in a public data collection, which would be released no later than a year of publication.
So yes, the people funding the science, and the people publishing the science, who sometimes have to force the people doing the science to actually be more open.
They would, for example, only publish the backbone structure of a protein as a drawing, and not as 3D coordinates. In response, people would input the 2D coordinates into software that would try to reconstruct the 3D structure. (This was possible because the publication would often be as a split stereoscopic image.)
Finally, in the mid-1990s, the major journals put their feet down and said they would only publish of the coordinates were deposited in a public data collection, which would be released no later than a year of publication.
So yes, the people funding the science, and the people publishing the science, who sometimes have to force the people doing the science to actually be more open.