If I'm not mistaken, they mean meticulously proofred by us in reCAPTCHAs.
When I was proofreading on DP, all rounds of proofreading involved examining the scanned images and comparing it to the OCR text and making corrections. The later rounds of proofreading involved increasing attention to various details of correctness and formatting. All of this was done directly in the DP web interface. I didn't see any mention of the use of captchas in the OCR process.
If Microsoft want me to use their documentation they better make sure google indexes it in a way than matches my queries.
Perhaps you should switch to using Bing, then your query responses might match up with what Microsoft wants?
It conjures up images of some kind of celestial primate flinging high energy particles about.
At least that's better than the stuff they usually fling around...