It's a pattern. This is the same kind of response the EFF got to its FOIA requests about NSLs (national security letters). They were sent back a pack of blank pages as all the text had been redacted. When the ACLU issued a request to the FBI on GPS tracking activity, they were sent back 111 blank pages.

You have three instances...that meets your hurdle for statistical significance?

I should hope so! The government is rather more regular than a particularly noisy lab experiment.

Now, as to what it meets the hurdle for statistical significance for . . . well, I personally prefer to avoid conspiracy theories when possible. So I'll just be keeping an eye open for more information.

It proves that they responded to this FOIA request by releasing a blank document. There are many different possible reasons why they released a blank document, but that's another question. (One which we can hopefully answer by examining the history of previous FOIA responses, but still another question.)

What level of proof do you require?

Out of curiosity, have you filed any FOIA requests?

For proof that there's a transparency problem, I'd want to look at hundreds of FOIA requests across 10-20 categories, content of news conferences and press conferences, level of press access to government officials, responsiveness to press inquiries, responsiveness to Congressional inquiries, and I could probably think of a few more things if I put more time into it. Then I'd want to compare that to data from previous administrations.