[arcdrag]

There's often cases of the data literally being measured in tera or peta bytes. Releasing the raw data isn't always feasible.

[aperiodic]

The mother of one of my friends works with the Zooniverse[1] project, a citizen science effort. She was saying that one of their biggest problems is how to distribute their data to the public. I naïvely suggested BitTorrent, figuring the sets were in the 10s of gigabytes at most. She laughed and said that that wasn't a great way to distribute terabytes of data every few weeks.

[1]: http://www.zooniverse.org/

[noilly]

I also assumed that the storage/bandwidth/computational costs of particle accelerator data sets would make open access practically useless