You could also just be doing real-time sentiment analysis of the readers of hackernews right? I wonder how you'd prove or disprove this. I know I get a similar

You could also just be doing real-time sentiment analysis of the readers of hackernews right? I wonder how you'd prove or disprove this. I know I get a similar feeling lately about Reddit, that some topics seem to be machine generated due to pretty surprising typos (not the existence of them, but the kinds, and not the kinds you get used to from ESL students).