Nicely put together article. Could it be that Google et.al are using it to double check their classifiers. That is to say to make sure that when their classifiers are making a decision, humans agree with it? Overall a lot of good points and food for thought! Thank you.