The two questions that must be asked about this are: Are all documents created equal? Is one document a multi-page report containing detailed information, or i

The two questions that must be asked about this are: Are all documents created equal? Is one document a multi-page report containing detailed information, or is it a brief memorandum about something trivial? How much of these documents is boilerplate such as daily summaries that will repeat the same information as long as it's relevant? If I grab someone's mbox is that one document, or a collection of emails that each count as one document? Of course none of that matters if there the number is entirely fictional. But there is still the second question, does it matter how many documents were leaked? Would your opinion of Snowden change if the number were different? If you defend him, is there an upper limit after which revealing the information is no longer justified? If you think he's a traitor, is there a small enough number that you would not object to? I don't think it matters. The writers probably don't think so either and are thus using it not to add information to their articles but because sounds better. A person learning about this for the first time will see "1.7 million" and think either "Wow, that's a lot!" or "This reporter did a good job." A person writing "Snowden is reported to have stolen an unknown number of documents," will raise the question of who it is unknown to. The reader may think the reporter didn't do enough research. Or doubt the NSA's account of what happened. Adding an irrelevant statistic makes the article more believable. And the more specific the statistic the better. Would the effect be the same if it were "2 million", or the more cautious "over 1 million"? I bet media analysts have studied click rates and know just what kind of meaningless statistics can be added to an article to increase readership. This is what happens when success in journalism is measured by page views.