Larry's post was designed to quickly assuage our knee-jerk fears about a giant like google having anything to do with "PRISM". In that, it is effective.

I think for a lot of folks complaining, his language (i.e. "direct access") is just overly precise enough to leave too much room in the margins for technical loopholes concerning "who" has access to "what" data. When Larry defends their general policies stating google "pushes back on requests", I am not convinced when terms like "overly broad" and "correct process" are left undefined.

But it's a 4 paragraph blog post- what do I expect?

Even legal docs are open to interpretation - they're not math. It's hard to expect a concise and readable post, not couched in legal phrasing, to somehow exclude all possible alternate interpretations. Especially when it can't be absolute in phrasing because there are known (and accepted) exceptions.