That's an excellent interview question! "How do you get the maximum of data out before getting noticed?". I guess the most effective way is to tamper with source code and smuggle data out along with legitimate traffic.

You don't need to extract the whole lot.

"Thousands of random users' data has been downloaded from Facebook and published on Wikileaks" would be enough for a good media scare story.

Change Wikileaks to 4Chan.

That was their story last week. This week their story is "We're going to leak a bunch of folks' Swiss bank account details because some of them might be avoiding taxes!"

They didn't redact informants and collaborators names from many of the documents they leaked. I would hardly call that exercising caution.

I understand that in some cases redacting names weakens the impact of the leak and its political statement. However, releasing the name of some random informant working on the ground in Iraq has no substantial effect on the impact of the whole leak, but can have an enormous effect on that person's life, especially if he's killed.